<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Antoine Tilloy’s research log</title><link>https://atilloy.com/</link><description>orthodox and unorthodox explorations of Quantum Theory</description><language>en</language><lastBuildDate>Sat, 25 Apr 2026 10:17:25 +0200</lastBuildDate><atom:link href="https://atilloy.com/" rel="self" type="application/rss+xml"/><item><title>News: habilitation thesis</title><link>https://atilloy.com/2026/04/25/news-habilitation-thesis/</link><guid>https://atilloy.com/2026/04/25/news-habilitation-thesis/</guid><pubDate>Sat, 25 Apr 2026 10:17:25 +0200</pubDate><description>&lt;p&gt;After a long hiatus, I want to reactivate this blog. The latest (tough not so recent) piece of news is that I have defended my Habilitation thesis in January (2026). I can now officially supervise PhD students. Two of my PhD students had already defended their thesis before that, but I was never &lt;em&gt;officially&lt;/em&gt; their advisor (technically not even their co-advisor). Now finally, I can manage everything without bothering a more senior colleague.&lt;/p&gt;&#10;&lt;p&gt;Usually, the Habilitation thesis is more of a compilation of summaries, that one assembles last minute to fulfill the technical requirements for the defense &amp;ndash; the real evaluation being done from the candidate&amp;rsquo;s CV. But graphomania took me over, and I ended up writing something quite long about my recent adventure in quantum field theory. It&amp;rsquo;s an overview of the intense efforts to make relativistic continuous matrix product states work, with a discussion of recent developments (some of which were not fully finished when I wrote the manuscript). Because I couldn&amp;rsquo;t help it, I added some borderline philosophical discussion about why I thought it was a good idea in the first place. The topic is pretty niche, but I am hoping it might be interesting for the handful of people interested in this approach. The manuscript itself, with its admin part stripped off, is on &lt;a href="https://arxiv.org/abs/2604.10284"&gt;arxiv&lt;/a&gt; for everyone to read.&lt;/p&gt;&#10;</description></item><item><title>Réseaux de tenseurs, suprématie, et problème à N corps</title><link>https://atilloy.com/2024/09/09/reseaux-de-tenseurs-suprematie-et-probleme-a-n-corps/</link><guid>https://atilloy.com/2024/09/09/reseaux-de-tenseurs-suprematie-et-probleme-a-n-corps/</guid><pubDate>Mon, 09 Sep 2024 18:03:02 +0200</pubDate><description>&lt;p&gt;J&amp;rsquo;ai écrit un &lt;a href="https://www.pourlascience.fr/sd/physique-theorique/reseaux-de-tenseurs-les-ordinateurs-classiques-contre-attaquent-26738.php"&gt;petit article&lt;/a&gt; dans &lt;em&gt;Pour la Science&lt;/em&gt; sur les réseaux de tenseurs et le lien avec la suprématie quantique. J&amp;rsquo;ai essayé d&amp;rsquo;adopter un angle un peu différent en ne mentionnant la bataille sur les problèmes de suprématie pure qu&amp;rsquo;en incipit.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2024/09/capture_pls.png" width="191"&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;De mon point de vue, l&amp;rsquo;intérêt principal des réseaux de tenseurs comme des ordinateurs quantiques est de résoudre le problème à N corps quantique dans des cas difficiles. Je rappelle ainsi dans cet article principalement ce qu&amp;rsquo;est ce problème, et les stratégies utilisées pour le résoudre. La conclusion, qui rejoint par exemple l&amp;rsquo;avis de Garnett Chan que j&amp;rsquo;aurais dû citer, c&amp;rsquo;est que pour le problème à N corps, il n&amp;rsquo;est pas évident que les ordinateurs quantiques seront forcément la meilleure méthode pour les instances utiles. C&amp;rsquo;est possible, mais ça n&amp;rsquo;est pas évident, et les algorithmes classiques pourront résister plus longtemps pour les problèmes utiles que pour les problèmes de suprématie pure.&lt;/p&gt;&#10;&lt;p&gt;Évidemment, pour les rares applications classiques établies comme l&amp;rsquo;algorithme de Shor, l&amp;rsquo;avantage de l&amp;rsquo;ordinateur quantique est asymptotiquement clair. De ce point de vue, je ne suis pas du tout un sceptique de l&amp;rsquo;avantage quantique. Je trouverais fascinant pour l&amp;rsquo;humanité qu&amp;rsquo;une énorme machine factorise des gros nombres en utilisant les lois bizarres de la physique fondamentale.&lt;/p&gt;&#10;&lt;p&gt;Mais si l&amp;rsquo;avantage quantique est restreint à quelques problèmes de théorie des nombres (d&amp;rsquo;intérêt pratique à mon avis bien moindre que le problème à N corps), alors l&amp;rsquo;impact pour la société sera inférieur à ce que certains prévoient. Je pense que même si on travaille dans une équipe qui cherche ultimement à améliorer le hardware d&amp;rsquo;ordinateurs quantiques, ce qui est mon cas avec &lt;em&gt;Quantic&lt;/em&gt; (joint venture Inria, ENS, Mines), on doit rester lucide sur ce point.&lt;/p&gt;&#10;&lt;p&gt;&lt;strong&gt;note I&lt;/strong&gt; : au moins temporairement, l&amp;rsquo;article est lisible gratuitement sur la &lt;a href="https://www.pourlascience.fr/sd/informatique/pour-la-science-n563-26740.php"&gt;preview du magazine de septembre&lt;/a&gt; - mais n&amp;rsquo;hésitez pas à l&amp;rsquo;acheter pour soutenir le journal (ça ne change rien à ce que je touche pour la pige !)&lt;/p&gt;&#10;&lt;p&gt;&lt;strong&gt;note II&lt;/strong&gt; : dans la courte notice biographique, l&amp;rsquo;article ne mentionne que l&amp;rsquo;institution qui paie généreusement mon salaire, et à laquelle j&amp;rsquo;enseigne : le Centre Automatique et Systèmes de &lt;em&gt;Mines Paris - PSL&lt;/em&gt;. Mea culpa. Il aurait dû mentionner aussi l&amp;rsquo;institution qui m&amp;rsquo;héberge (ainsi que mon groupe) dans d&amp;rsquo;exceptionnelles conditions matérielles tout en me fournissant de brillants collègues : &lt;em&gt;Inria Paris&lt;/em&gt;. À ces deux honorables institutions s&amp;rsquo;ajoute le laboratoire de physique de l&amp;rsquo;ENS-PSL, qui fait lui aussi partie de l&amp;rsquo;équipe projet &lt;em&gt;Quantic&lt;/em&gt;, et en héberge la partie expérimentale. J&amp;rsquo;y suis aussi mécaniquement affilié. Le laboratoire de Physique de l&amp;rsquo;ENS-PSL étant une unité mixte de recherche avec le CNRS, il faut donc aussi le remercier. Si l&amp;rsquo;article vous déplaît, vous avez ainsi un large choix d&amp;rsquo;institutions auprès desquelles vous plaindre !&lt;/p&gt;&#10;</description></item><item><title>Notes d’optimisation</title><link>https://atilloy.com/2023/06/06/notes-doptimisation/</link><guid>https://atilloy.com/2023/06/06/notes-doptimisation/</guid><pubDate>Tue, 06 Jun 2023 09:31:40 +0200</pubDate><description>&lt;p&gt;J&amp;rsquo;ai donné pour la deuxième année consécutive le TD du cours d&amp;rsquo;optimisation convexe de première année des Mines Paris. J&amp;rsquo;ai écrit quelques notes pour aider les étudiants à visualiser les théorèmes du cours. Elles peuvent peut-être en aider d&amp;rsquo;autres&amp;hellip; À lire en restant vigilant, car c&amp;rsquo;est un domaine où je ne suis qu&amp;rsquo;un amateur (dans les deux sens du terme : de celui qui débute, et de celui qui apprécie &amp;ndash; double sens auquel m&amp;rsquo;a initié Denis Bernard, mon directeur de thèse) . J&amp;rsquo;ai pas mal utilisé les méthodes d&amp;rsquo;optimisation (surtout dans le cas différentiable et sans contrainte) pour ma recherche, mais je n&amp;rsquo;ai appris la théorie (et les cas plus fins, par exemple non différentiables) essentiellement qu&amp;rsquo;avec ces TD. Pour ceux qui veulent aller plus loin que ces notes ultra basiques, le mieux est de jeter un œil à &lt;em&gt;Numerical Optimization&lt;/em&gt; de Nocedal et Wright.&lt;/p&gt;&#10;&#10;&lt;div class="pdf"&gt;&#10; &lt;object data="/wp-content/uploads/2023/06/notes_optimisation.pdf" type="application/pdf"&gt;&lt;/object&gt;&#10; &lt;p&gt;&lt;a href="https://atilloy.com/wp-content/uploads/2023/06/notes_optimisation.pdf"&gt;📄 notes_optimisation&lt;/a&gt; (PDF)&lt;/p&gt;&#10;&lt;/div&gt;&#10;&#10;</description></item><item><title>Postdoc offer</title><link>https://atilloy.com/2022/12/16/postdoc-offer/</link><guid>https://atilloy.com/2022/12/16/postdoc-offer/</guid><pubDate>Fri, 16 Dec 2022 15:02:53 +0100</pubDate><description>&lt;p&gt;I put a first postdoc offer on &lt;a href="https://inspirehep.net/jobs/2613102"&gt;Inspire&lt;/a&gt; a few days ago. I am looking for smart independent people, who are ideally willing to push the use of tensor network methods in quantum field theory.&lt;/p&gt;&#10;&lt;p&gt;The application has a very close deadline in the advertisement, so that I can make offers to candidates in sync with the standard high energy theory postdoc cycle (to which I was only partly familiar with a few days ago). If you are interested and cannot assemble all the documents by the deadline, feel free to contact me and to apply still. If you really arrive too late, I will open another postdoc position later, still funded by the ERC.&lt;/p&gt;&#10;</description></item><item><title>ERC funding</title><link>https://atilloy.com/2022/11/22/erc-funding/</link><guid>https://atilloy.com/2022/11/22/erc-funding/</guid><pubDate>Tue, 22 Nov 2022 18:03:02 +0100</pubDate><description>&lt;p&gt;It is now official: the ERC awarded me generous research funding to work on the intersection of tensor networks and quantum field theory (see e.g. &lt;a href="https://www.minesparis.psl.eu/Actualites/ERC-Antoine-Tilloy-laureat/5787"&gt;here&lt;/a&gt;). The proposal that got me the money is &lt;a href="https://atilloy.com/wp-content/uploads/2022/11/erc_b2_tilloy_210412-1.pdf"&gt;here&lt;/a&gt; and, as far as I am concerned, it is public (I will be happy if you can solve the open problems faster than my collaborators and I). Some formal job openings will follow but you can already check the &amp;ldquo;&lt;a href="https://atilloy.com/working-with-me-2/"&gt;working with me&amp;rdquo; section&lt;/a&gt;.&lt;/p&gt;&#10;&lt;p&gt;Soon, I will have to work. But now I thank the people who helped me. Thanks to those who proofread my proposal and pushed me to apply (in particular Pierre Rouchon). Thanks also to Jean-Pierre Banatre who helped me prepare for the interview. Thanks to Mathieu Reboul at &lt;em&gt;Armines&lt;/em&gt; who took care of the admin part and allowed me to focus on the science. Thanks, to the UK for leaving the ERC, which apparently allowed my project to go from &amp;ldquo;&lt;em&gt;recommended for funding&lt;/em&gt;&amp;rdquo; to &amp;ldquo;&lt;em&gt;funded&lt;/em&gt;&amp;rdquo;. Thanks finally to &lt;em&gt;randomness&lt;/em&gt; &amp;ndash; whether it comes from the splitting of many worlds, the collapse of the wave function, or the last digits of the Bohmian positions at the beginning of time &amp;ndash; it certainly helped me.&lt;/p&gt;&#10;</description></item><item><title>Research news</title><link>https://atilloy.com/2022/09/18/research-news/</link><guid>https://atilloy.com/2022/09/18/research-news/</guid><pubDate>Sun, 18 Sep 2022 09:46:17 +0200</pubDate><description>&lt;p&gt;I am at the &lt;em&gt;Erwin Schrödinger Institute&lt;/em&gt; in Vienna for 1 more week, where I get a perfusion of tensor networks straight into my veins. There, I unsurprisingly talked about my recent work on continuous matrix product states. The blackboard talk was recorded, but sadly I wrote a little too small so I have to squint to read the equations. It is adapted to an audience knowing tensor networks very well, but quantum field theory slightly less.&lt;/p&gt;&#10;&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;"&gt;&#10;&#9;&#9;&#9;&lt;iframe allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share; fullscreen" loading="eager" referrerpolicy="strict-origin-when-cross-origin" src="https://www.youtube.com/embed/F5p9WSQsgSg?autoplay=0&amp;amp;controls=1&amp;amp;end=0&amp;amp;loop=0&amp;amp;mute=0&amp;amp;start=0" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" title="YouTube video"&gt;&lt;/iframe&gt;&#10;&#9;&#9;&lt;/div&gt;&#10;&#10;&lt;p&gt;To have more things to discuss at the breaks, I put on &lt;a href="https://arxiv.org/abs/2209.05341"&gt;arxiv&lt;/a&gt; the results of my exploration of the ground state of the Sinh-Gordon model with relativistic continuous matrix product state. This is not a final version (some references are missing, and I will likely update the discussion), but I had to put it online at some point. I spent quite a lot of time optimizing the numerics but at the end still couldn&amp;rsquo;t fully settle the situation of the model near its self dual point. I still think the data is interesting nonetheless: it is obtained in the thermodynamic limit, and is fairly precise for moderate coupling. I hope it will also push people to develop better contraction routines, as ramping up the bond dimension could be enough to clarify the matter (at least numerically, before mathematical physicists and mathematicians nail it).&lt;/p&gt;&#10;</description></item><item><title>Smart person needed for a PhD!</title><link>https://atilloy.com/2022/03/14/smart-person-needed-for-a-phd/</link><guid>https://atilloy.com/2022/03/14/smart-person-needed-for-a-phd/</guid><pubDate>Mon, 14 Mar 2022 11:26:16 +0100</pubDate><description>&lt;p&gt;&lt;strong&gt;update&lt;/strong&gt;: this position is now filled, but there will be new PhD and postdoc offerings very soon, on similar topics.&lt;/p&gt;&#10;&lt;hr&gt;&#10;&lt;p&gt;I am now looking for a PhD student to work with me, starting next year (December 2022 at the latest). I have the money, I just need a smart person. The ideal topic, described in more detail in the offer below, would be tensor networks and their use in non-perturbative quantum field theory. However, I am open to inquiries on any other of my past subjects, so if there is something you are interested in, feel free to contact me!&lt;/p&gt;&#10;&#10;&lt;div class="pdf"&gt;&#10; &lt;object data="/wp-content/uploads/2022/03/phd_proposal_tilloy.pdf" type="application/pdf"&gt;&lt;/object&gt;&#10; &lt;p&gt;&lt;a href="https://atilloy.com/wp-content/uploads/2022/03/phd_proposal_tilloy.pdf"&gt;📄 phd_proposal_tilloy&lt;/a&gt; (PDF)&lt;/p&gt;&#10;&lt;/div&gt;&#10;&#10;&lt;p&gt;The best is to start with an internship, to get a feel for the subject. This is why it is important to contact me asap if you&amp;rsquo;re interested, even before the offer is formally advertised on the website of the école doctorale.&lt;/p&gt;&#10;</description></item><item><title>Various news</title><link>https://atilloy.com/2021/10/26/various-news/</link><guid>https://atilloy.com/2021/10/26/various-news/</guid><pubDate>Tue, 26 Oct 2021 15:00:21 +0200</pubDate><description>&lt;p&gt;1- Since October 4th, I am back in Paris with a position at &lt;em&gt;Mines Paristech&lt;/em&gt;, which is a French &amp;ldquo;grande école&amp;rdquo; part of Paris Science et Lettres (PSL). I am also a member of the Quantic group, which is a joint venture with &lt;em&gt;Inria&lt;/em&gt; and &lt;em&gt;École Normale Supérieure&lt;/em&gt; exploring superconducting circuits for quantum information processing (theoretically and experimentally). I am very happy about this move back to Paris after 5 great years in Munich where I learned immensely and became more of an &amp;ldquo;adult&amp;rdquo; theoretical physicist. In Paris I will keep an interest in my old subjects: quantum field theory with variational methods, many body physics, quantum foundations (and others..). But I will try to develop tighter connection between state compression methods (tensor networks) and what is done in the group especially with cat qubits and error correction. For all these subjects, I will probably have some money soon from various sources. Students interested in any of these themes (even the exotic stuff) should feel free to contact me in advance to discuss possible PhDs or internships.&lt;/p&gt;&#10;&lt;p&gt;2- I gave a rather conceptual talk on collapse models at the &lt;a href="https://quantumboundaries2021.wordpress.com/"&gt;Quantum Boundaries workshop&lt;/a&gt;. The talk was recorded:&lt;/p&gt;&#10;&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;"&gt;&#10;&#9;&#9;&#9;&lt;iframe allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share; fullscreen" loading="eager" referrerpolicy="strict-origin-when-cross-origin" src="https://www.youtube.com/embed/U25Mq-V5fUs?autoplay=0&amp;amp;controls=1&amp;amp;end=0&amp;amp;loop=0&amp;amp;mute=0&amp;amp;start=0" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" title="YouTube video"&gt;&lt;/iframe&gt;&#10;&#9;&#9;&lt;/div&gt;&#10;&#10;&lt;p&gt;3- For those who are in Paris, I will speak at &lt;em&gt;&lt;a href="https://foundations2020.sciencesconf.org/resource/page/id/8"&gt;Foundation 2020&lt;/a&gt;&lt;/em&gt; (which was delayed because of covid) on Thursday October 28th. I will present our latest results with Howard Wiseman on the equivalence between non-Markovian collapse models and Bohmian theories (&lt;a href="https://arxiv.org/abs/2105.06115#"&gt;this preprint&lt;/a&gt;).&lt;/p&gt;&#10;</description></item><item><title>Notes on continuous tensor networks</title><link>https://atilloy.com/2021/09/30/notes-on-continuous-tensor-networks/</link><guid>https://atilloy.com/2021/09/30/notes-on-continuous-tensor-networks/</guid><pubDate>Thu, 30 Sep 2021 14:54:08 +0200</pubDate><description>&lt;p&gt;For the European Tensor Network school 2021 at ICCUB in Barcelona, I started to write some lecture notes on continuous tensor network states. So far they cover only continuous matrix product states, which is a pretty stabilized subject. I tried to make them very detailed about the computations which can be scary for beginners. I plan (or rather hope) to extend the notes to include more recent developments like relativistic extensions in 1 space dimension, and non-relativistic extensions in 2 space dimensions and more.&lt;/p&gt;&#10;&#10;&lt;div class="pdf"&gt;&#10; &lt;p&gt;&lt;a href="https://atilloy.com/wp-content/uploads/2021/09/lecture_tensor21.pdf"&gt;📄 lecture_tensor21&lt;/a&gt; (PDF)&lt;/p&gt;&#10;&lt;/div&gt;&#10;&#10;</description></item><item><title>Une discussion sur la gravité (non)-quantique</title><link>https://atilloy.com/2021/08/10/une-discussion-sur-la-gravite-non-quantique/</link><guid>https://atilloy.com/2021/08/10/une-discussion-sur-la-gravite-non-quantique/</guid><pubDate>Tue, 10 Aug 2021 14:37:25 +0200</pubDate><description>&lt;p&gt;J&amp;rsquo;ai discuté il y a quelques jours avec Vincent Debierre (du podcast Libaca) du problème de l&amp;rsquo;unification (ou de la jonction) entre modèle standard et gravité. J&amp;rsquo;ai surtout essayé de clarifier la logique : qu&amp;rsquo;est-ce qu&amp;rsquo;on cherche à faire, qu&amp;rsquo;est-ce qui est a priori possible et qu&amp;rsquo;est-ce qui ne l&amp;rsquo;est pas ?