Search

We set out to explore the possibility of investigating the critical behavior of systems with first-order phase transition using deep machine learning. We propose a machine learning protocol with ternary classification of instantaneous spin configurations using known values of disordered phase energy and ordered phase energy. The trained neural network is used to predict whether a given sample belong to one or the other phase of matter. This allows us to estimate the probability that configurations with a certain energy belong to the ordered phase, mixed phase and, disordered phase. From these probabilities, we obtained estimates of the values of the critical energies and the latent heat for the Potts model with 10 and 20 components, which undergoes a strong discontinuous transition. We also find that the probabilities can reflect geometric transitions in the mixed phase., Comment: 5 pages, 5 figures, 4 tables

Long-time series of neuronal recordings are resulting from the activity of connected neuronal networks. Yet how neuronal properties can be extracted remains empirical. We review here the data analysis based on network models to recover physiological parameters from electrophysiological and calcium recordings in neurons and astrocytes. After, we present the recording techniques and activation events, such as burst and interburst and Up and Down states. We then describe time-serie segmentation methods developed to detect and to segment these events. To interpret the statistics extracted from time series, we present computational models of neuronal populations based on synaptic short-term plasticity and After hyperpolarization. We discuss how these models are calibrated so that they can reproduce the statistics observed in the experimental time series. They serve to extract specific parameters by comparing numerical and experimental statistical moment or entire distributions. Finally, we discus cases where calibrated models are used to predict the selective impact of some parameters on the circuit behavior, properties that would otherwise be difficult to dissect experimentally.

We describe how to use quantum linear algebra to simulate a physically realistic model of disordered non-interacting electrons on exponentially many lattice sites. The physics of disordered electrons outside of one dimension challenges classical computation due to the critical nature of the Anderson localization transition or exponential localization lengths, while the atypical distribution of the local density of states limits the power of disorder averaged approaches. We overcome this by simulating an exponentially large disorder instance using a block-encoded hopping matrix of physical form where disorder is introduced by pseudorandom functions. Key physical quantities, including the reduced density matrix, Green's function, and local density of states, as well as bulk-averaged observables such as the linear conductivity, can then be computed using quantum singular value transformation, quantum amplitude estimation, and trace estimation., Comment: 12 pages, 2 figures

Scale-free localization in non-Hermitian systems is a distinctive type of localization where the localization length of certain eigenstates, known as scale-free localized (SFL) states, scales proportionally with the system size. Unlike skin states, where the localization length is independent of the system size, SFL states maintain a spatial profile that remains invariant as the system size changes. We consider a model involving a single non-Hermitian impurity in an otherwise Hermitian one-dimensional lattice. Introducing disorder into this system transforms SFL states into Anderson-localized states. In contrast to the Hatano-Nelson model, where disorder typically leads to the localization of skin states and a size-independent Anderson transition, the scale-free localization in our model causes a size-dependent Anderson transition., Comment: 7 pages, 5 figures

Lagrangian Neural Networks (LNNs) are a powerful tool for addressing physical systems, particularly those governed by conservation laws. LNNs can parametrize the Lagrangian of a system to predict trajectories with nearly conserved energy. These techniques have proven effective in unconstrained systems as well as those with holonomic constraints. In this work, we adapt LNN techniques to mechanical systems with nonholonomic constraints. We test our approach on some well-known examples with nonholonomic constraints, showing that incorporating these restrictions into the neural network's learning improves not only trajectory estimation accuracy but also ensures adherence to constraints and exhibits better energy behavior compared to the unconstrained counterpart.

Scrambling unitary dynamics in a quantum system transmutes local quantum information into a non-local web of correlations which manifests itself in a complex spatio-temporal pattern of entanglement. In such a context, we show there can exist three distinct dynamical phases characterised by qualitatively different forms of quantum correlations between two disjoint subsystems of the system. Transitions between these phases are driven by the relative sizes of the subsystems and the degree scrambling that the dynamics effects. Besides a phase which has no quantum correlations as manifested by vanishing entanglement between the parts and a phase which has non-trivial quantum correlations quantified by a finite entanglement monotone, we reveal a new phase transition within the entangled phase which separates phases wherein the quantum correlations are invisible or visible to measurements on one of the subsystems. This is encoded in the qualitatively different properties of the ensemble of states on one of the subsystems conditioned on the various measurement outcomes on the other subsystem. This provides a new characterisation of entanglement phases in terms of their response to measurements instead of the more ubiquitous measurement-induced entanglement transitions. Our results have implications for the kind of tasks that can be performed using measurement feedback within the framework of quantum interactive dynamics., Comment: 7 pages, 1 figure + Supplementary Material (5 pages, 2 figures)

We propose an entropy-related function (non-repeatability) that describes dynamical behaviors in complex systems. A normalized version of this function (mutability) has been previously used in statistical physics. To illustrate their characteristics, we apply these functions to different systems: (a) magnetic moments on a square lattice and (b) real seismic data extracted from the IPOC-2007-2014 catalog. These systems are well-established in the literature, making them suitable benchmarks for testing the new approach. Shannon entropy is used as a reference to facilitate comparison, enabling us to highlight similarities, differences, and the potential benefits of the new measure. Notably, non-repeatability and mutability are sensitive to the order in which the data sequence is collected, distinguishing them from traditional entropy measures.

