#PhysicsJournalClub
"Dark Matter Search Results from 4.2  Tonne−Years of Exposure of the LUX-ZEPLIN (LZ) Experiment"
by a LOT of people

https://doi.org/10.1103/4dyc-z8zf

tl;dr: No, they have found no trace of Dark Matter.

#PhysicsJournalClub
"Training of Physical Neural Networks"
by a LOT of people (including @FMarquardtGroup and @sylvaingigan, who I think are the only ones with a presence here in Mastodon)

https://arxiv.org/pdf/2406.03372

Neural Networks are currently the most popular architecture for Machine Learning, but their digital implementation (as powerful as it is) is rapidly hitting a point of diminishing returns, with skyrocketing costs and infrastructure requirements.
Physical neural networks attempt to side-step this problem by letting Physics to naturally do as much of the work as possible. A fundamental issue with such an approach is that, contrary to a digital model, you often don't have immediate and fast access to all the nodes and weights making up the network, which makes training a non-trivial problem.
This review gives a overview of the various approaches people have developed to perform training of a physical neural network in an efficient way, discussing in details when and under which conditions we can expect a physical neural network to have an advantage over its digital counterpart.

The biggest selling point of this review is that it is both legible and understandable (which is a rarity in the machine learning literature). Kudos to whoever of the authors actually did the writing 🙂

ps
Before people start jumping up and down: we are NOT talking about generative AI here.

#PhysicsJournalClub
"Model-free estimation of the Cramér–Rao bound for deep learning microscopy in complex media"
by I. Starshynov et al.

Nat. Photon. (2025)
https://doi.org/10.1038/s41566-025-01657-6

As everybody who ever tried to orient themselves while immersed in thick fog knows, scattering scrambles information. The question "how much information is still there?" is not particularly interesting as the answer is "essentially all of it", as elastic scattering can't destroy information. A much more interesting question is "how much information can we retrieve?" In order to even try to give an answer we need to be a bit more specific, so the authors placed a small reflective surface behind a scattering layer and asked how much information about its transverse position could be retrieved. This is a well-posed question, and the answer takes the form of a "Cramér–Rao bound" (https://en.wikipedia.org/wiki/Cram%C3%A9r%E2%80%93Rao_bound).
After estimating this upper bound, the authors investigate how well a trained neural network can do at this task, and show that a specifically built convolutional neural network can almost reach the theoretical bound.

[Conflict of interest: Ilya Starshynov (the first author) did his PhD in my group.]

#Physics #InformationTheory #Optics #MachineLearning

#PhysicsJournalClub
"Temperature as joules per bit"
by C.A. Bédard, S. Berthelette, X. Coiteux-Roy, and S. Wolf

Am. J. Phys. 93, 390 (2025)
https://doi.org/10.1119/5.0198820

Entropy is an important but largely misunderstood quantity. A lot of this confusion arise from its original formulation within the framework of Thermodynamics. Looking at it from a microscopic point of view (i.e. approaching it as a Statistical Mechanics problem) makes it a lot more digestible, but its ties to Thermodynamics still creates a lot of unnecessary complications.
In this paper the authors suggest that by removing the forced connection between entropy and the Kelvin temperature scale, one can rethink entropy purely in terms of information capacity of a Physical system, which takes away a lot of the difficulties usually plaguing the understanding of what entropy is actually about.
I don't think the SI will ever consider their suggestion to remove Kelvins as a fundamental unit and include bits, but this paper will be a great boon to any student banging their head against the idea of entropy for the first (or second, or third) time.

#Physics #Entropy #Thermodynamics #StatisticalMechanics

#PhysicsJournalClub
"Direct observation of colloidal quasicrystallization"
by Y. Gao, B. Sprinkle, DWM Marr, and N. Wu

Nat. Phys. (2025)
doi.org/10.1038/s41567-025-02859-z

Quasicrystals are weird. When you solidify something it tends to get into a high-order state: a crystal. If you cool it down too fast so it doesn't manage to make a monocrystal it will form a polycrystalline state or, worse case scenario, something completely amorphous like a glass.
But quasicrystals are weird. They are ordered structures that lack a periodicity and making them is not easy.
In this paper the authors show how paramagnetic colloidal microspheres (i.e. big enough to be clearly visible under a microscope) subject to an electromagnetic field spontaneously arrange themselves into a quasicrystal.

This is 100% not my field, but the ability to create quasicrystals on demand looks so cool!