&lt;/p&gt;&#10;&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;"&gt;&#10;&#9;&#9;&#9;&lt;iframe allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share; fullscreen" loading="eager" referrerpolicy="strict-origin-when-cross-origin" src="https://www.youtube.com/embed/7H-clEQ4JR0?autoplay=0&amp;amp;controls=1&amp;amp;end=0&amp;amp;loop=0&amp;amp;mute=0&amp;amp;start=0" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" title="YouTube video"&gt;&lt;/iframe&gt;&#10;&#9;&#9;&lt;/div&gt;&#10;&#10;&lt;p&gt;Pour ceux qui peuvent passer outre ma grammaire floue et mon élocution un peu non-linéaire, c&amp;rsquo;est un moyen d&amp;rsquo;avoir une idée de ce domaine d&amp;rsquo;un point de vue peut-être hétérodoxe. Je suis aussi preneur de commentaires éventuellement critiques sur la présentation du problème, car je ne suis évidemment pas sans biais.&lt;/p&gt;&#10;&lt;p&gt;À la fin, je mentionne aussi les réseaux de tenseurs. C&amp;rsquo;est une nouvelle méthode pour résoudre numériquement des modèles existants difficiles (et non un nouveau modèle de la nature) qui est assez à la mode et qui m&amp;rsquo;occupe aujourd&amp;rsquo;hui davantage que la gravité. Je suis curieux de savoir si la philosophie en est compréhensible en quelques phrases.&lt;/p&gt;&#10;</description></item><item><title>Progress in the variational method for QFT</title><link>https://atilloy.com/2021/07/06/progress-in-the-variational-method-for-qft/</link><guid>https://atilloy.com/2021/07/06/progress-in-the-variational-method-for-qft/</guid><pubDate>Tue, 06 Jul 2021 08:16:36 +0200</pubDate><description>&lt;p&gt;Today I put online a major update of my pair of papers on the variational method for quantum field theory (&lt;a href="https://arxiv.org/abs/2102.07733"&gt;short here&lt;/a&gt;, &lt;a href="https://arxiv.org/abs/2102.07741"&gt;long here&lt;/a&gt;). The idea is still to use the same class of variational wave functions (relativistic continuous matrix product states) to find the ground state of (so far bosonic) quantum field theories in 1+1 dimensions. The novelty comes from the algorithm I now use to compute expectation values, that has a cost only proportional to &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D^3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; where &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6833em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; is the bond dimension. Using backpropagation techniques, the cost of computing the gradient of observables with respect to the parameters is also only &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D^3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;. This is basically the same asymptotic scaling as standard continuous matrix product states.&lt;/p&gt;&#10;&lt;p&gt;Previously, the method was a nice theoretical advance as it worked without any cutoff, but it was not numerically competitive compared to bruteforce discretization + standard tensor methods (at least not competitive for most observables insensitive to the UV). With the new algorithm with improved scaling, I can go fairly easily from &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;9&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D=9&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6833em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;9&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; to &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;32&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D=32&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6833em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;32&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, which gives &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mo&gt;≃&lt;/mo&gt;&lt;msup&gt;&lt;mn&gt;10&lt;/mn&gt;&lt;mrow&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mn&gt;5&lt;/mn&gt;&lt;/mrow&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\simeq 10^{-5}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.4637em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;≃&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord"&gt;0&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mtight"&gt;−&lt;/span&gt;&lt;span class="mord mtight"&gt;5&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; relative error for the renormalized energy density at a coupling of order 1. Crucially, the error really seems to decrease exponentially as a function of the bond dimension, and thus with only a slightly higher numerical effort (say 100 times more) one could probably get close to machine precision. Already at &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mn&gt;10&lt;/mn&gt;&lt;mrow&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mn&gt;5&lt;/mn&gt;&lt;/mrow&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;10^{-5}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord"&gt;0&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mtight"&gt;−&lt;/span&gt;&lt;span class="mord mtight"&gt;5&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; relative error for the renormalized energy density, that is a quantity where the leading lattice contribution has been subtracted, I doubt methods relying on a discretization can compete. This makes me a bit more confident that methods working directly in the continuum are a promising way forward even if only for numerics.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2021/07/joint.png"&#10;&#9;&#9;&#9;alt="Energy density and relative error for ϕ4\phi^4ϕ4 theory with RCMPS. RHT is the state-of-the-art renormalized Hamiltonian truncation result which is manifestly less precise"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Energy density and relative error for &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0085em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; theory with RCMPS. RHT is the state-of-the-art renormalized Hamiltonian truncation result which is manifestly less precise&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Now, let me go a bit more into the technicalities. Computing expectation values of local functions of the field &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8889em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; with such a low cost &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D^3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; seems difficult at first because the ansatz is not written in terms of local functions of &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8889em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;. Naively this should at the very least square the cost to &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;mn&gt;6&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D^6&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;6&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;. The main idea to obtain the cheap scaling is to realize that the expectation value of vertex operators, i.e. operators of the form &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mo&gt;:&lt;/mo&gt;&lt;mi&gt;exp&lt;/mi&gt;&lt;mo&gt;⁡&lt;/mo&gt;&lt;mo stretchy="false"&gt;(&lt;/mo&gt;&lt;mi&gt;α&lt;/mi&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mo stretchy="false"&gt;)&lt;/mo&gt;&lt;mo&gt;:&lt;/mo&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;: \exp( \alpha \phi):&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.4306em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;:&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1em;vertical-align:-0.25em;"&gt;&lt;/span&gt;&lt;span class="mop"&gt;exp&lt;/span&gt;&lt;span class="mopen"&gt;(&lt;/span&gt;&lt;span class="mord mathnormal" style="margin-right:0.0037em;"&gt;α&lt;/span&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="mclose"&gt;)&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;:&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, can be computed by solving an ordinary differential equation (ODE) where the generator has a cost &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;D&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;D^3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0278em;"&gt;D&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;. Basically, computing vertex operators for relativistic CMPS is as expensive as computing field expectation values for standard non-translation-invariant CMPS. To solve this ODE, one can use powerful method with extremely quickly decaying errors as a function of the discretization step (e.g. very high order Runge-Kutta). So vertex operators are, in fact cheap. But local powers of the field are merely differentials of vertex operators, and thus can be computed as well for the same cost. Finally, to get the gradient, one can differentiate through the ODE with backpropagation, and obtain the result for only twice the cost. This allows the full variational optimization for all well-defined bosonic Hamiltonians with polynomial and exponential potentials.&lt;/p&gt;&#10;</description></item><item><title>Two recent articles on tensor networks: Monte-Carlo chimera and the phi^4 supremacy</title><link>https://atilloy.com/2021/04/28/two-recent-articles-on-tensor-networks-monte-carlo-chimera-and-the-phi4-supremacy/</link><guid>https://atilloy.com/2021/04/28/two-recent-articles-on-tensor-networks-monte-carlo-chimera-and-the-phi4-supremacy/</guid><pubDate>Wed, 28 Apr 2021 14:28:10 +0200</pubDate><description>&lt;p&gt;There are 2 recent preprints on tensor networks I found interesting.&lt;/p&gt;&#10;&lt;p&gt;The first is &lt;em&gt;&lt;strong&gt;Collective Monte Carlo updates through tensor network renormalization&lt;/strong&gt;&lt;/em&gt; (&lt;a href="https://arxiv.org/pdf/2104.13264.pdf"&gt;arxiv:2104.13264&lt;/a&gt;) by Miguel Frias-Perez, Michael Marien, David Perez Garcia, Mari Carmen Banuls and Sofyan Iblisdir. I am certainly a bit biased because two authors come from our group at MPQ, and I heard about the result twice in seminars, but I think it is a genuinely new and interesting use of tensor networks. The main idea is to reduce the rejection rate of Markov Chain Monte Carlo (MCMC) by crudely estimating the probability distribution to be sampled from with tensor network renormalization. This combines the advantage of the Monte Carlo method (it is exact asymptotically, but slow especially for frustrated systems) with the advantage of tensor renormalization (it is very fast for approximations).&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2021/04/capture_2104.13264.png"&#10;&#9;&#9;&#9;alt="Figure taken from arxiv:2104.13264" width="429"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Figure taken from &lt;a href="https://arxiv.org/pdf/2104.13264.pdf"&gt;arxiv:2104.13264&lt;/a&gt;&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;A quick word of caution for purists, technically they use the tensor renormalization group (TRG), which is the simplest approximate contraction method for 2d tensor networks. Tensor network renormalization (TNR) can refer to a specific (and more subtle) renormalization method introduced by Evenbly and Vidal (&lt;a href="https://arxiv.org/abs/1412.0732"&gt;arxiv:1412.0732&lt;/a&gt;) that is in fact closer to what the Wilsonian RG does. This is just a matter of terminology.&lt;/p&gt;&#10;&lt;p&gt;Terminology aside, it seems the hybrid Monte Carlo method they obtain is dramatically faster than standard algorithms, at least in number of steps to thermalize the Markov chain (which remains of order 1 even for reasonably low bond dimensions). These steps are however necessarily more costly than for typical methods because sampling from the approximate distribution generated with TRG gets increasingly expensive as the bond dimension increases. As a result, for a large frustrated system, it is not clear to me how much faster the method can be in true computing time. I think in the future it would be really interesting to have a benchmark to see if the method can crush the most competitive Markov chain heuristics on genuine hard problems, like spin glasses or hard spheres for example. It is known that TRG struggles to get precise results for such frustrated problems, and so the approximation will surely degrade. But perhaps there is a sweet spot where the crude approximation with TRG (of high but manageable bond dimension) is sufficient to have a rejection rate in the MCMC at 10^(-4) instead of 10^(-15) for other methods in a hard-to-sample phase.&lt;/p&gt;&#10;&lt;p&gt;The second one is &lt;em&gt;&lt;strong&gt;Entanglement scaling for &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;λ&lt;/mi&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\lambda\phi^4_2&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal"&gt;λ&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/strong&gt;&lt;/em&gt; by Bram Vanhecke, Frank Verstraete, and Karel Van Acoleyen. I am of course interested because it is on &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;λ&lt;/mi&gt;&lt;msup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\lambda \phi^4&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0085em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal"&gt;λ&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, my favorite toy theory of relativistic quantum field theory. There are many ways to solve this non-trivial theory numerically, and I really find them all interesting. Recently, I explored ways to work straight in the continuum, which is the most robust or rigorous perhaps, but so far not the most efficient for most observables.&lt;/p&gt;&#10;&lt;p&gt;So far, the most efficient method is to discretize the model, use the most powerful lattice tensor techniques to solve it, and then extrapolate the results to the continuum limit. The extrapolation step is crucial. Indeed, such relativistic field theories are free conformal field theories at short distance. This implies that for a fixed error, the bond dimension of the tensor representation (and thus the computational cost) explodes as the lattice spacing goes to zero. One can take the lattice spacing small, but not arbitrarily small.&lt;/p&gt;&#10;&lt;p&gt;With Clément Delcamp, we explored this strategy in the most naive way. We had used a very powerful and to some extent subtle algorithm (tensor network renormalization with graph independent local truncation, aka GILT-TNR) to solve the lattice theory. We pushed the bond dimension to the max that could run on our cluster, plotted 10 points and extrapolated. At the smallest lattice spacing, we were already very close to the real continuum theory, and a fairly easy 1-dimensional fit with 3 parameters gave us accurate continuum limit results. A crucial parameter for &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;λ&lt;/mi&gt;&lt;msup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\lambda \phi^4&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0085em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal"&gt;λ&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; theory is the value of the critical coupling &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;f&lt;/mi&gt;&lt;mi&gt;c&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;f_c&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8889em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.1076em;"&gt;f&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1514em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.1076em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;c&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, and our estimate was &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;f&lt;/mi&gt;&lt;mi&gt;c&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;11.0861&lt;/mn&gt;&lt;mo stretchy="false"&gt;(&lt;/mo&gt;&lt;mn&gt;90&lt;/mn&gt;&lt;mo stretchy="false"&gt;)&lt;/mo&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;f_c = 11.0861(90)&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8889em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.1076em;"&gt;f&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1514em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.1076em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;c&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1em;vertical-align:-0.25em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;11.0861&lt;/span&gt;&lt;span class="mopen"&gt;(&lt;/span&gt;&lt;span class="mord"&gt;90&lt;/span&gt;&lt;span class="mclose"&gt;)&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;. We estimated it naively: by putting our lattice models exactly criticality, and extrapolating the line of results to the continuum limit. This was in 2020, and back then this was the best estimate. It was the best because of the exceptional accuracy of GILT-TNR, which allowed us to simulate well systems even near criticality, and allowed us to go closer to the continuum limit than Monte Carlo people.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2021/04/capture_gilt.png"&#10;&#9;&#9;&#9;alt="Extrapolation of the critical coupling as the lattice spacing (lambda is a proxy for it here) goes to zero in our paper with Clément." width="510"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Extrapolation of the critical coupling as the lattice spacing (lambda is a proxy for it here) goes to zero in our paper with Clément.&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Bram Vanhecke and collaborators follow a philosophically different strategy. They use an arguably simpler method to solve the lattice theory, boundary matrix product states. Compared to GILT-TNR, this method is simpler but a priori less efficient to solve an exactly critical problem. &lt;strong&gt;But, &lt;em&gt;spoiler&lt;/em&gt;, they ultimately get better estimates of the critical coupling than we did.&lt;/strong&gt; How is that possible? Instead of simulating exactly the critical theory with the highest possible bond dimensions, the authors simulate many points away from criticality and with different bond dimension (see image below). Then they use an hypothesis about how the results should scale as a function of the bond dimension and distance from criticality. This allows them to fit the whole theory manifold around the critical point without ever simulating it. This approach is much more data driven in a way. It swallows the bullet that some form of extrapolation is going to be needed anyway and turns it into an advantage. Since the extrapolation is done with respect to more parameters (not just lattice spacing), the whole manifold of models is fitted instead of simply a critical line. This gives the fit more rigidity. By putting sufficiently many points, which are cheap to get since they are not exactly at criticality or with maximal D, one can get high precision estimates. They find &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;f&lt;/mi&gt;&lt;mi&gt;c&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;11.09698&lt;/mn&gt;&lt;mo stretchy="false"&gt;(&lt;/mo&gt;&lt;mn&gt;31&lt;/mn&gt;&lt;mo stretchy="false"&gt;)&lt;/mo&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;f_c = 11.09698(31)&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8889em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.1076em;"&gt;f&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1514em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.1076em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;c&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1em;vertical-align:-0.25em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;11.09698&lt;/span&gt;&lt;span class="mopen"&gt;(&lt;/span&gt;&lt;span class="mord"&gt;31&lt;/span&gt;&lt;span class="mclose"&gt;)&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, which if their error estimate is correct, is about 1 and a half digits better than our earlier estimate (which is a posteriori confirmed, since the new result is about 1 standard deviation away from the mean we had proposed).&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2021/04/capture_entanglementscaling.png"&#10;&#9;&#9;&#9;alt="The points sampled by Bram Vanhecke et al around the critical line (also for different bond dimensions not shown)" width="484"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;The points sampled by Bram Vanhecke et al around the critical line (also for different bond dimensions not shown)&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Naturally, an extrapolation is only as good as the hypothesis one has for the functional form of the fit. There, I like to hope that our earlier paper with Clément helped. By looking at the critical scaling as a function of the lattice spacing, and going closer to the continuum limit than before, we could see that the best fit clearly contained logarithms and was not just a very high order polynomial (see above). I think it was the first time one could be 100 % sure that such non-obvious log corrections existed (previous Monte Carlo papers had ruled it out after suspecting it), the &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;χ&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\chi^2&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0085em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;χ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; left no doubt. Interestingly, Bram and his Ghent collaborators had tried their smart extrapolations techniques back in 2019 (&lt;a href="https://arxiv.org/abs/1907.08603"&gt;arxiv:1907.08603&lt;/a&gt;) but their ansatz for the functional form of the critical coupling as a function of the lattice spacing was polynomial. As a result their estimates looked precise, between 11.06 and 11.07, but were in fact underestimating the true value (like almost all papers in the literature did).&lt;/p&gt;&#10;&lt;p&gt;Could this scaling have been found theoretically? Perhaps. In this new paper, Vanhecke and collaborators provide a perturbative justification. My understanding (which may be updated) is that such a derivation is only a hint of the functional form because the critical point is deep in the non-perturbative regime and thus all diagrams contribute with a comparable weight. For example, individual log contributions from infinitely many diagrams could get re-summed into polynomial ones, e.g. &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mo&gt;∑&lt;/mo&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;/msub&gt;&lt;mi&gt;log&lt;/mi&gt;&lt;mo&gt;⁡&lt;/mo&gt;&lt;mo stretchy="false"&gt;(&lt;/mo&gt;&lt;mi&gt;λ&lt;/mi&gt;&lt;msup&gt;&lt;mo stretchy="false"&gt;)&lt;/mo&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;/msup&gt;&lt;mi mathvariant="normal"&gt;/&lt;/mi&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo stretchy="false"&gt;!&lt;/mo&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mi&gt;λ&lt;/mi&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\sum_k \log(\lambda)^k/k! = \lambda&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.1488em;vertical-align:-0.2997em;"&gt;&lt;/span&gt;&lt;span class="mop"&gt;&lt;span class="mop op-symbol small-op" style="position:relative;top:0em;"&gt;∑&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1864em;"&gt;&lt;span style="top:-2.4003em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2997em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.1667em;"&gt;&lt;/span&gt;&lt;span class="mop"&gt;lo&lt;span style="margin-right:0.0139em;"&gt;g&lt;/span&gt;&lt;/span&gt;&lt;span class="mopen"&gt;(&lt;/span&gt;&lt;span class="mord mathnormal"&gt;λ&lt;/span&gt;&lt;span class="mclose"&gt;&lt;span class="mclose"&gt;)&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8491em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mord"&gt;/&lt;/span&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="mclose"&gt;!&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6944em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal"&gt;λ&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; non-perturbatively. In any case, I think (or rather hope&amp;hellip;) that precise but &amp;ldquo;less extrapolated&amp;rdquo; methods will remain complementary and even guide the hypothesis for more data driven estimates.&lt;/p&gt;&#10;&lt;p&gt;In the end, congratulations to Bram, Frank, and Karel for holding the critical &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_2&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; crown now. I will dearly miss it, and the bar is now so high that I am not sure we can contest it&amp;hellip;&lt;/p&gt;&#10;</description></item><item><title>Variational method in field theory - videos</title><link>https://atilloy.com/2021/03/23/variational-method-in-field-theory-videos/</link><guid>https://atilloy.com/2021/03/23/variational-method-in-field-theory-videos/</guid><pubDate>Tue, 23 Mar 2021 17:25:57 +0100</pubDate><description>&lt;p&gt;A bit more than a month ago, I put on arxiv two preprints (&lt;a href="https://arxiv.org/abs/2102.07733"&gt;short&lt;/a&gt; and &lt;a href="https://arxiv.org/abs/2102.07741"&gt;long&lt;/a&gt;) that summarize my recent work on applying the variational method in relativistic quantum field theory. I am happy that they were (so far) well received, and I got the chance to present the corresponding results in a few seminars already. The last two, at the University of Helsinki and EPFL, were recorded. I had a better microphone at EPFL, and expanded more on tensor networks, so I embed the video below (&lt;a href="https://tube.switch.ch/videos/0hSQb2G7MU"&gt;or direct link here&lt;/a&gt;).&lt;/p&gt;&#10;&lt;figure class="video"&gt;&lt;video controls preload="metadata" src="https://tube.switch.ch/download/video/0hSQb2G7MU"&gt;&lt;/video&gt;&lt;/figure&gt;&#10;&#10;&lt;p&gt;I skipped the introduction to the basics of relativistic quantum field theory at EPFL and so you may prefer the &lt;a href="https://unitube.it.helsinki.fi/unitube/embed.html?id=faf16af2-0fc6-4ea2-a7b3-3dbdb2f00097"&gt;Helsinki recording&lt;/a&gt; if this is not well known to you (as I was a basically two years ago). As the typical Frenchman, I say &amp;ldquo;eeeeuuuh&amp;rdquo; a lot when presenting, which makes the whole experience particularly awful for people listening. Now that I am painfully aware of it, I will try to work on it ifor future presentations&amp;hellip;&lt;/p&gt;&#10;</description></item><item><title>Random news</title><link>https://atilloy.com/2020/12/04/random-news-2/</link><guid>https://atilloy.com/2020/12/04/random-news-2/</guid><pubDate>Fri, 04 Dec 2020 10:05:08 +0100</pubDate><description>&lt;p&gt;The preprint on subcritical reactors I had written about two months ago has finally appeared on &lt;a href="https://arxiv.org/abs/2012.01566"&gt;arxiv&lt;/a&gt;. Apparently it was &amp;ldquo;on hold&amp;rdquo; for that long because the primary category (applied physics) was not the right one, and the moderators thought &amp;ldquo;instrumentation and detectors&amp;rdquo; was better. Honestly, I am not sure they are right, given how theoretical the paper is, but I am just glad the paper appeared (I would have just preferred it to be faster).&lt;/p&gt;&#10;&lt;div class="gallery"&gt;&#10; &lt;a href="https://atilloy.com/wp-content/uploads/2020/12/subcritical_1.png"&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/12/subcritical_1.png" alt="" loading="lazy"&gt;&lt;/a&gt;&#10; &lt;a href="https://atilloy.com/wp-content/uploads/2020/12/subcritical2.png"&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/12/subcritical2.png" alt="" loading="lazy"&gt;&lt;/a&gt;&#10; &lt;a href="https://atilloy.com/wp-content/uploads/2020/12/subcritical3.png"&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/12/subcritical3.png" alt="" loading="lazy"&gt;&lt;/a&gt;&#10; &lt;a href="https://atilloy.com/wp-content/uploads/2020/12/subcritical4.png"&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/12/subcritical4.png" alt="" loading="lazy"&gt;&lt;/a&gt;&#10;&lt;/div&gt;&#10;&#10;&lt;p&gt;I also have one article written with Vincent Vennin and one interview by Philippe Pajot in the latest edition of &lt;em&gt;La Recherche&lt;/em&gt;. Both are, of course, in French. This edition, with a new format, features many interesting articles, in particular an interview of the president of the &lt;em&gt;Max Planck Society&lt;/em&gt;.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/12/capture_larecherche.png"&#10;&#9;&#9;&#9;alt="Le dernier La Recherche en kiosques" width="346"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Le dernier La Recherche en kiosques&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;</description></item><item><title>Sustaining the chain reaction without criticality</title><link>https://atilloy.com/2020/08/03/sustaining-the-chain-reaction-without-criticality/</link><guid>https://atilloy.com/2020/08/03/sustaining-the-chain-reaction-without-criticality/</guid><pubDate>Mon, 03 Aug 2020 12:50:04 +0200</pubDate><description>&lt;p&gt;I have found a neat nuclear physics problem, and have written a &lt;a href="https://atilloy.com/wp-content/uploads/2020/08/neutron_multiplier_tilloy_200803.pdf"&gt;draft of article&lt;/a&gt; about it. Below I explain how I came to be interested in this problem, give some context most physicists may not be familiar with, and explain the result briefly.&lt;/p&gt;&#10;&lt;p&gt;A while back, I took a small part in the organization of the French public debate on radioactive waste management. My work was not very technical, but got me interested in the rich physics involved. The debate itself didn&amp;rsquo;t allow me to do real research on the subject: it was certainly not what was asked from me, and there was already so much to learn about the practical details.&lt;/p&gt;&#10;&lt;p&gt;After the debate ended, I kept on reading about the subject on the side. In particular, I read more on advanced reactor designs, where I found a neat theoretical question which I believe was unanswered.&lt;/p&gt;&#10;&lt;p&gt;I stumbled upon the problem that interested me in the last 6-9 months by reading Sylvain David&amp;rsquo;s PhD thesis on accelerator driven subcritical reactors (ADSR). Such reactors, popularized by Nobel Prize winner Carlo Rubia, work with a fissile core that is not critical, and where the chain reaction thus dies off exponentially. Before explaining why it is interesting and what questions it brings, I think it makes sense to explain how a standard reactor works.&lt;/p&gt;&#10;&lt;p&gt;Standard reactors have to be exactly critical, that is each fission should produce exactly one fission on average. Away from this point the chain reaction dies off or multiplies exponentially fast. It is quite surprising that this sweet spot, where the output power is flat, can be reached in practice. This is doable by exploiting primarily two mechanisms: the existence of delayed neutrons, and the temperature dependence of reactivity. Delayed neutrons explain why the timescales involved are manageable. Such neutrons are not immediately produced by the fission (like prompt neutrons), but by the subsequent decay of fission products. The latter takes place on much longer time scales (seconds, minutes, or more). A standard reactor is slightly subcritical in prompt neutrons alone, and gets critical only with delayed neutrons. The latter are only a small fraction of all neutrons, but this fraction is sufficient to create a narrow band of slow controllable evolution around criticality. This allows a real time tuning of reactivity by operators. In addition, for typical fuel loadings, an increase in temperature tends to reduce reactivity, which has a stabilizing effect.&lt;/p&gt;&#10;&lt;p&gt;This all works because of delayed neutrons, which are highly dependent on the type of fuel used. In particular, adding minor actinides (an annoying part of nuclear waste) into the fuel to burn them, tends to reduce the fraction of delayed neutrons. This a priori forbids making a pure waste burner: a critical reactor working with minor actinides as only fissile fuel. The latter would have too few delayed neutrons to be controllable.&lt;/p&gt;&#10;&lt;p&gt;One objective of subcritical reactors is to be less dependent on the characteristics of the fuel for safety. Such reactors are subcritical even accounting for delayed neutrons, the chain reaction always decays exponentially fast (and this &amp;ldquo;fast&amp;rdquo; is really fast, since the timescale is given by prompt neutrons). For such a reactor to sustain fissions and produce power, it needs to be externally driven by an intense source. The only source that is intense enough, with current designs, is a spallation source driven by an extremely high current proton accelerator. When a fast proton hits a heavy nucleus target like lead, the shock frees neutrons, which can then induce fissions in a subcritical core. Quite surprisingly (at least for me) this is energetically feasible: if the fission energy is turned into electricity, there is more than enough to drive the accelerator and give some to the grid. However, to make electricity production or waste incineration meaningful, the accelerator has to have a proton current that is higher than what can be found off shelf. This makes the accelerator part of such reactors quite expensive, tricky to construct, and potentially unreliable. It&amp;rsquo;s not the only objection to such designs, and more generally to a waste transmutation strategy, but it is certainly one that makes testing a realistic prototype complex and costly.&lt;/p&gt;&#10;&lt;p&gt;The question of the source requirements for subcritical reactors is really interesting and more subtle than it may seem. The main parameter for a subcritical reactor is its level of subcriticality, which is one proxy for its safety. Typically it is proposed in the range 0.95-0.99, meaning that one neutron injected produces on average and asymptotically 0.95 to 0.99 new neutrons per generation. It also suggests that for one neutron injected by the source, 20 (for &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;0.95&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}=0.95&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;0.95&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;) to 100 (for &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;0.99&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}=0.99&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;0.99&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;) fissions occur in the core. This rule of thumb sets the accelerator requirements for a given fission power output. What is surprising, and that I first discovered in Sylvain David&amp;rsquo;s thesis, is that there is in fact no rigorous link whatsoever between keff and the source requirements. More shockingly, for a fixed &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, say &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;0.97&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}=0.97&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;0.97&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, it is in principle possible have an arbitrarily faint source produce an arbitrarily large number of fissions. This seems outrageously wrong, which is of course why it got me very curious. What I did in the last months was understanding better how this is possible, and demonstrating that it is doable with physically realistic materials.&lt;/p&gt;&#10;&lt;p&gt;The main idea to get is that criticality is an asymptotic property while the number of fissions per source neutron is related to transcients. To be more precise, we need to consider to the average multiplication per generation &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_i&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;. The source neutron produces on average &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_1&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; fissions, which then produces on average &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_2&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; fissions, etc. Now there are two important quantities to consider.&lt;/p&gt;&#10;&lt;p&gt;The first is the &lt;em&gt;asymptotic multiplication&lt;/em&gt;&lt;/p&gt;&#10;&lt;span class="katex-display"&gt;&lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML" display="block"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mo&gt;&lt;mover&gt;&lt;mo&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;/mo&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;d&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;.&lt;/mi&gt;&lt;/mrow&gt;&lt;/mover&gt;&lt;/mo&gt;&lt;munder&gt;&lt;mrow&gt;&lt;mi&gt;lim&lt;/mi&gt;&lt;mo&gt;⁡&lt;/mo&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;mo&gt;→&lt;/mo&gt;&lt;mo&gt;+&lt;/mo&gt;&lt;mi mathvariant="normal"&gt;∞&lt;/mi&gt;&lt;/mrow&gt;&lt;/munder&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;&#10;k_\mathrm{eff}=\overset{\mathrm{def.}}{=} \lim_{i\rightarrow +\infty} k_i&#10;&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.153em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;&lt;span class="mop op-limits"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:1.153em;"&gt;&lt;span style="top:-3em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span&gt;&lt;span class="mop"&gt;=&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.5669em;margin-left:0em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight"&gt;def.