#Physics #Chemistry #Quasicrystal

#PhysicsJournalClub
F. Diacu "The Solution of the n-body Problem"
Math. Intell. 18, 66 (1996)
https://doi.org/10.1007/BF03024313

In Physics we are often guilty of cutting corners when it comes to Mathematics (to be honest, we are in very good company). One example is in Chaos theory, where statements like "the 3-body problem is unsolvable" are common but misleading.
The author of this paper actually complains about his fellow Mathematicians colleagues (not about Physicists), and then explain in a very understandable way how the sentence "the 3-body problem is unsolvable" must be understood (spoiler: it just means we have fewer conserved quantities than degrees of freedom), how a full solution to the n-body problem in terms of an infinite but convergent series had been found already in 1991 by a Chinese student, and how this solution (albeit correct) is completely useless for any practical purpose, as it converges horribly slow.

An easy and very interesting reading if you know just a little bit about classical mechanics.

#Chaos #3bodyProblem

#PhysicsJournalClub
Today a paper I contributed to!

S. Kendall et al. "Dynamically reconfigurable 2D polarization-agnostic image edge-detection using nonvolatile phase-change metasurfaces"

Optics Express 33, 8971 (2025)
https://doi.org/10.1364/OE.543602

Phase-change materials are media with two metastable solid phases (usually an amorphous and a crystalline phase) with different optical properties. They are commonly used in CD/DVDs where you can use a laser to melt a small volume of the material and let it solidify in the phase you prefer (depending on how fast you let it cool down), creating a binary pattern in the refractive index which encode whatever data you wanted to write.

Beside their use in data storage, phase-changing materials can be used to create structures with two different optical responses, that can be switched at will*.

In this paper we show (simulations only so far, but the experimental results are coming) how to design a structure that is either transparent, or highlight the edges of any image passing through it.

* Caveats apply.

#PhysicsJournalClub
"Three-dimensional holographic imaging of incoherent objects through scattering media"
by Y. Baek, H. de Aguiar and @sylvaingigan
https://arxiv.org/abs/2502.01475
#optics #physics #imaging

As you daily experience anytime you look at anything, light scattering severely impairs your ability to image (mild scattering like mist makes things in a distance fuzzy, strong scattering like your own body makes it completely impossible to see what is happening inside or behind it). On one hand this is good, as it allows us to see where (e.g.) trees are so we don't bump into them. On the other hand there are a LOT of situations where you would really like to see what is going on behind a scattering medium (surely it would save a lot of exploratory surgeries).
The problem of imaging through a scattering medium is largely unsolvable in its most general form, but there are a lot of special cases where you can go surprisingly far, and people (me included) have spent a lot of time checking exactly how far.

In this paper the authors consider a set of small fluorescent objects behind a not-too-thick scattering medium, and look for a way to retrieve their 3D arrangement.
Problem: fluorescent emission means incoherent emission, so the phase information (which encodes a lot of information about position) is lost. Still, we can rely on the assumption that there is a finite (ideally not too large) amount of point emitters. Since each emitter is point-like, if we only measure the light that reaches us through the scattering medium at a single frequency (to be more realistic, a small bandwidth), we will see the incoherent sum of a speckle pattern per fluorescent emitter.
1/2

#PhysicsJournalClub
"Emergence of collective oscillations in massive human crowds"
by Francois Gu et al.
Nature 638, 112 (2025)
https://doi.org/10.1038/s41586-024-08514-6

The flow of granular media (think nuts or sand in a pipe) is a notoriously difficult system to deal with, with a smooth flow suddenly turning into a jam that completely prevents any movement.
People moving is even harder, because most of us (not all) look where we go and make some sort of informed decision about how to move next depending on our surrounding. If there is a lot of space things tend to go smoothly, but what if there is a LOT of people?

In the paper the authors record and analyse people movement at the San Fermín festival in Pamplona (Spain). Before the beginning of the run, there is a lot of people pressed in a not too big square. When the density passes 4 people per m² they observe the spontaneous creation of localized group movement along circles ("vortices") for no apparent reason. The paper contains a lot of discussion on how one can model this.

Do I understand the deep reason for this? NO.
Is this utterly fascinating? Yes.
Do I wish I could replicate this at Saint Ubaldo day in Gubbio (https://en.wikipedia.org/wiki/Saint_Ubaldo_Day)? Totally YES!!! 😀