&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.4804em;vertical-align:-0.786em;"&gt;&lt;/span&gt;&lt;span class="mop op-limits"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.6944em;"&gt;&lt;span style="top:-2.3723em;margin-left:0em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;span class="mrel mtight"&gt;→&lt;/span&gt;&lt;span class="mord mtight"&gt;+&lt;/span&gt;&lt;span class="mord mtight"&gt;∞&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span&gt;&lt;span class="mop"&gt;lim&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.786em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.1667em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;p&gt;which tells us about criticality. If &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;&amp;gt;&lt;/mo&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff} &amp;gt; 1&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;&amp;gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; the number of neutrons will explode exponentially with a timescale proportional to &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;log&lt;/mi&gt;&lt;mo&gt;⁡&lt;/mo&gt;&lt;mo stretchy="false"&gt;(&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo stretchy="false"&gt;)&lt;/mo&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\log (k_\mathrm{eff})&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1em;vertical-align:-0.25em;"&gt;&lt;/span&gt;&lt;span class="mop"&gt;lo&lt;span style="margin-right:0.0139em;"&gt;g&lt;/span&gt;&lt;/span&gt;&lt;span class="mopen"&gt;(&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mclose"&gt;)&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;. Conversely, if &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;&amp;lt;&lt;/mo&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff} &amp;lt; 1&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;&amp;lt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, then the number of neutron dies out at long times, and &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;log&lt;/mi&gt;&lt;mo&gt;⁡&lt;/mo&gt;&lt;mo stretchy="false"&gt;(&lt;/mo&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo stretchy="false"&gt;)&lt;/mo&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\log (- k_\mathrm{eff})&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1em;vertical-align:-0.25em;"&gt;&lt;/span&gt;&lt;span class="mop"&gt;lo&lt;span style="margin-right:0.0139em;"&gt;g&lt;/span&gt;&lt;/span&gt;&lt;span class="mopen"&gt;(&lt;/span&gt;&lt;span class="mord"&gt;−&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mclose"&gt;)&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; again gives the associated timescale for the decay. This is the metric one needs to look at for safety.&lt;/p&gt;&#10;&lt;p&gt;On the other hand, what is crucial for neutron economy and amplification is the total number of neutrons produced&lt;/p&gt;&#10;&lt;span class="katex-display"&gt;&lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML" display="block"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;mi&gt;N&lt;/mi&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/msub&gt;&lt;mo&gt;+&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/msub&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;/msub&gt;&lt;mo&gt;+&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/msub&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;/msub&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;/msub&gt;&lt;mo&gt;+&lt;/mo&gt;&lt;mo&gt;⋯&lt;/mo&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;munderover&gt;&lt;mo&gt;∑&lt;/mo&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mo&gt;+&lt;/mo&gt;&lt;mi mathvariant="normal"&gt;∞&lt;/mi&gt;&lt;/mrow&gt;&lt;/munderover&gt;&lt;munderover&gt;&lt;mo&gt;∏&lt;/mo&gt;&lt;mrow&gt;&lt;mi&gt;j&lt;/mi&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/munderover&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;j&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;&lt;mover&gt;&lt;mo&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;/mo&gt;&lt;mtext&gt;def.&lt;/mtext&gt;&lt;/mover&gt;&lt;/mo&gt;&lt;mfrac&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;mrow&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;s&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;/mfrac&gt;&lt;mtext&gt; &lt;/mtext&gt;&lt;mi mathvariant="normal"&gt;.&lt;/mi&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;&#10;N = k_1 + k_1 k_2 + k_1 k_2 k_3 + \cdots = \sum_{i=1}^{+\infty} \prod_{j=1}^i k_j \overset{\text{def.}}{=} \frac{1}{1-k_s} \, .&#10;&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6833em;"&gt;&lt;/span&gt;&lt;span class="mord mathnormal" style="margin-right:0.109em;"&gt;N&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;span class="mbin"&gt;+&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;span class="mbin"&gt;+&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3011em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;span class="mbin"&gt;+&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.3669em;"&gt;&lt;/span&gt;&lt;span class="minner"&gt;⋯&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:3.2254em;vertical-align:-1.4138em;"&gt;&lt;/span&gt;&lt;span class="mop op-limits"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:1.7583em;"&gt;&lt;span style="top:-1.8723em;margin-left:0em;"&gt;&lt;span class="pstrut" style="height:3.05em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;span class="mrel mtight"&gt;=&lt;/span&gt;&lt;span class="mord mtight"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.05em;"&gt;&lt;span class="pstrut" style="height:3.05em;"&gt;&lt;/span&gt;&lt;span&gt;&lt;span class="mop op-symbol large-op"&gt;∑&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-4.3em;margin-left:0em;"&gt;&lt;span class="pstrut" style="height:3.05em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mtight"&gt;+&lt;/span&gt;&lt;span class="mord mtight"&gt;∞&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:1.2777em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.1667em;"&gt;&lt;/span&gt;&lt;span class="mop op-limits"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:1.8117em;"&gt;&lt;span style="top:-1.8723em;margin-left:0em;"&gt;&lt;span class="pstrut" style="height:3.05em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathnormal mtight" style="margin-right:0.0572em;"&gt;j&lt;/span&gt;&lt;span class="mrel mtight"&gt;=&lt;/span&gt;&lt;span class="mord mtight"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.05em;"&gt;&lt;span class="pstrut" style="height:3.05em;"&gt;&lt;/span&gt;&lt;span&gt;&lt;span class="mop op-symbol large-op"&gt;∏&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-4.3em;margin-left:0em;"&gt;&lt;span class="pstrut" style="height:3.05em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:1.4138em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.1667em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight" style="margin-right:0.0572em;"&gt;j&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2861em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;&lt;span class="mop op-limits"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:1.153em;"&gt;&lt;span style="top:-3em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span&gt;&lt;span class="mop"&gt;=&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.5669em;margin-left:0em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord text mtight"&gt;&lt;span class="mord mtight"&gt;def.&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:2.1574em;vertical-align:-0.836em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mopen nulldelimiter"&gt;&lt;/span&gt;&lt;span class="mfrac"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:1.3214em;"&gt;&lt;span style="top:-2.314em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;span class="mbin"&gt;−&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2222em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1514em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;s&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.23em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span class="frac-line" style="border-bottom-width:0.04em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.677em;"&gt;&lt;span class="pstrut" style="height:3em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.836em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mclose nulldelimiter"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.1667em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;.&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;p&gt;Importantly, N can be very large if the first &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_i&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; are larger than 1. One sometimes introduces &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;s&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_s&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1514em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;s&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, which is what the multiplication per generation &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_i&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; would have had to be to get the same total number of neutrons had it been generation independent. If &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_i&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; is approximately generation independent, then &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;s&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_s=k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1514em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;s&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, but in general it is not the case.&lt;/p&gt;&#10;&lt;p&gt;The typical example is a just-barely-subcritical plutonium sphere with a source in the center: clearly the first generations, near the source, can have &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;&amp;gt;&lt;/mo&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_i&amp;gt;1&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;&amp;gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; as the neutrons are likely to meet plutonium nuclei, but &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;&amp;lt;&lt;/mo&gt;&lt;mn&gt;1&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_i&amp;lt;1&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3117em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;i&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;&amp;lt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; for the subsequent generations as neutrons are more likely to leak away. This makes it intuitive that &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mi&gt;s&lt;/mi&gt;&lt;/msub&gt;&lt;mo mathvariant="normal"&gt;≠&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_s \neq k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8889em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.1514em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mathnormal mtight"&gt;s&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;&lt;span class="mrel"&gt;&lt;span class="mord katex-vbox"&gt;&lt;span class="katex-thinbox"&gt;&lt;span class="rlap"&gt;&lt;span class="katex-strut" style="height:0.8889em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="katex-inner"&gt;&lt;span class="mord"&gt;&lt;span class="mrel"&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-fix"&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace nobreak"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, but does not yet show they can be arbitrarily different.&lt;/p&gt;&#10;&lt;p&gt;Typically, if you put together two pieces of fissile material that independently have a &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; close to 1, say &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mn&gt;0.97&lt;/mn&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}=0.97&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;span class="mrel"&gt;=&lt;/span&gt;&lt;span class="mspace" style="margin-right:0.2778em;"&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.6444em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;0.97&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, the pair will become critical, so you can&amp;rsquo;t just increase multiplication this way. The trick is to create some decoupling in the neutron flux. Imagine you have a perfect neutron diode. Now put the two pieces together and separate them with the diode. If there is a source in the first piece, it will create a few neutrons that will leak away into the second one through the diode, where they will create more fissions and neutrons. However, the neutrons created in the second piece can&amp;rsquo;t flow back into the first. It is thus clear that the &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; of the pair is the same as the original. The &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; is the same, but the multiplication has increased. If we go from 2 to n pieces, the multiplication of the stack is increased exponentially in n while &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; is constant.&lt;/p&gt;&#10;&lt;p&gt;The problem is that there are no perfect neutron diodes in nature. In 1957 people had proposed combining cadmium, a thermal neutron absorber, and a moderator: fissions produce fast neutrons, that pass through the cadmium as if it were transparent, then go through a moderator, and then induce new fissions in a new stage. The neutrons from the second fission stage can go back through the moderator, but are then thermalized and thus get absorbed by cadmium. This produces an approximate diode.&lt;/p&gt;&#10;&lt;p&gt;How far can this be pushed? As far as I can tell from carrying a rather tedious review of the literature, not much has been tried. Some physicists have made back of the envelope calculations about what multiplication could be achieved assuming some reflection and transmission coefficients between stages are given. The multiplication reached per stage is huge but of course the result is rather trivial since the assumptions are ad hoc and it&amp;rsquo;s hard to know if they are realistic (with known materials and reasonable reactor geometry). On the other hand, some physicists have tried to improve existing reactor designs by making them two stage. This is the other extreme: very realistic, since it&amp;rsquo;s a full reactor model, but the gain is fixed and moderate, because there are only two stages. I wanted to fill the space in between: without making a super realistic reactor, is it possible to achieve arbitrary multiplication with a simple scalable design made of materials that exist? Basically I wanted to be physically realistic, without necessarily being realistic from an engineering perspective.&lt;/p&gt;&#10;&lt;p&gt;I found that they were basically 4 physically distinct strategies, 3 being usable if the design is to be scalable (by that I mean that it is possible stack as many stages as one likes). I then dreamed of a design incorporating all 3 strategies, and carried neutron transport Monte-Carlo simulations to see how it fared. Since I thought the results were cool, I wrote them down in &lt;a href="https://atilloy.com/wp-content/uploads/2020/08/neutron_multiplier_tilloy_200803.pdf"&gt;an article draft&lt;/a&gt; (comments are very welcome, since I am a bit out of my turf).&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/08/set.png"&#10;&#9;&#9;&#9;alt="set" width="469"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;The multilayer setup I considered&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;In a nutshell, arbitrary amplification is feasible at fixed &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; , with known fissile materials, reflectors and moderators. This opens the way, in principle, to have subcritical reactors driven by arbitrarily faint sources, even based on radioactive decay. Of course, the &amp;ldquo;thing&amp;rdquo; I simulated is far from a complete reactor, and its nice feature (arbitrarily large subcritical amplification) is paid with other safety problems that may be unsolvable. Yet I think it deserves to be explored further and I would like better designs to be proposed. Sure, such fancy designs will probably never be implemented, but from a theoretical point of view they are stimulating. For example, I don&amp;rsquo;t think it is known how low &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; can be while allowing arbitrary amplification in a multistage system. In my simulations, I had to keep &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; quite close to criticality: unless the reactor is taken huge, it is difficult to have &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mrow&gt;&lt;mi mathvariant="normal"&gt;e&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;k_\mathrm{eff}&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8444em;vertical-align:-0.15em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal" style="margin-right:0.0315em;"&gt;k&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.3361em;"&gt;&lt;span style="top:-2.55em;margin-left:-0.0315em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;&lt;span class="mord mathrm mtight" style="margin-right:0.0778em;"&gt;eff&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.15em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; below 0.95 at least with the strategies I manually explored. Could one go as low as 0.8 for example, to make the reactor robust to massive changes in geometry, coolant loss, etc.? How small can a subcritical system amplifying neutrons by a factor &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mn&gt;10&lt;/mn&gt;&lt;mn&gt;9&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;10^9&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:0.8141em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;1&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord"&gt;0&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;9&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; be? Can such subcritical systems be used as pulsed neutron sources, as alternatives to high flux critical reactors?&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/08/amp.png"&#10;&#9;&#9;&#9;alt="amp" width="382"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Number of fissions per stage and neutron flux in the multilayer system I considered&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;As far as I know, these rather fundamental questions have not been explored much. It is possible that practionners have found a long time ago that there were major drawbacks to multistage systems that I have not seen. People I talked to didn&amp;rsquo;t know of any show stopper. If anyone of my readers has hints, or wants to explore related questions further, I am of course very interested in discussing.&lt;/p&gt;&#10;&lt;p&gt;To summarize, it&amp;rsquo;s physically possible to have a strictly subcritical reactor of standard power driven by an arbitrarily weak neutron source. This brings a lot of interesting theoretical questions, and makes me wonder what the main drawbacks are. For more, just read &lt;a href="https://atilloy.com/wp-content/uploads/2020/08/neutron_multiplier_tilloy_200803.pdf"&gt;the draft&lt;/a&gt;, which I hope to finalize after the holidays.&lt;/p&gt;&#10;</description></item><item><title>The sound of quantum jumps</title><link>https://atilloy.com/2020/07/31/the-sound-of-quantum-jumps/</link><guid>https://atilloy.com/2020/07/31/the-sound-of-quantum-jumps/</guid><pubDate>Fri, 31 Jul 2020 09:36:26 +0200</pubDate><description>&lt;p&gt;Before I finally go on holidays, I put on arxiv &lt;a href="https://arxiv.org/abs/2007.15420"&gt;an essay on quantum jumps&lt;/a&gt;, in fact rather on collapse models, that I initially submitted to the FQXI essay contest.&lt;/p&gt;&#10;&lt;p&gt;In this essay/paper, I just make a simple point which I have made orally for years at conferences. Every time, people looked quite surprised and so I thought it made sense to write it down.&lt;/p&gt;&#10;&lt;p&gt;The argument is simple enough that I can try to reproduce it here. Collapse models are stochastic non-linear modifications of the Schrödinger equation. The modification is meant to solve the measurement problem. The measurement problem is the fact that in ordinary quantum mechanics, what happens in measurement situations is postulated rather than derived from the dynamics. This is a real problem (contrary to what some may say), the dynamics should say what can be measured and how, it makes no sense to have an independent axiom (and it could bring contradictions). Decoherence explains why the measurement problem does not bring contradictions for all practical purposes, but again contrary to what some may say, it certainly does not solve the measurement problem. So the measurement problem is a real problem and collapse models provide a solution that works.&lt;/p&gt;&#10;&lt;p&gt;The stochastic non-linearity brought by collapse models creates minor deviations from the standard quantum mechanical predictions (it makes sense, the dynamics has been modified). This is often seen, paradoxically perhaps, as a good thing, because it makes the approach falsifiable. It is true that collapse models are falsifiable. What is not true, is that collapse models modify the predictions of quantum mechanics understand broadly. This is what is more surprising, sometimes seems contradictory with the previous point, and is the subject of my essay.&lt;/p&gt;&#10;&lt;p&gt;How is it possible? Collapse models are non-linear and stochastic, surely ordinary quantum mechanics cannot reproduce that? But in fact it can. As was understood when collapse models were constructed in the eighties, the non-linearity of collapse models, which is useful to solve the measurement problem, has to vanish upon averaging the randomness away. Since we have no a priori access to this randomness, all the things we can measure in practice can be deduced from linear equations, even in the context of collapse models. This linear equation is not the Schrödinger equation, but one that does not preserve purity, the Lindblad equation. However, it is also known that by enlarging the Hilbert space (essentially assuming hidden particles), Lindblad dynamics can be reproduced by Schrödinger dynamics. Hence, the predictions of collapse models can always be reproduced exactly by a purely quantum theory (linear and deterministic) at the price enlarging the Hilbert space with extra degrees of freedom. Collapse models do not deviate from quantum theory, they deviate from the Standard Model of particle physics, which is an instantiation of quantum theory. Even if experiments showed precisely the kind of deviations predicted by collapse models, one could still defend orthodox quantum mechanics (not that it would necessarily be advisable to do so).&lt;/p&gt;&#10;&lt;p&gt;Collapse models are still useful in that they solve the measurement problem, which is an ontological problem (what the theory says the world is like or what the world is made of). However, the empirical content of collapse models (what the theory predicts) is less singular that one might think. In the essay, I essentially make this point in a more precise way, and illustrate it on what I believe is the most shocking example, the &lt;em&gt;sound of quantum jumps&lt;/em&gt;, borrowed from a &lt;a href="https://arxiv.org/abs/1109.6579"&gt;paper by Feldmann and Tumulka&lt;/a&gt;. I doesn&amp;rsquo;t make sense to write more here since I will end up paraphrasing the essay, but I encourage whoever is interested to read it &lt;a href="https://arxiv.org/abs/2007.15420"&gt;here&lt;/a&gt;.&lt;/p&gt;&#10;</description></item><item><title>The skyscraper and pile of dirt approaches to QFT</title><link>https://atilloy.com/2020/06/09/the-skyscraper-and-pile-of-dirt-approaches-to-qft/</link><guid>https://atilloy.com/2020/06/09/the-skyscraper-and-pile-of-dirt-approaches-to-qft/</guid><pubDate>Tue, 09 Jun 2020 10:18:01 +0200</pubDate><description>&lt;p&gt;Quantum field theory (QFT) is the main tool we use to understand the fundamental particles and their interactions. It also appears in the context of condensed matter physics, as an effective description. But it is unfortunately also a notoriously difficult subject: first because it is tricky to define non-trivial instances rigorously (it&amp;rsquo;s not known for any one that exists in Nature), and also because even assuming it can be done, it is then very difficult to solve to extract accurate predictions.&lt;/p&gt;&#10;&lt;p&gt;There is a subset of QFTs where there is no difficulty: free QFTs. Free QFTs are easy because one can essentially define them in a non-rigorous way first, physicist style, then &amp;ldquo;solve&amp;rdquo; them exactly, and finally take the solution itself as a rigorous definition of what we actually meant in the first place. Then, to define the interacting theories, the historical solution has been to see them as perturbations of the free ones. This comes with well known problems: interacting theories are not as close to free ones as one would naively think, so the expansions one obtains are weird: they diverge term by term, and if the divergences are subtracted in a smart way (renormalization), the expansions still diverge as a whole.&lt;/p&gt;&#10;&lt;p&gt;This has not been such a terrible problem historically because in the regime where we observe it, the most important force between particles, the electromagnetic force, is &amp;ldquo;weak&amp;rdquo; in the sense of perturbation theory, and thus perturbative expansions still give accurate results at least at reasonably low energy. Renormalizing term by term and truncating the series to the first few terms gives numbers that match what is measured, at least for what can be probed with particle accelerators fitting on earth. Then the problem that maybe the theory itself does not exist is evacuated as secondary. However, for the strong force, described by quantum chromodynamics (QCD) this does not work. Perturbation theory is inaccurate at least at low energy, where quarks get strongly bound and are not approximately non-interacting. So it is difficult to make computations, and as a result the question of whether or not it is even possible in principle to define the theory becomes important: if you can no longer compute stuff, you are not as sure that the object you manipulate even makes sense. In fact, one of the Millenium prizes of 1 million dollars will be awarded to whoever can define a quantum field theory like QCD rigorously. It&amp;rsquo;s probably the most difficult way to make 1 million dollars.&lt;/p&gt;&#10;&lt;p&gt;So, to summarize, there exist specific models of quantum field theories like quantum chromodynamics, that are useful in the real world, that we are more or less convinced are right (in some regime, the model predicts exactly what is observed) but that we don&amp;rsquo;t know how to define rigorously and for which the standard computation methods (which we could potentially use in a yolo fashion) don&amp;rsquo;t work.&lt;/p&gt;&#10;&lt;p&gt;Now we get to what I find interesting and perhaps insufficiently mentioned usually: this problem of QFT has been attacked from two completely different angles, by two different communities, and with a radically different philosophy. In a nutshell the philosophies are:&lt;/p&gt;&#10;&lt;ol&gt;&#10;&lt;li&gt;&lt;em&gt;Make the problem simpler&lt;/em&gt;, hoping the tools you build to define and solve the simple instances are more or less generic, and work your way up to real QCD.&lt;/li&gt;&#10;&lt;li&gt;&lt;em&gt;Make the problem more complex&lt;/em&gt;, so that it can miraculously be solved exactly thanks to the extra structure, and progressively reduce or deform the structure to get to real QCD.&lt;/li&gt;&#10;&lt;/ol&gt;&#10;&lt;p&gt;I won&amp;rsquo;t hide that I prefer the first approach: it&amp;rsquo;s the one I am most interested in and to which I very modestly tried to contribute recently. However, the second one deserves to be presented because I think it is paradoxically the most mainstream, at least in terms of fame of those who defend it, and perhaps the most surprising. It&amp;rsquo;s one that illuminates best the mindset of theoretical physicists and high-energy physicists in particular.&lt;/p&gt;&#10;&lt;p&gt;At first sight it seems weird that making things more complex makes the problem solvable. But I think it&amp;rsquo;s a quite standard pattern in theoretical physics, at least in the last century. Typically you start from a simple non-trivial model. You realize that the things you would want to compute can be expressed as a sum of infinitely many complicated terms. It is hopeless, too generic, almost random. Then, you complexify the model to add &amp;ldquo;structure&amp;rdquo; in such a way that this infinite mess orders itself into a few simple terms, through magical cancellations, or because all the terms now make a geometric series. At the same time you can use additional structure to cure divergences in a neater way than with standard renormalization. The most famous extra structure is supersymmetry, but in lower dimensions (1+1 instead of 3+1) there is also an infinite family of quantum field theories precisely designed to make the difficulty go away (the so called &amp;ldquo;integrable&amp;rdquo; QFTs).&lt;/p&gt;&#10;&lt;p&gt;The idea of this approach is to reproduce the &amp;ldquo;exactly solvable free field theory&amp;rdquo; + &amp;ldquo;perturbation theory&amp;rdquo; miracle, but this time with new candidates for the starting points, exactly solvable theories. As for free field theories, exactly solvable theories can be defined (more or less rigorously) implicitly through their solutions, so this problem seems dealt with as well. This approach has mesmerized physicists: the infinite richness of exactly solvable theories has provided nice qualitative understanding of some weird phenomena like quark confinement. However, I think it will not go beyond that, because we see in these models only the structure we put in. This ordering, this simplification of the mess is precisely what does not happen in most of the real quantum field theories we see in Nature.&lt;/p&gt;&#10;&lt;p&gt;It&amp;rsquo;s a bit like trying to learn about mountains and volcanoes by studying the Eiffel tower or the Chrysler building. They are all pointy things going upward, but what we learn from one by studying the other is necessarily limited. Further the Chrysler building has beautiful metallic gargoyles, which end up attracting all the attention. In the same way, the idiosyncrasies of the structured solvable models become the center of attention, an interest on their own, and we end up forgetting what we came here for.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/06/new_york_city_chrysler_building_02.jpg"&#10;&#9;&#9;&#9;alt="New_York_City_Chrysler_Building_02" width="307"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;One of the ornaments of the Chrysler building - Norbert Nagel / Wikimedia Commons CC-BY-SA 3.0&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;A historical hope was that, deep down, Nature was like the Chrysler building, that there was structure all the way down. But there is not much empirical evidence for this belief. It seems the complex models that make things simpler are really man-made, and don&amp;rsquo;t describe what we see. Now I would say most people in the field no longer use this argument as a defense of the study of highly structured theories, but rather say that these well understood theories give us a window into what really happens in those, more realistic, that we cannot solve. Ultimately however, I think we all get too fascinated by the gargoyles.&lt;/p&gt;&#10;&lt;p&gt;The alternative is to really make things simpler, but try to keep the mess that makes the theories what they are. It&amp;rsquo;s like studying a pile of dirt, then a small hill, in the hope that we will understand mountains one day. Down this road, there is no discovery of rich structure, and no &amp;ldquo;two birds one shot&amp;rdquo; miracle of defining the theory through its exact solution. You keep two distinct problems, and you lose the shininess of polished chromium. The first steps were made by the warrior monks of theoretical physicists, the constructive field theorists, starting in the seventies. They focused on defining the theories rigorously, and managed to go up to the self-interacting scalar field in 2+1 dimensions, aka &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, which is to QCD what Zugspitze is to the Everest. &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; is not QCD, but it does not cheat, a large part of the mess is still there. But then going beyond this remarkable achievement turned out difficult. The mess of &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; is maximal is you want to solve it, but if you merely want to define the theory, the difficulty is milder than for QCD. Technically, the theory is superrenormalizable, whereas QCD is just renormalizable. So far no one managed to get results as strong as what was obtained for &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; while removing the &amp;ldquo;super&amp;rdquo;. Unfortunately adding one dimension to the self interacting scalar field, with &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_4&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, which would constitute the natural next step and bring us to the right number of physical dimensions, is not possible because while renormalizable in the usual sense, it turns out that the theory does not exist. This is a subtle result: while &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_4&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; was long suspected to be ill-defined, the model was definitely slayed by Aizenman and Duminil-Copin only a few months ago.&lt;/p&gt;&#10;&lt;p&gt;The constructive field theorists solved part of the problem (the definition), without cheating, but only up to a point. This is a success or a failure depending on how you look at it. To me it&amp;rsquo;s mostly a success. However, computing what the models predict, not just defining them, remains to be done.&lt;/p&gt;&#10;&lt;p&gt;Because of the limitations of the perturbative approach to QCD, the problem of computing its predictions in the strongly coupled regime was tackled by the hyper pragmatic lattice gauge theorists. They went for bruteforce and used the universal bazooka of physics, the Monte-Carlo method, to get pretty good estimates of many quantities of interest (like the masses of composite particles). This is a remarkable achievement. The Monte Carlo method is universal, it thrives in the mess and absence of structure, it works in theory for basically any physics problem. It is the ultimate joker, what to try when all hope is lost. But it is a costly joker. With arrays of supercomputers scattered all around the world, the lattice gauge theorists only managed to reach a limited precision, and increasing the quality of the estimates will require a tremendous increase in computational power.&lt;/p&gt;&#10;&lt;p&gt;Until recently, Monte-Carlo was the only game in town. It still is if one wants to attack the mountain of QCD directly, but what about the smaller hills the constructive field theorists made sure exist? There was an increase of interest recently, which I mentioned in a previous post. It&amp;rsquo;s difficult to build new methods directly for QCD, but for simpler theories like &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_2&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; which are brutal simplifications, it is manageable. I think it started being interesting when Slava Rychkov and his collaborators managed to beat Monte-Carlo with a method called renormalized Hamiltonian truncation (RHT), for this very simple model. Then the Monte-Carlo people improved their simulations, and beat RHT again. The perturbation theory people pushed their method to the edge by Borel resumming their expansions to order 8, and arrived roughly in the same league. Meanwhile, the tensor network approach, which we develop on in Munich, was used by other groups to reach roughly the same precision as the latest generation Monte-Carlo. Together with Clément Delcamp at the Max Planck Institute, we found this game fun and used the most advanced tensor network renormalization method together with a smarter discretization to obtain &lt;a href="https://arxiv.org/abs/2003.12993"&gt;slightly more precise results&lt;/a&gt; than every one else, at least until another method beats us in the next few months.&lt;/p&gt;&#10;&lt;p&gt;Of course, we are falling into the same kind of trap as the gargoyle fans. We image to amazing precision a small hill that is not what we ultimately care about. It&amp;rsquo;s a good benchmark, since it does not cheat with extra structure, but its relevance is limited because all the methods we use (apart from Monte-Carlo) do not trivially extend to more complicated problems. Recently, the RHT people have &lt;a href="https://arxiv.org/abs/2003.08405"&gt;boldly attacked&lt;/a&gt; &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, which is basically as far as the constructive field theorists went. We expect our tensor networks to be competitive there as well, but clearly we need to work more for this to happen. Meanwhile QCD remains ruled by the lattice gauge theorists, and will likely remain so for quite some time.&lt;/p&gt;&#10;&lt;p&gt;Quantum field theory, in the number of dimensions of the real world (3+1), remains impossible to define and hard to solve away from trivial theories. The high-energy theorists tried to find the answer in structure, to bypass the mess. This is the skyscraper approach. The constructive field theorists tried to make sure the whole enterprise made sense and the theories existed. The lattice gauge theorists attacked the problem head on obtaining results at a massive computational cost. Finally smaller communities started competing to solve the simpler models the constructive field theorists had defined, hoping to one day escape the tremendous computational thirst of Monte-Carlo. I put the last one in the pile of dirt approach. Less shiny for sure, but the mess has to be dealt with.&lt;/p&gt;&#10;</description></item><item><title>Bien plus qu’il n’en faut sur le cycle du combustible nucléaire et le taux de recyclage associé</title><link>https://atilloy.com/2020/01/14/bien-plus-quil-nen-faut-sur-le-cycle-du-combustible-nucleaire-et-le-taux-de-recyclage-associe-soon-online/</link><guid>https://atilloy.com/2020/01/14/bien-plus-quil-nen-faut-sur-le-cycle-du-combustible-nucleaire-et-le-taux-de-recyclage-associe-soon-online/</guid><pubDate>Tue, 14 Jan 2020 18:56:39 +0100</pubDate><description>&lt;p&gt;&lt;em&gt;Les combustibles nucléaires usés issus des réacteurs électrogènes français sont recyclés. D&amp;rsquo;aucuns arguent que ce recyclage permet de récupérer 96% du combustible, l&amp;rsquo;industrie est très vertueuse et exemplaire ! D&amp;rsquo;autres disent au contraire que le recyclage ne permet qu'1% de réutilisation, le recyclage c&amp;rsquo;est du bullshit ! Qui a raison ?&lt;/em&gt;&lt;/p&gt;&#10;&lt;p&gt;&lt;em&gt;Je trouve que cette controverse est un bon prétexte pour expliquer la physique associée, qui est intéressante. Mon objectif est d&amp;rsquo;expliquer en détails l&amp;rsquo;aval du cycle du combustible pour son intérêt propre, la résolution de la controverse étant ensuite un corollaire trivial. Au passage, c&amp;rsquo;est l&amp;rsquo;occasion d&amp;rsquo;apprendre un peu plus sur l&amp;rsquo;histoire civile et militaire du nucléaire, le principe d&amp;rsquo;un réacteur nucléaire, et les subtilités des différents isotopes de l&amp;rsquo;uranium et du plutonium.&lt;/em&gt;&lt;/p&gt;&#10;&lt;h2 id="la-fission-nucléaire-et-la-réaction-en-chaîne"&gt;La fission nucléaire et la réaction en chaîne&lt;/h2&gt;&#10;&lt;h3 id="la-base"&gt;La base&lt;/h3&gt;&#10;&lt;p&gt;Les réacteurs à eau (légère), qui sont ultra majoritaires dans le monde, produisent de la chaleur par fission d&amp;rsquo;uranium enrichi. D&amp;rsquo;abord pourquoi de l&amp;rsquo;eau, et pourquoi enrichi ?&lt;/p&gt;&#10;&lt;p&gt;Dans la nature, l&amp;rsquo;uranium contient environ 0.7 % d&amp;rsquo;uranium 235 et 99.3% d&amp;rsquo;uranium 238. Seul le premier, le 235 est fissile, c&amp;rsquo;est à dire qu&amp;rsquo;il se casse en produisant de l&amp;rsquo;énergie lorsqu&amp;rsquo;il est soumis à un flux de neutrons.&lt;/p&gt;&#10;&lt;p&gt;Lorsqu&amp;rsquo;un noyau d&amp;rsquo;uranium 235 fissionne, il produit 2 ou 3 neutrons qui peuvent eux mêmes faire fissionner un atome d&amp;rsquo;uranium 235, c&amp;rsquo;est la réaction en chaîne, qui permet de se passer d&amp;rsquo;une source externe de neutrons. Pour que la réaction en chaîne se maintienne, il faut qu&amp;rsquo;au moins un neutron en moyenne issu d&amp;rsquo;une fission engendre une autre fission.&lt;/p&gt;&#10;&lt;p&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/1000px-fission_chain_reaction.svg_.png" alt="1000px-Fission_chain_reaction.svg"&gt;&lt;/p&gt;&#10;&lt;p&gt;Qu&amp;rsquo;est-ce qui peut arriver d&amp;rsquo;autre au neutron ? Essentiellement, être capturé par un atome de 238 ou simplement sortir du bloc d&amp;rsquo;uranium. Plus le bloc est gros, moins on perd de neutrons, et donc plus on s&amp;rsquo;approche de la masse critique, c&amp;rsquo;est à dire le point où la réaction en chaîne se maintient. Le problème c&amp;rsquo;est que même pour un bloc de taille infinie, c&amp;rsquo;est à dire avec aucune perte de neutrons à l&amp;rsquo;extérieur, il faut au moins environ 10% d&amp;rsquo;uranium 235 pour atteindre la masse critique, sans quoi l&amp;rsquo;uranium 238 capture trop de neutrons. Bref on peut rassembler autant d&amp;rsquo;uranium naturel qu&amp;rsquo;on veut, on n&amp;rsquo;atteindra jamais la masse critique tout seul.&lt;/p&gt;&#10;&lt;p&gt;Face à ce problème, les physiciens ont trouvé deux solutions : l&amp;rsquo;&lt;em&gt;enrichissement&lt;/em&gt; et la &lt;em&gt;modération&lt;/em&gt;.&lt;/p&gt;&#10;&lt;h3&gt;&lt;/h3&gt;&#10;&lt;h3 id="atteindre-la-masse-critique"&gt;Atteindre la masse critique&lt;/h3&gt;&#10;&lt;p&gt;L&amp;rsquo;enrichissement est la solution la plus évidente : augmenter la concentration d&amp;rsquo;uranium 235 pour réduire le taux de capture par 238. C&amp;rsquo;est une procédure pas évidente car 235 et 238 sont chimiquement indistinguables et donc il faut ruser en exploitant leur légère différence de masse. Aujourd&amp;rsquo;hui, on fait cela par centrifugation et avant ça, on le faisait par diffusion gazeuse. C&amp;rsquo;était très cher et très énergivore, ça l&amp;rsquo;est un peu moins maintenant, mais ça reste une opération non triviale.&lt;/p&gt;&#10;&lt;p&gt;Une autre solution est de modérer (c&amp;rsquo;est à dire ralentir) les neutrons. Là on touche à de la physique franchement pas évidente. Il se trouve que lorsque les neutrons vont vite l&amp;rsquo;uranium 235 leur apparaît petit et 238 gros. Mais si les neutrons vont moins vite alors l&amp;rsquo;uranium 235 apparaît de plus en plus gros et 238 proportionnellement plus petit, et donc la probabilité de collision avec 235 augmente.&lt;/p&gt;&#10;&lt;p&gt;Pourquoi ? En général à ce moment on invoque la mécanique quantique, et la théorie quantique des champs, en faisant des grands gestes avec les mains. Je dois dire que ça reste peu évident pour moi, je sais que c&amp;rsquo;est un effet qu&amp;rsquo;on mesure mais je ne sais pas jusqu&amp;rsquo;à quel point on sait le dériver des équations fondamentales de la physique. On a probablement des modèles assez phénoménologiques sur la structure des noyaux pour l&amp;rsquo;expliquer, mais je n&amp;rsquo;en sais pas plus.&lt;/p&gt;&#10;&lt;p&gt;Pour ralentir les neutrons, il suffit de les faire passer dans un milieu qui les laisse rebondir pour qu&amp;rsquo;ils perdent progressivement leur énergie mais ne les capture pas trop. L&amp;rsquo;eau du robinet est un bon modérateur, et le graphite (des mines de crayon) et l&amp;rsquo;eau lourde (de l&amp;rsquo;eau où l&amp;rsquo;hydrogène est remplacé par son isotope lourd, le deuterium) en sont d&amp;rsquo;encore meilleurs.&lt;/p&gt;&#10;&lt;h3 id="retour-sur-lhistoire-du-nucléaire"&gt;Retour sur l&amp;rsquo;histoire du nucléaire&lt;/h3&gt;&#10;&lt;p&gt;Aux tous débuts du nucléaire, dans les années 40, les physiciens ont commencé par modérer, parce que c&amp;rsquo;était le plus simple. Les américains autour d&amp;rsquo;Enrico Fermi ont utilisé du graphite, et les allemands, principalement autour d&amp;rsquo;Heisenberg, ont utilisé l&amp;rsquo;eau lourde. Dans les deux cas avec un combustible d&amp;rsquo;uranium naturel.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/pile-e1579014368980.jpg"&#10;&#9;&#9;&#9;alt="scanned: May 2001 by Image Delivery Systems LLC" width="346"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Une des piles de graphite du projet Manhattan&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Le réacteur de Fermi, modéré au graphite, a donné naissance à la première génération de réacteurs nucléaires exploités à travers le monde, et notamment les réacteurs à uranium naturel graphite gaz (UNGG) français (aujourd&amp;rsquo;hui tous en démantèlement), et les tristement célèbres &lt;em&gt;Reaktor Bolshoy Moshchnosti Kanalnyi&lt;/em&gt; (RBMK) [qui fonctionnent en fait à l&amp;rsquo;uranium légèrement enrichi] dont ceux qui ont vu la série &lt;em&gt;Chernobyl&lt;/em&gt; connaissent désormais tout.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/saint_laurent.jpg"&#10;&#9;&#9;&#9;alt="saint_laurent" width="414"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Réacteurs A1 et A1 à l&amp;rsquo;uranium naturel (UNGG) de la centrale de Saint Laurent&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Si les nazis ne sont pas parvenus à fabriquer la bombe nucléaire, les réacteurs à eau lourde qu&amp;rsquo;ils exploraient ont plus tard été industrialisés et sont utilisés aujourd&amp;rsquo;hui par le Canada et l&amp;rsquo;Inde.&lt;/p&gt;&#10;&lt;p&gt;Aujourd&amp;rsquo;hui la France, comme la plupart des pays produisant de l&amp;rsquo;énergie nucléaire, exploite des réacteurs utilisant de l&amp;rsquo;eau comme modérateur et de l&amp;rsquo;uranium enrichi à environ 4% comme combustible. La masse critique est ainsi obtenue en combinant enrichissement et modération. Cette combinaison est celle des réacteurs de deuxième génération (comme les 58 réacteurs actuellement exploités en France) et de troisième génération (comme l&amp;rsquo;EPR).&lt;/p&gt;&#10;&lt;p&gt;En fait, du point de vue physique, cette combinaison correspond aussi à la génération 0 de réacteurs. Il y a environ 2 milliards d&amp;rsquo;années, l&amp;rsquo;uranium naturel avait une concentration plus forte en 235 (3.6%) qu&amp;rsquo;aujourd&amp;rsquo;hui (0.7%). Le 235 a une demi vie plus courte que le 238, et donc la concentration en 235 diminue au cours du temps. Par rapport à l&amp;rsquo;uranium actuel, l&amp;rsquo;uranium naturel d&amp;rsquo;il y a 2 milliards d&amp;rsquo;années était enrichi, à peu près autant que celui des centrales françaises. À Oklo au Gabon, des infiltrations d&amp;rsquo;eau (faisant office de modérateur) dans un gisement d&amp;rsquo;uranium ont ainsi permis de maintenir une réaction en chaîne pendant plusieurs centaines de milliers d&amp;rsquo;années.&lt;/p&gt;&#10;&lt;p&gt;En résumé, que ce soit dans la nature ou dans un réacteur, pour obtenir une réaction nucléaire auto-entretenue, il faut modérer, enrichir, ou les deux.&lt;/p&gt;&#10;&lt;p&gt;Avant de passer à la suite, et regarder ce qui sort d&amp;rsquo;un réacteur, on peut noter qu&amp;rsquo;on a déjà un sous produit de la fabrication de combustible à l&amp;rsquo;uranium enrichi : l&amp;rsquo;uranium appauvri. On part d&amp;rsquo;un uranium naturel à 0.7%, et après enrichissement environ un dixième devient de l&amp;rsquo;uranium enrichi à 4% et le reste de l&amp;rsquo;uranium appauvri à 0.3% (le taux est variable). Donc on produit beaucoup d&amp;rsquo;uranium appauvri, a priori inutile.&lt;/p&gt;&#10;&lt;p&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/uapp.png" alt="uapp"&gt;&lt;/p&gt;&#10;&lt;h3 id="le-combustible-usé"&gt;Le combustible usé&lt;/h3&gt;&#10;&lt;p&gt;Maintenant qu&amp;rsquo;est ce qui sort du réacteur après environ 3 ans de réactions ? Naïvement, à quoi peut-on s&amp;rsquo;attendre ? On a dû consommer une fraction de l&amp;rsquo;uranium 235 présent, et donc, en démarrant avec 100% d&amp;rsquo;uranium (lui même à 4%) de 235, on a dû passer typiquement à 97% d&amp;rsquo;uranium (à environ 1% de 235) et 3% de produits de fission, c&amp;rsquo;est à dire de fragments issus de la fission d&amp;rsquo;uranium.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/naive_composition.png"&#10;&#9;&#9;&#9;alt="naive_composition" width="344"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Composition naïve du combustible usé&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;En fait, ça n&amp;rsquo;est pas ce qu&amp;rsquo;on obtient. La concentration de l&amp;rsquo;uranium est effectivement tombée à 1% de 235 environ mais au lieu d&amp;rsquo;avoir 97% d&amp;rsquo;uranium au total on en a 95%. On a en plus 4% de produits de fission (au lieu des 3% attendus), 1% environ de plutonium et environ 0.1% de d&amp;rsquo;autres atomes lourds (les actinides mineurs). Que s&amp;rsquo;est il passé ?&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/realcomposition.png"&#10;&#9;&#9;&#9;alt="realcomposition" width="408"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Composition réelle du combustible usé&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;h2 id="transmutation-de-luranium-238"&gt;Transmutation de l&amp;rsquo;uranium 238&lt;/h2&gt;&#10;&lt;h3 id="principe"&gt;Principe&lt;/h3&gt;&#10;&lt;p&gt;Pour comprendre il faut revenir au tout début et les captures parasites de neutrons par l&amp;rsquo;uranium 238. L&amp;rsquo;idée de la modération était de les réduire pour favoriser les fissions de 235 et atteindre la masse critique. Mais il en reste malgré tout un certain nombre. Il faut maintenant suivre cet uranium 238. Lorsqu&amp;rsquo;il capture un neutron, l&amp;rsquo;uranium 238 devient uranium 239. C&amp;rsquo;est un noyau très instable qui se désintègre rapidement en neptunium 239 (désintégration beta -, essentiellement un neutron du noyau se transforme en proton). Le neptunium 239 est toujours instable et se désintègre ensuite en plutonium 239 qui est stable (au moins pour quelques dizaines de milliers d&amp;rsquo;années). Ainsi, à cause de ces captures, une partie de l&amp;rsquo;uranium 238, a priori spectateur, est &lt;em&gt;transmutée&lt;/em&gt; en plutonium 239.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/generation_pu239.jpg"&#10;&#9;&#9;&#9;alt="generation_pu239" width="500"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Génération du plutonium 239 par capture (&lt;a href="http://www.laradioactivite.com/site/pages/laformationduplutonium239.htm"&gt;IN2P3&lt;/a&gt;)&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Le plutonium 239 possède, en première approximation, les mêmes propriétés que l&amp;rsquo;uranium 235 du point de vue de la fission nucléaire. Les captures permettent ainsi de &amp;ldquo;régénérer&amp;rdquo; partiellement le combustible : à mesure que de l&amp;rsquo;uranium 235 est consommé, on produit du plutonium 239 essentiellement substituable, qui vient en partie compenser la perte de 235. L&amp;rsquo;uranium 238 n&amp;rsquo;est pas fissile, mais il est &lt;em&gt;fertile&lt;/em&gt;, c&amp;rsquo;est à dire qu&amp;rsquo;il peut produire de la matière fissile.&lt;/p&gt;&#10;&lt;p&gt;On peut ainsi expliquer la composition du combustible usé en sortie de réacteur : au total 2% du total d&amp;rsquo;uranium 238 a été transmuté en plutonium. 1% de ce plutonium a fissionné et ajoute 1% aux produits de fission, et 1% reste intact dans le combustible usé. Les 0.1% d&amp;rsquo;actinides mineurs correspondent à des atomes lourds produits par de multiples captures de neutrons, on reviendra sur eux plus tard.&lt;/p&gt;&#10;&lt;h3 id="applications-militaires-du-plutonium"&gt;Applications militaires du plutonium&lt;/h3&gt;&#10;&lt;p&gt;Dès les débuts du projet Manhattan, les américains ont compris l&amp;rsquo;intérêt militaire du plutonium. L&amp;rsquo;uranium est très difficile à enrichir pour atteindre de hauts pourcentages (+ de 95%) d&amp;rsquo;uranium 235 fissile. Les deux isotopes de l&amp;rsquo;uranium sont quasi indistinguables (ils réagissent pareil, fondent pareil, etc.) et seule leur petite différente de masse peut permettre de les séparer. C&amp;rsquo;est cher, lent et compliqué.&lt;/p&gt;&#10;&lt;p&gt;Le plutonium est produit en petite quantité dans les réacteurs, mais il est facilement séparable du reste par des réactions chimiques. Ainsi, si l&amp;rsquo;on souhaite produire une bombe, et pour peu que l&amp;rsquo;on possède un réacteur adapté, il peut être beaucoup plus simple de le faire à partir de plutonium extrait de combustibles préalablement irradiés qu&amp;rsquo;à partir d&amp;rsquo;uranium coûteusement enrichi.&lt;/p&gt;&#10;&lt;p&gt;C&amp;rsquo;est ce qui s&amp;rsquo;est produit historiquement. Les américains, après un énorme effort de guerre, ont péniblement réussi à enrichir 64 kg d&amp;rsquo;uranium pour produire la bombe Little Boy qui explosa sur Hiroshima.&lt;/p&gt;&#10;&lt;p&gt;Mais c&amp;rsquo;est leur programme de production de plutonium qui permit de faire la première bombe atomique de l&amp;rsquo;essai Trinity, puis Fat man qui explosa sur Nagazaki et ensuite le gros de leur arsenal.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/trinity-e1579016571135.jpg"&#10;&#9;&#9;&#9;alt="trinity.jpg" width="5554"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Bombe au plutonium (6kg) de l&amp;rsquo;essai Trinity&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;En France, la première série de réacteurs du CEA (G1 G2 G3), uranium naturel graphite gaz (UNGG) avait aussi pour objectif la production de plutonium pour la force de dissuasion. Plus généralement, posséder un réacteur donne accès au plutonium comme sous produit, qui donne potentiellement accès à une production industrielle d&amp;rsquo;armes nucléaires.&lt;/p&gt;&#10;&lt;p&gt;C&amp;rsquo;est le risque de prolifération lié aux activités nucléaires même a priori civiles qui explique que l&amp;rsquo;on soit modérément serein quand un pays instable démarre un réacteur nucléaire, sauf à pouvoir contrôler précisément ce qui entre et sort de ses réacteurs. On va voir plus tard que ce point est à nuancer pour les réacteurs actuels, qui ne permettent pas facilement un détournement militaire, à cause de la faible qualité du plutonium qui en sort.&lt;/p&gt;&#10;&lt;h3 id="le-retraitement-des-combustibles-nucléaires-usés-et-recyclage"&gt;Le retraitement des combustibles nucléaires usés et recyclage&lt;/h3&gt;&#10;&lt;p&gt;Après cette digression, revenons sur le combustible usé en sortie de réacteur&lt;/p&gt;&#10;&lt;p&gt;Le combustible usé contient 4% de produits de fission et actinides mineurs. Ils sont très radioactifs et pas simplement fissiles, bref non seulement inutiles mais aussi pénibles à gérer. Le combustible usé contient aussi 1% de plutonium fissile, et 95% d&amp;rsquo;uranium lui même à 1% environ d&amp;rsquo;uranium 235 fissile.&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/acu.png" alt="acu"&gt;&lt;/p&gt;&#10;&lt;p&gt;L&amp;rsquo;idée du retraitement est de séparer ce qui est potentiellement réutilisable de ce qui ne l&amp;rsquo;est pas. Les produits de fission et les actinides mineurs sont dangereux et inutiles, ils sont donc séparés et vitrifiés (c&amp;rsquo;est à dire mélangés à un verre spécial et coulés dans des fûts en inox). Le plutonium et l&amp;rsquo;uranium sont extraits.&lt;/p&gt;&#10;&lt;p&gt;Le plutonium est mélangé à de l&amp;rsquo;uranium appauvri en substitut de l&amp;rsquo;uranium 235 pour produire un nouveau combustible, le MOX (pour mixed oxydes). Ce dernier contient aujourd&amp;rsquo;hui environ 9% de plutonium (on verra pourquoi c&amp;rsquo;est beaucoup plus que les 4% d&amp;rsquo;uranium 235 du combustible standard plus tard). On produit environ 1 MOX pour 10 combustibles UNE retraités.&lt;/p&gt;&#10;&lt;p&gt;L&amp;rsquo;uranium extrait est appelé uranium de retraitement (URT). Aujourd&amp;rsquo;hui il est simplement entreposé en attente de réutilisation. Mais jusqu&amp;rsquo;en 2013 (et probablement de nouveau à partir de 2023) il était réenrichi pour donner de l&amp;rsquo;uranium de retraitement réenrichi (URE) à grosso modo 6% de 235 contre 1% pour l&amp;rsquo;URT avant réenrichissement (on verra aussi plus tard pourquoi c&amp;rsquo;est plus que 4%). On produisait (et produira) environ 1.5 combustible URE pour 10 UNE retraités.&lt;/p&gt;&#10;&lt;p&gt;Les réacteurs doivent être adaptés pour utiliser ces combustibles dont la réactivité est un peu différente de celle des UNE, mais la technologie reste celle des réacteurs actuels. En France ce sont les réacteurs 900MW, donc les plus anciens réacteurs à eau (dont Fessenheim) qui ont été adaptés pour brûler du MOX (&amp;ldquo;moxés&amp;rdquo; pour faire fancy). L&amp;rsquo;URE était quant à lui brûlé dans les réacteurs de la centrale de Cruas (900MW aussi) jusqu&amp;rsquo;en 2013 (mais je ne sais pas où EDF prévoit de le faire ensuite).&lt;/p&gt;&#10;&lt;p&gt;En sortie de réacteur, les MOX et URE usés contiennent toujours de l&amp;rsquo;uranium et du plutonium qu&amp;rsquo;on pourrait être tenté de réutiliser.&lt;/p&gt;&#10;&lt;p&gt;Mais aujourd&amp;rsquo;hui ça n&amp;rsquo;est pas le cas et la France se contente de &amp;ldquo;monorecycler&amp;rdquo; ses combustibles usés. MOX et URE usés sont donc entreposés en l&amp;rsquo;état dans les piscines de la Hague. En résumé on a donc quelque chose comme ça :&lt;/p&gt;&#10;&lt;p&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/summary.png" alt="summary"&gt;&lt;/p&gt;&#10;&lt;h2 id="le-multirecyclage"&gt;Le multirecyclage&lt;/h2&gt;&#10;&lt;h3 id="tous-les-plutonium-ne-se-valent-pas"&gt;Tous les plutonium ne se valent pas&lt;/h3&gt;&#10;&lt;p&gt;Le MOX usé contient toujours environ 5% de plutonium, bien plus que les UNE usés, et une ressource a priori énorme. Pourquoi ne s&amp;rsquo;en sert-on pas ? Plus suspect encore le MOX était chargé à 9% de plutonium alors qu&amp;rsquo;on a dit que le plutonium 239 avait grosso modo les mêmes propriétés que l&amp;rsquo;uranium 235 (et on n&amp;rsquo;a besoin que de 4% de ce dernier). Pourquoi ça ne nous pète pas à la figure ?&lt;/p&gt;&#10;&lt;p&gt;En fait on a trop simplifié la situation. Les captures de neutrons ne s&amp;rsquo;arrêtent pas au plutonium 239. Lui même peut capturer un neutron sans fissionner pour donner du plutonium 240. Ce dernier peut encore capturer un neutron pour produire du plutonium 241, qui se désintègre ou, s&amp;rsquo;il capture encore un neutron, donner du plutonium 242. Les isotopes impairs du plutonium sont fissiles, mais les isotopes pairs ne le sont pas dans les réacteurs actuels. La proportion d&amp;rsquo;isotopes impairs détermine la &amp;ldquo;qualité&amp;rdquo; du plutonium. Tous les plutoniums ne se valent pas.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/formationplutonium.jpg"&#10;&#9;&#9;&#9;alt="FormationPlutonium" width="464"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Différents isotopes du plutonium (&lt;a href="http://www.laradioactivite.com/site/pages/lesisotopesduplutonium.htm"&gt;IN2P3&lt;/a&gt;)&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Le plutonium du combustible UNE usé ne contient que 60% de plutonium 239 fissile, et environ 30% de 240 qui est au contraire un poison neutronique (il capture beaucoup les neutrons et réduit la réactivité). C&amp;rsquo;est pour ça qu&amp;rsquo;il fallait monter à 9% pour faire un MOX réactif. Maintenant en sortie le MOX usé contient certes 5% de plutonium, mais ce dernier a encore moins de 239 (moins de 40%) car les captures se sont multipliées pendant que les isotopes fissiles étaient consommés (augmentant mécaniquement la fraction d&amp;rsquo;isotopes pairs non fissiles).&lt;/p&gt;&#10;&lt;p&gt;Maintenant, pourquoi ne pas faire du &amp;ldquo;super MOX&amp;rdquo; à 15% de plutonium pour compenser la dégradation de sa qualité ? Pour comprendre le problème, il faut remonter au tout début et la question de la modération. On l&amp;rsquo;a vu la présence d&amp;rsquo;eau comme modérateur (et caloporteur) dans les réacteurs actuels favorise la réaction en ralentissant les neutrons, et en facilitant les fissions d&amp;rsquo;uranium 235 (ou plutonium impair) par rapport aux captures. Les isotopes pairs du plutonium, on l&amp;rsquo;a dit, ne sont pas fissiles pour des neutrons modérés, mais ils le deviennent pour des neutrons rapides, non ralentis.&lt;/p&gt;&#10;&lt;p&gt;Ainsi, si on réduit la modération, on réduit la réactivité en augmentant les captures, mais on l&amp;rsquo;augmente en rendant fissile les isotopes pairs du plutonium. Tant qu&amp;rsquo;il n&amp;rsquo;y a pas trop de plutonium, le premier effet domine et la perte d&amp;rsquo;eau réduit la réactivité. Mais passé environ 12%, l&amp;rsquo;augmentation de réactivité des isotopes pairs du plutonium domine, l&amp;rsquo;effet d&amp;rsquo;une réduction de modération est une augmentation de la réactivité. Un réacteur chargé avec un tel combustible serait donc instable en cas de perte de l&amp;rsquo;eau dans le circuit primaire, ce qui est inacceptable.&lt;/p&gt;&#10;&lt;p&gt;Pourquoi ne pas enrichir le plutonium comme on enrichit l&amp;rsquo;uranium, par centrifugation ? L&amp;rsquo;uranium est peu radioactif, c&amp;rsquo;est un émetteur beta de longue période. On peut le passer dans des centrifugeuses et plus généralement des usines sans protection massive. À l&amp;rsquo;inverse le plutonium possède une durée de vie plus courte et est un puissant émetteur alpha, rayonnement fortement énergétique. Il ainsi beaucoup plus compliqué à manipuler ce qui rend son enrichissement difficile.&lt;/p&gt;&#10;&lt;p&gt;On est ainsi limité à environ 9% pour la concentration en plutonium dans du combustible. Le plutonium du MOX usé est ainsi de qualité trop faible pour fabriquer un combustible capable d&amp;rsquo;alimenter une réaction en chaîne car il faudrait monter à des pourcentages supérieurs. On va voir qu&amp;rsquo;il y a malgré tout des solutions.&lt;/p&gt;&#10;&lt;h3 id="retour-sur-lutilisation-militaire-du-plutonium-et-risque-de-prolifération"&gt;Retour sur l&amp;rsquo;utilisation militaire du plutonium et risque de prolifération&lt;/h3&gt;&#10;&lt;p&gt;Comme on a parlé de la question de la qualité isotopique du plutonium on peut revenir sur les applications militaires et le risque de prolifération. Aujourd&amp;rsquo;hui, on ne sait pas faire de bombe avec du plutonium contenant moins de 90% de plutonium 239.&lt;/p&gt;&#10;&lt;p&gt;Comment l&amp;rsquo;obtenir ?&lt;/p&gt;&#10;&lt;p&gt;Pour obtenir du plutonium ultra pur en 239 il faut éviter les captures additionnelles, et donc irradier un très court moment, typiquement seulement quelques jours au lieu de 3 ans. Pour ça il faut avoir un réacteur qui permet de faire des irradiations très courtes d&amp;rsquo;assemblages de combustible sans tout perturber.&lt;/p&gt;&#10;&lt;p&gt;Les réacteurs au graphite le permettaient très facilement, les réacteurs à eau pressurisée actuels rendent ça très compliqué car il faut arrêter le réacteur à chaque déplacement d&amp;rsquo;un assemblage. De surcroît, ces arrêts sont visibles dans la production électrique : un pays qui tenterait malgré tout de produire du plutonium militaire avec un réacteur à eau civil n&amp;rsquo;est pas trop difficile à démasquer.&lt;/p&gt;&#10;&lt;p&gt;Le fait que les réacteurs à eau produisent un plutonium très dégradé en conditions normales et qu&amp;rsquo;il soit difficile de faire des irradiations courtes avec est ainsi un gros avantage du point de vue de la prolifération.&lt;/p&gt;&#10;&lt;h3 id="multirecyclage-du-plutonium-dans-les-réacteurs-actuels"&gt;Multirecyclage du plutonium dans les réacteurs actuels&lt;/h3&gt;&#10;&lt;p&gt;Donc on résume, le plutonium qui sort d&amp;rsquo;un réacteur à eau possède un mélange d&amp;rsquo;isotopes dont seule une fraction est fissile.&lt;/p&gt;&#10;&lt;p&gt;Après un second passage en réacteur, ce plutonium contient tellement d&amp;rsquo;isotopes non fissiles qu&amp;rsquo;on ne peut plus en faire du combustible sans dépasser un taux limite qui rendrait le réacteur instable. Peut-on passer outre ?&lt;/p&gt;&#10;&lt;p&gt;Oui, a priori on peut. Pour ça on peut imaginer ajouter de l&amp;rsquo;uranium 235 issu de l&amp;rsquo;enrichissement pour donner un coup de boost au plutonium dégradé. Là où le plutonium 239 permettait de remplacer l&amp;rsquo;uranium 235, on peut remplacer une partie du plutonium de mauvaise qualité par de l&amp;rsquo;uranium 235 pour augmenter sa réactivité.&lt;/p&gt;&#10;&lt;p&gt;Naturellement, l&amp;rsquo;intérêt est moindre car il faut ajouter de l&amp;rsquo;uranium enrichi pour brûler le plutonium. Comme cela conduit à mettre des isotopes plus lourds du plutonium plus longtemps en réacteur, ça conduit aussi à produire plus d&amp;rsquo;actinides mineurs.&lt;/p&gt;&#10;&lt;p&gt;Le plutonium se dégrade aussi de plus en plus pour arriver à une valeur d&amp;rsquo;équilibre autour de 30% en plutonium 239, on a donc asymptotiquement un plutonium de très basse qualité (même si la qualité ne tend pas vers 0).&lt;/p&gt;&#10;&lt;p&gt;Ce multirecyclage du plutonium sur base uranium enrichi, aujourd&amp;rsquo;hui, n&amp;rsquo;est pas fait, mais c&amp;rsquo;est dans les tiroirs en France pour l&amp;rsquo;après 2030. Je me garderai bien de dire si c&amp;rsquo;est positif ou pas, il y a des avantages et des défauts et le bilan n&amp;rsquo;est pas évident.&lt;/p&gt;&#10;&lt;h3 id="pollution-isotopique-de-luranium"&gt;Pollution isotopique de l&amp;rsquo;uranium&lt;/h3&gt;&#10;&lt;p&gt;Maintenant on pourrait imaginer réutiliser l&amp;rsquo;uranium contenu dans les URE usés. La difficulté est similaire à celle que l&amp;rsquo;on rencontre pour le plutonium. Souvenons nous que pour le combustible à l&amp;rsquo;uranium réenrichi, on montait à 6% de 235 au lieu de 4% dans des combustibles non-recyclés. La raison, c&amp;rsquo;est que l&amp;rsquo;uranium qui est déjà passé en réacteur contient un peu d&amp;rsquo;uranium 236, qui vient de l&amp;rsquo;uranium 235 qui n&amp;rsquo;a pas fissionné et a capturé un neutron.&lt;/p&gt;&#10;&lt;p&gt;Cet uranium 236 est un puissant poison neutronique, c&amp;rsquo;est à dire qu&amp;rsquo;il capte des neutrons sans rien faire, et donc réduit fortement la réactivité.&lt;/p&gt;&#10;&lt;p&gt;Après un deuxième passage en réacteur, l&amp;rsquo;uranium de l&amp;rsquo;URE usé contient beaucoup d&amp;rsquo;uranium 236 et il est donc plus difficile d&amp;rsquo;en faire un combustible recyclé. Ici l&amp;rsquo;obstacle est je crois davantage économique que technique : comme il y a peu d&amp;rsquo;URE usé, qu&amp;rsquo;il faudrait réenrichir beaucoup pour produire peu de nouveau combustibles, ça ne vaut pas le coup.&lt;/p&gt;&#10;&lt;h3 id="multirecyclage-croisé-"&gt;Multirecyclage croisé ??&lt;/h3&gt;&#10;&lt;p&gt;Maintenant on pourrait être encore plus vicieux et réutiliser l&amp;rsquo;uranium du MOX usé et le plutonium de l&amp;rsquo;URE usé.&lt;/p&gt;&#10;&lt;p&gt;La réutilisation de l&amp;rsquo;uranium du MOX usé est impossible sans multirecyclage de son plutonium pour des raisons réglementaires. La France a signé des traités de non-prolifération qui lui interdisent d&amp;rsquo;accumuler trop de plutonium séparé, c&amp;rsquo;est à dire hors d&amp;rsquo;un assemblage de combustible prêt à être utilisé. A priori ce plutonium de faible qualité n&amp;rsquo;est pas proliférant, mais j&amp;rsquo;imagine que les traités ne mentionnent pas l&amp;rsquo;isotopie.&lt;/p&gt;&#10;&lt;p&gt;Pour le recyclage du plutonium de l&amp;rsquo;URE usé a priori pas de problème de physique évident, mais probablement un obstacle économique qui est qu&amp;rsquo;aujourd&amp;rsquo;hui il y a très peu d&amp;rsquo;URE usé. Peut-être que cela changera quand le réenrichissement de l&amp;rsquo;uranium de retraitement sera repris.&lt;/p&gt;&#10;&lt;h2 id="multirecyclage-en-réacteurs-à-neutrons-rapides"&gt;Multirecyclage en réacteurs à neutrons rapides&lt;/h2&gt;&#10;&lt;p&gt;Jusqu&amp;rsquo;à maintenant on parle de recyclage à petite échelle et il faut maintenant mentionner une option qui changerait complètement les ordres de grandeur. Pour ça il faut revenir au concept de modération.&lt;/p&gt;&#10;&lt;p&gt;On l&amp;rsquo;a vu la modération, en augmentant la réactivité, permet de faire fonctionner des réacteurs avec des concentrations &amp;ldquo;faibles&amp;rdquo; de matière fissile. La contrepartie, c&amp;rsquo;est que les neutrons modérés font fissionner uniquement les isotopes impairs du plutonium et que ce dernier accumule ainsi les isotopes pairs.&lt;/p&gt;&#10;&lt;p&gt;L&amp;rsquo;idée des réacteurs à neutrons rapides est simplement de ne pas modérer en utilisant un combustible très concentré en matières fissiles, c&amp;rsquo;est à dire à l&amp;rsquo;uranium très enrichi, ou fortement concentré en plutonium.&lt;/p&gt;&#10;&lt;p&gt;Dans un tel réacteur il se passe plusieurs choses intéressantes. D&amp;rsquo;abord, l&amp;rsquo;intégralité des isotopes du plutonium sont fissiles, ce qui permet de stabiliser sa qualité à une valeur plus élevée que le multirecyclage en réacteurs modérés (à eau légère).&lt;/p&gt;&#10;&lt;p&gt;Mais ça n&amp;rsquo;est pas le plus intéressant&amp;hellip;&lt;/p&gt;&#10;&lt;h3 id="régénération-et-surgénération"&gt;Régénération et surgénération&lt;/h3&gt;&#10;&lt;p&gt;On l&amp;rsquo;a vu, un réacteur à eau légère produit du plutonium mais ce dernier ne compense pas la quantité d&amp;rsquo;uranium consommée. Il y a trop de pertes de neutrons. Maintenant, la fission d&amp;rsquo;un noyau de plutonium 239 produit environ 3 neutrons en moyenne. Ainsi, pour peu que l&amp;rsquo;on ne perde pas trop de neutrons à droit à gauche, on doit pouvoir en utiliser 1 pour faire de nouvelles fissions et 1 (voire plus) pour régénérer la matière fissile à partir de l&amp;rsquo;uranium 238.&lt;/p&gt;&#10;&lt;p&gt;C&amp;rsquo;est précisément ce qu&amp;rsquo;on peut faire dans un réacteur à neutrons rapide. En imaginant que ce dernier est démarré avec un mélange uranium 238 / plutonium 239, on peut faire fissionner tout le 239 tout en le régénérant intégralement par transmutation de l&amp;rsquo;uranium 238. En fait on peut même en théorie surgénérer, et produire plus de plutonium qu&amp;rsquo;on n&amp;rsquo;en consomme pour alimenter d&amp;rsquo;autres réacteurs plus standards. Ainsi, au lieu de consommer environ 0.5% de l&amp;rsquo;uranium naturel, on pourrait en théorie monter à 100%, c&amp;rsquo;est à dire gagner un facteur 200 dans l&amp;rsquo;utilisation de la ressource naturelle.&lt;/p&gt;&#10;&lt;p&gt;Ça c&amp;rsquo;est en théorie. En pratique, les réacteurs à neutrons rapides sont aujourd&amp;rsquo;hui un peu plus compliqués à faire. Il y a des questions de sûreté et de prolifération potentiellement moins simples que pour les réacteurs à eau et donc il y en a beaucoup moins en service.&lt;/p&gt;&#10;&lt;p&gt;En France il y avait Superphénix, arrêté dans les années 1990 pour des raisons à la fois techniques et politiques, Phénix (plus petit, démarré avant et arrêté après) et il devait y avoir Astrid, projet dont la réalisation pratique a récemment été repoussée à l&amp;rsquo;après 2050.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2020/01/superphenix_reactor_south.jpg"&#10;&#9;&#9;&#9;alt="Superphenix_reactor_south" width="407"&gt;&lt;figcaption&gt;&#10;&#9;&#9;&#9;&lt;p&gt;Superphénix (© &lt;a href="https://commons.wikimedia.org/wiki/User:Yann" title="User:Yann"&gt;Yann Forget&lt;/a&gt; / &lt;a href="https://commons.wikimedia.org/" class="external text"&gt;Wikimedia Commons&lt;/a&gt; / &lt;span class="plainlinks noprint"&gt;&lt;a href="https://creativecommons.org/licenses/by-sa/3.0/" class="external text" rel="nofollow"&gt;CC-BY-SA&lt;/a&gt;)&lt;/span&gt;&lt;/p&gt;&#10;&#9;&#9;&lt;/figcaption&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Les russes ont divers prototypes de taille quasi industrielle comme le BN800 à Beloyarsk. Les réacteurs à neutrons rapides sont donc faisables, ça n&amp;rsquo;est pas de la science fiction ni quelque chose d&amp;rsquo;incertain comme la fusion nucléaire, mais le fait est qu&amp;rsquo;il y en a peu.&lt;/p&gt;&#10;&lt;h2 id="taux-de-recyclage-et-économie-de-ressources"&gt;Taux de recyclage et économie de ressources&lt;/h2&gt;&#10;&lt;p&gt;Maintenant qu&amp;rsquo;on a dit tout ça, on peut se demander combien ces diverses réutilisations de matière permettent comme économies ou recyclage suivant ce qu&amp;rsquo;on choisit comme stratégie.&lt;/p&gt;&#10;&lt;p&gt;Commençons par le chiffre généralement mis en avant par les partisans du recyclage:&lt;/p&gt;&#10;&lt;h3 id="96-du-combustible-est-recyclable---trompeur"&gt;96% du combustible est recyclable - trompeur&lt;/h3&gt;&#10;&lt;p&gt;&lt;em&gt;Techniquement c&amp;rsquo;est vrai&lt;/em&gt;, effectivement on sort du combustible usé l&amp;rsquo;uranium de retraitement et le plutonium. Restent 4% qu&amp;rsquo;on vitrifie. Donc on a bien 96% de recyclable. Mais c&amp;rsquo;est franchement trompeur : déjà, ces 96% ne sont pas uniformes et énergétiquement et il y a autant dans les 95% d&amp;rsquo;uranium faiblement enrichi que dans le 1% de plutonium. Il ne reste plus grand chose à essorer de ce dernier, et une fois essoré on ne sait pas bien quoi faire du reliquat appauvri (à moins de le fertiliser en réacteur à neutrons rapides&amp;hellip;).&lt;/p&gt;&#10;&lt;p&gt;De plus, si on passait à une régénération intégrale du plutonium en réacteur à neutrons rapide, c&amp;rsquo;est à dire si on gagnait un facteur 200 d&amp;rsquo;efficacité énergétique, ce chiffre de 96% ne changerait pas, il pourrait même devenir moins bon ! En effet, si on régénère le plutonium à un meilleur taux et en imaginant que les gaines tiennent le coup, on pourrait faire tourner le réacteur plus longtemps, et augmenter mécaniquement le pourcentage de produits de fission et d&amp;rsquo;actinides mineurs en sortie. On aurait donc un taux de recyclage inférieur alors que la situation est clairement meilleure.&lt;/p&gt;&#10;&lt;p&gt;À l&amp;rsquo;inverse, imaginons qu&amp;rsquo;on réduise artificiellement le temps passé en réacteur des combustibles (le burnup) et qu&amp;rsquo;on ne produise ainsi qu'1% de produits de fission. Alors 99% du combustible serait recyclable.&lt;/p&gt;&#10;&lt;p&gt;Bref, ce chiffre n&amp;rsquo;est pas incorrect mais il ne permet pas de comparaison raisonnable, et induit au moins superficiellement une opinion fausse. C&amp;rsquo;est une mauvaise métrique.&lt;/p&gt;&#10;&lt;h3 id="1-du-combustible-est-effectivement-recyclé---trompeur-aussi"&gt;1% du combustible est effectivement recyclé - trompeur aussi&lt;/h3&gt;&#10;&lt;p&gt;Ce chiffre est en général donné par les opposants au recyclage dans l&amp;rsquo;hypothèse que l&amp;rsquo;on ne fait que du MOx et donc que seul le plutonium est réutilisé et une unique fois (mais c&amp;rsquo;est légitime, c&amp;rsquo;est effectivement ce qu&amp;rsquo;on fait aujourd&amp;rsquo;hui).&lt;/p&gt;&#10;&lt;p&gt;Savoir ce qu&amp;rsquo; &amp;ldquo;&lt;em&gt;effectivement recyclé&lt;/em&gt;&amp;rdquo; veut dire n&amp;rsquo;est pas tout à fait clair, mais on comprend bien l&amp;rsquo;idée : on ne réutilise que le plutonium, et il représente 1% du combustible en masse.&lt;/p&gt;&#10;&lt;p&gt;Ça n&amp;rsquo;est pas totalement faux mais c&amp;rsquo;est aussi franchement trompeur : ce plutonium possède une énorme densité énergétique. De plus, quand EDF réutilisera de nouveau l&amp;rsquo;uranium de retraitement, est-ce qu&amp;rsquo;on passera d&amp;rsquo;un coup à 96% avec ce calcul ? Ou est-ce qu&amp;rsquo;on ne compte que l&amp;rsquo;uranium 235 contenu dans l&amp;rsquo;uranium de retraitement ? Mais alors comment prendre en compte la portion de plutonium régénérée en réacteur ? Le problème d&amp;rsquo;un réacteur nucléaire c&amp;rsquo;et qu&amp;rsquo;il brûle typiquement plus de matière fissile que ce qu&amp;rsquo;on met dedans en entrée&amp;hellip;&lt;/p&gt;&#10;&lt;p&gt;Le pourcentage effectivement recyclé est trop ambigu pour être défini de manière univoque. C&amp;rsquo;est aussi une mauvaise métrique.&lt;/p&gt;&#10;&lt;h3 id="quest-ce-quune-bonne-mesure-"&gt;Qu&amp;rsquo;est ce qu&amp;rsquo;une bonne mesure ?&lt;/h3&gt;&#10;&lt;p&gt;En physique, on réfléchit souvent à la question de &amp;ldquo;la bonne mesure&amp;rdquo;. Et il y a quelques principes simples qu&amp;rsquo;on peut utiliser pour trouver une bonne mesure.&lt;/p&gt;&#10;&lt;p&gt;Le principal, c&amp;rsquo;est qu&amp;rsquo;il faut que la valeur ne change pas par un simple &lt;em&gt;relabelling&lt;/em&gt; ou redéfinition des objets.&lt;/p&gt;&#10;&lt;p&gt;Par exemple imaginons que je rajoute un bloc de bois qui ne sert à rien à côté du combustible et que je déclare qu&amp;rsquo;il fait aussi partie du combustible. Ou que je me mette à compter les gaines dans le combustible, ou les murs de la centrale. Ça ne doit rien changer en termes de recyclage, c&amp;rsquo;est la même situation. Or pour les deux valeurs précédemment mentionnées, ça change tout !&lt;/p&gt;&#10;&lt;p&gt;Après avoir pas mal réfléchi à la question, je pense (je ne dis pas que c&amp;rsquo;est original) que la seule manière raisonnable d&amp;rsquo;estimer ce qui est recyclé est de regarder combien on économise de ressources naturelles à service identique (ici la production d&amp;rsquo;électrique).&lt;/p&gt;&#10;&lt;p&gt;On imagine un cycle ouvert, c&amp;rsquo;est à dire une configuration où le combustible usé n&amp;rsquo;est pas réutilisé. Et on le compare aux alternatives et on regarde l&amp;rsquo;économie de ressources obtenue, c&amp;rsquo;est à dire combien on a dû miner d&amp;rsquo;uranium dans les deux configurations. On obtient alors :&lt;/p&gt;&#10;&lt;ol&gt;&#10;&lt;li&gt;Réutilisation du plutonium 1 fois avec du MOX (monorecyclage) &lt;strong&gt;10% d&amp;rsquo;économie&lt;/strong&gt; [pour 10 UNE, on produit 1 MOX]&lt;/li&gt;&#10;&lt;li&gt;Réutilisation du plutonium avec le MOx et de l&amp;rsquo;uranium avec l&amp;rsquo;URE &lt;strong&gt;25% d&amp;rsquo;économie&lt;/strong&gt; [pour 10 UNE, on produit 1 MOX et 1.5 URE]&lt;/li&gt;&#10;&lt;li&gt;Monorecyclage de l&amp;rsquo;URE + multirecyclage du plutonium en réacteurs à eau légère &lt;strong&gt;30% d&amp;rsquo;économie&lt;/strong&gt; [là la comptabilité est plus compliquée&amp;hellip;]&lt;/li&gt;&#10;&lt;li&gt;Multirecyclage en réacteur à neutrons rapide en théorie jusqu&amp;rsquo;à &lt;strong&gt;99.5% d&amp;rsquo;économie&lt;/strong&gt;&lt;/li&gt;&#10;&lt;/ol&gt;&#10;&lt;p&gt;Je pense que ce sont les chiffres pertinents à avoir en tête et probablement les seuls cohérents. Quitte à passer pour un scientiste, je prends le risque de dire que ce n&amp;rsquo;est pas une question de point de vue. Les autres chiffres sont non seulement trompeurs mais incohérents (soit à cause de leur périmètre ambigu, soit à cause de la régénération des matières fissiles).&lt;/p&gt;&#10;&lt;p&gt;On voit qu&amp;rsquo;en poussant comme un fou les réacteurs actuels on monte à 30% et il paraît difficile, même en principe, de dépasser 40%. En revanche, si on s&amp;rsquo;autorise des réacteurs à neutrons rapides, alors on change complètement de ligue.&lt;/p&gt;&#10;&lt;h3 id="quel-type-de-recyclage-choisir-"&gt;Quel type de recyclage choisir ?&lt;/h3&gt;&#10;&lt;p&gt;Est-ce qu&amp;rsquo;il est toujours mieux de recycler plus ? C&amp;rsquo;est une question bien plus large que ce qu&amp;rsquo;on aborde ici, et je voudrais être clair sur le fait que je n&amp;rsquo;ai pas cherché à y répondre. Ce recyclage a plein d&amp;rsquo;impacts secondaires : quand on recycle, on fait davantage d&amp;rsquo;opérations sur des matières radioactives (négatif), ce qui reste est au total moins radiotoxique (positif), le volume de déchets finaux est réduit (positif, même si l&amp;rsquo;ampleur est discutée), le conditionnement des déchets est meilleur (positif), on dispose d&amp;rsquo;une technologie de séparation du plutonium qui pourrait être utilisée à d&amp;rsquo;autres fins (négatif). Et il y a encore plein d&amp;rsquo;autres critères&amp;hellip;&lt;/p&gt;&#10;&lt;p&gt;Par ailleurs, tant que l&amp;rsquo;on monorecycle ou multirecycle dans les réacteurs actuels, c&amp;rsquo;est à dire qu&amp;rsquo;on est autour de 10 - 30% d&amp;rsquo;économie, alors l&amp;rsquo;économie d&amp;rsquo;énergie n&amp;rsquo;est pas de manière évidente le critère dominant permettant de choisir parmi les précédentes nuances. C&amp;rsquo;est le saut vers le multirecyclage en réacteurs à neutrons rapides qui change réellement la donne.&lt;/p&gt;&#10;&lt;p&gt;Ce qui est sûr c&amp;rsquo;est qu&amp;rsquo;à très long terme, si le nucléaire se maintient et se développe, notamment via la Chine et que l&amp;rsquo;on reste sur un cycle uranium/plutonium (pas de thorium) alors la limitation des ressources naturelles forcera le multirecyclage en réacteurs à neutrons rapides. En attendant, la ressource naturelle est peu chère dans le coût total de l&amp;rsquo;énergie nucléaire, notamment parce que des progrès importants ont été réalisés sur l&amp;rsquo;extraction et l&amp;rsquo;enrichissement de l&amp;rsquo;uranium, et donc il n&amp;rsquo;y a pas une très forte pression ce qui explique (je me garde de dire si ça justifie) le fait que la puissance publique ait repoussé récemment l&amp;rsquo;horizon de développement du multirecyclage en réacteurs à neutrons rapides.&lt;/p&gt;&#10;&lt;h3 id="retour-sur-le-contexte-et-le-débat-public"&gt;Retour sur le contexte et le débat public&lt;/h3&gt;&#10;&lt;p&gt;La question de savoir s&amp;rsquo;il fallait recycler ou non le combustible usé (et le point incroyablement plus précis et restreint de savoir quelle économie de ressources le recyclage permet) était posée dans le débat public de 2019 sur la gestion des matières et déchets radioactifs. J&amp;rsquo;étais membre de la commission qui organisait ce débat. On avait organisé un exercice de &lt;a href="https://pngmdr.debatpublic.fr/approfondir/clarification-des-controverses-techniques"&gt;clarification des controverses techniques&lt;/a&gt; avec des experts de tous horizons pour se mettre d&amp;rsquo;accord sur les conséquences du recyclage des combustibles usés (à défaut de trancher sur sa pertinence).&lt;/p&gt;&#10;&lt;p&gt;La question du taux de recyclage avait fait débat, mais on avait convenu dans la synthèse, non sans mal, que l&amp;rsquo;économie de ressources naturelles était la manière pertinente de mesurer la chose. Ensuite, durant le débat lui même, on avait pu voir que le débat était très actif sur le choix d&amp;rsquo;une stratégie plutôt qu&amp;rsquo;une autre, avec une situation qui peut sembler à front renversé pour le béotien. Les &amp;ldquo;pro-nucléaire&amp;rdquo; sont quasi uniformément favorables à la stratégie la plus ambitieuse possible de recyclage, les &amp;ldquo;anti-nucléaire&amp;rdquo;, dont un certain nombre d&amp;rsquo;associations environnementales, souhaitant à l&amp;rsquo;inverse son arrêt immédiat. C&amp;rsquo;est que derrière tous les aspects positifs ou négatifs du recyclage, il y a surtout la question de la poursuite du nucléaire, le recyclage ancrant son utilisation dans la durée. Mais c&amp;rsquo;est une autre histoire&amp;hellip;&lt;/p&gt;&#10;&lt;p&gt;&lt;em&gt;Il est difficile d&amp;rsquo;écrire un blogpost pareil sans faire d&amp;rsquo;erreurs, n&amp;rsquo;hésitez pas à les corriger dans les commentaires.&lt;/em&gt;&lt;/p&gt;&#10;</description></item><item><title>Nice quantum field theory videos</title><link>https://atilloy.com/2019/10/29/nice-quantum-field-theory-videos/</link><guid>https://atilloy.com/2019/10/29/nice-quantum-field-theory-videos/</guid><pubDate>Tue, 29 Oct 2019 16:43:09 +0100</pubDate><description>&lt;p&gt;These days I am trying to improve my understanding of quantum field theory with as little perturbation theory as possible. I came across videos from a &lt;a href="https://www.youtube.com/playlist?list=PLx5f8IelFRgHkCsp-XyPEUn8caIUENB4j"&gt;workshop at IHES&lt;/a&gt; in Bures-sur-Yvette on Hamiltonian methods for QFT and videos from &lt;a href="https://www.newton.ac.uk/event/srq/seminars"&gt;a semester at the Newton institute&lt;/a&gt; in Cambridge, which both happened about a year ago. Both events are quite well filmed (especially at IHES), most presentations are made on a blackboard, and most talks I checked were well explained and interesting so I definitely recommend them.&lt;/p&gt;&#10;&lt;p&gt;The workshop at IHES could have been called &lt;em&gt;50 shades of&lt;/em&gt; &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_2&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;2&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;, since many talks try to find the critical point of the theory with more or less elaborate methods (8 loop perturbation theory, and various non-perturbative Hamiltonian methods). I recommend in particular &lt;a href="https://www.youtube.com/watch?v=4KB5hd3N1rU&amp;amp;list=PLx5f8IelFRgHkCsp-XyPEUn8caIUENB4j&amp;amp;index=8&amp;amp;t=0s"&gt;the talk of Joan Elias Miro on renormalized Hamiltonian truncation methods&lt;/a&gt;, which I found very clear and interesting. There are also nice tensor network talks by the usual suspects (Mari Carmen Banuls, Frank Pollman, Guifre Vidal, Karen Van Acoleyen, Philippe Corboz). Finally there is an intriguing talk by Giuseppe Mussardo on the sinh-Gordon model.&lt;/p&gt;&#10;&lt;p&gt;The semester at the Newton Institute was clearly geared more towards mathematics, with important emphasis on modern probabilistic approaches, starting from the stochastic quantization of Euclidean field theories. The semester opens with 4 really amazing lectures by Antti Kupiainen on the renormalization group (supplemented by &lt;a href="https://www.newton.ac.uk/files/seminar/20180903100011001-1370338.pdf"&gt;lecture notes&lt;/a&gt;). He works with Euclidean &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0085em;vertical-align:-0.1944em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt; in all dimensions, on the lattice and in the continuum limit, and explains everything that can happen. He distinguishes very well the IR scaling limit and UV continuum limit problems, the various fixed point structures, the easy and hard problems, many issues which had always been quite confused in my mind. It&amp;rsquo;s a pleasure to listen to people who understand what they are doing. There is another talk, more like a work in progress, where Martin Hairer attempts the stochastic quantization of Yang-Mills (which starts from a quite original explanation of what a gauge theory is!). I have not had much time to check the other talks, but the whole program looks really interesting (with a lot of different ways to define rigorously &lt;span class="katex"&gt;&lt;span class="katex-mathml"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML"&gt;&lt;semantics&gt;&lt;mrow&gt;&lt;msubsup&gt;&lt;mi&gt;ϕ&lt;/mi&gt;&lt;mn&gt;3&lt;/mn&gt;&lt;mn&gt;4&lt;/mn&gt;&lt;/msubsup&gt;&lt;/mrow&gt;&lt;annotation encoding="application/x-tex"&gt;\phi^4_3&lt;/annotation&gt;&lt;/semantics&gt;&lt;/math&gt;&lt;/span&gt;&lt;span class="katex-html" aria-hidden="true"&gt;&lt;span class="katex-base"&gt;&lt;span class="katex-strut" style="height:1.0622em;vertical-align:-0.2481em;"&gt;&lt;/span&gt;&lt;span class="mord"&gt;&lt;span class="mord mathnormal"&gt;ϕ&lt;/span&gt;&lt;span class="msupsub"&gt;&lt;span class="vlist-t vlist-t2"&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.8141em;"&gt;&lt;span style="top:-2.4519em;margin-left:0em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;3&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span style="top:-3.063em;margin-right:0.05em;"&gt;&lt;span class="pstrut" style="height:2.7em;"&gt;&lt;/span&gt;&lt;span class="katex-sizing reset-size6 size3 mtight"&gt;&lt;span class="mord mtight"&gt;4&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-s"&gt;​&lt;/span&gt;&lt;/span&gt;&lt;span class="vlist-r"&gt;&lt;span class="vlist" style="height:0.2481em;"&gt;&lt;span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;&lt;/span&gt;). I watch these while ironing my shirts, so I will know more at the next laundry.&lt;/p&gt;&#10;</description></item><item><title>Great work from friends</title><link>https://atilloy.com/2019/03/13/great-work-from-friends/</link><guid>https://atilloy.com/2019/03/13/great-work-from-friends/</guid><pubDate>Wed, 13 Mar 2019 10:30:15 +0100</pubDate><description>&lt;p&gt;My smart friends have been doing great work recently and I think it deserves attention.&lt;/p&gt;&#10;&lt;h2 id="i-understanding-deep-neural-networks-theoretically"&gt;I Understanding deep neural networks theoretically&lt;/h2&gt;&#10;&lt;p&gt;&lt;a href="https://fr.linkedin.com/in/jonathan-donier-b1151461"&gt;Jonathan Donier&lt;/a&gt;, who now works for &lt;em&gt;Spotify&lt;/em&gt; in London after a PhD in applied maths in Paris, has put a series of 3 fundamental articles on theoretical machine learning:&lt;/p&gt;&#10;&lt;p&gt;1) &lt;strong&gt;Capacity allocation analysis of neural networks: A tool for principled architecture design&lt;/strong&gt; &amp;ndash; &lt;a href="http://arxiv.org/abs/1902.04485"&gt;arXiv:1902.04485&lt;/a&gt;&lt;br&gt;&#10;2) &lt;strong&gt;Capacity allocation through neural network layers&lt;/strong&gt; &amp;ndash; &lt;a href="http://arxiv.org/abs/1902.08572"&gt;arXiv:1902.08572&lt;/a&gt;&lt;br&gt;&#10;3) &lt;strong&gt;Scaling up deep neural networks: a capacity allocation perspective&lt;/strong&gt; &amp;ndash; &lt;a href="http://arxiv.org/abs/1903.04455"&gt;arXiv:1903.04455&lt;/a&gt;&lt;/p&gt;&#10;&lt;p&gt;In these papers, Jonathan defines and explores the notion of &lt;em&gt;capacity&lt;/em&gt; &lt;em&gt;allocation&lt;/em&gt; of a neural network, which formalizes the intuitive idea that some parts of a network encode more information about certain parts of the input space. The objective is to understand how a given architecture of network manages to capture the structure of correlations in the input. Ultimately, this should allow one to go beyond fuzzily grounded heuristics and expensive trial and error in order to design networks with a topology adapted to the problem right from the start.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2019/03/image.png"&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Jonathan very progressively builds up the theory from basic definitions to non-trivial scaling prescriptions for deep networks. The first paper defines the &lt;em&gt;capacity&lt;/em&gt; rigorously in the simplest settings and deals mostly with the linear case. The second one considers special non-linear settings where the capacity analysis can still be carried out exactly and where one gets insights about the decoupling role of non-linearity. The final one puts all the pieces together and, among other things, allows to rigorously recover many initialization prescriptions for deep networks that where known only from heuristics. This super quick summary does not do justice to the content: this series of papers is, in my opinion, a major advance in the theoretical understanding of deep neural networks.&lt;/p&gt;&#10;&lt;h2 id="ii-making-measurements-crystal-clear-in-bohmian-mechanics"&gt;II Making measurements crystal clear in Bohmian mechanics&lt;/h2&gt;&#10;&lt;p&gt;&lt;a href="https://dustinlazarovici.com/"&gt;Dustin Lazarovici&lt;/a&gt;, who is now a philosopher of physics in Lausanne after a PhD in mathematical physics in Munich, has put online a very clear paper explaining how position measurements work in Bohmian mechanics and what their relation with particle positions is.&lt;/p&gt;&#10;&lt;p&gt;&lt;strong&gt;Position Measurements and the Empirical Status of Particles in Bohmian Mechanics&lt;/strong&gt; &amp;ndash; &lt;a href="https://arxiv.org/abs/1903.04555"&gt;arXiv:1903.04555&lt;/a&gt;&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2019/03/image-1.png"&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Dustin is perhaps one of the people who has the clearest mind on foundations and Bohmian mechanics in particular. The notion of measurement in Bohmian mechanics is usually so deeply misunderstood that Dustin&amp;rsquo;s concise explanation is a great reference for anyone interested in these questions. I particularly enjoyed the very end, where the link (or lack of link) with consciousness is precisely discussed. I think it exemplifies what useful work by philosophers of physics can be like: not muddling the water (as physicists usually think philosophers do) but sharpening the reasoning to save physicists from their own confusion.&lt;/p&gt;&#10;&lt;h2 id="iii-popularizing-tricky-mathematical-notions"&gt;III Popularizing tricky mathematical notions&lt;/h2&gt;&#10;&lt;p&gt;&lt;a href="http://www.antoinebourget.org/blog/"&gt;Antoine Bourget&lt;/a&gt;, who is now a postdoc at &lt;em&gt;Imperial College&lt;/em&gt; in London, after a postdoc in Oviedo and a PhD at ENS in Paris (in the same office as me), has put a series of pedagogical videos on Youtube, through his account &lt;a href="https://www.youtube.com/user/antoinebrgt"&gt;&lt;em&gt;Scientia Egregia&lt;/em&gt;&lt;/a&gt;.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://atilloy.com/wp-content/uploads/2019/03/image-2.png"&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;The videos are in French, and I recommend in particular the &lt;em&gt;&lt;a href="https://www.youtube.com/watch?v=SU6TsD3TM4k"&gt;dictionnaire entre algèbre et géométrie&lt;/a&gt;&lt;/em&gt;. Antoine starts with many simple examples to show the subtleties and motivate the definitions. He explains very well how one constructs mathematical notions to fit a certain intuition, a certain purpose, and thereby manages to make &amp;ldquo;obvious&amp;rdquo; really non-trivial concepts. Go check his &lt;a href="https://www.youtube.com/user/antoinebrgt"&gt;videos&lt;/a&gt; so that he gets pressure to make more.&lt;/p&gt;&#10;</description></item></channel></rss>