Symmetry of non-Hermitian matrices underpins many physical phenomena. In particular, chaotic open quantum systems exhibit universal bulk spectral correlations classified on the basis of time-reversal symmetry$^{\dagger}$ (TRS$^{\dagger}$), coinciding with those of non-Hermitian random matrices in the same symmetry class. Here, we analytically study the spectral correlations of non-Hermitian random matrices in the presence of TRS$^{\dagger}$ with signs $+1$ and $-1$, corresponding to symmetry classes AI$^{\dagger}$ and AII$^{\dagger}$, respectively. Using the fermionic replica non-linear sigma model approach, we derive $n$-fold integral expressions for the $n$th moment of the one-point and two-point characteristic polynomials. Performing the replica limit $n\to 0$, we qualitatively reproduce the density of states and level-level correlations of non-Hermitian random matrices with TRS$^{\dagger}$., Comment: 21 pages, 9 figures

A neural network-based machine learning potential energy surface (PES) expressed in a matrix product operator (NN-MPO) is proposed. The MPO form enables efficient evaluation of high-dimensional integrals that arise in solving the time-dependent and time-independent Schr\"odinger equation and effectively overcomes the so-called curse of dimensionality. This starkly contrasts with other neural network-based machine learning PES methods, such as multi-layer perceptrons (MLPs), where evaluating high-dimensional integrals is not straightforward due to the fully connected topology in their backbone architecture. Nevertheless, the NN-MPO retains the high representational capacity of neural networks. NN-MPO can achieve spectroscopic accuracy with a test mean absolute error (MAE) of 3.03 cm$^{-1}$ for a fully coupled six-dimensional ab initio PES, using only 625 training points distributed across a 0 to 17,000 cm$^{-1}$ energy range. Our Python implementation is available at https://github.com/KenHino/Pompon., Comment: 11 pages, 10 figures

The Gardner transition in structural glasses is characterized by full-replica symmetry breaking of the free-energy landscape and the onset of anomalous aging dynamics due to marginal stability. Here we show that this transition also has a structural signature in finite-dimensional glasses consisting of hard spheres and disks. By analyzing the distribution of inter-particle gaps in the simulated static configurations at different pressures, we find that the Gardner transition coincides with the emergence of two well-known jamming scalings in the gap distribution, which enables the extraction of a structural order parameter. The jamming scalings reflect a compressible effective force network formed by contact and quasi-contact gaps, while non-contact gaps that do not participate in the effective force network are incompressible. Our results suggest that the Gardner transition in hard-particle glasses is a precursor of the jamming transition. The proposed structural signature and order parameter provide a convenient approach to detecting the Gardner transition in future granular experiments., Comment: 8 pages, 8 figures

Triadic interactions are general mechanisms by which a node or neuron can regulate directly the link or synapse between other two neurons. The regulation takes place in a familiar way by either depressing or facilitating synaptic transmission. Such interactions are ubiquitous in neural systems, accounting for axo-axonic synapses and tripartite synapses mediated by astrocytes, for instance, and have been related to neuronal and synaptic processes at different time-scales, including short and long-term synaptic plasticity. In the field of network science, triadic interactions have been shown to produce complex spatio-temporal patterns of connectivity. Here, we investigate the emergent behavior of an in silico neural medium constituted by a population of leaky integrate-and-fire neurons with triadic interactions. We observe that, depending on relevant parameters defining triadic interactions, different activity patterns emerge. These include i) a silent phase, ii) a low-activity phase in which complex spatio-temporal patterns of low neuronal firing rate emerge that propagate through the medium, iii) a high-activity phase characterized by complex spatio-temporal patterns of high neuronal firing rate that propagate through the neural medium as waves of high firing activity over a bulk of low activity neurons, and iv) a pseudo-blinking phase in which the neural medium switches between high and low activity states, in a similar fashion to up/down state transitions. Here we analyse in depth the features of such patterns and relate our findings to the recently proposed model of triadic percolation., Comment: 22 pages, 9 figures

Inhomogeneous quantum chains have recently been considered in the context of developing novel discrete realizations of holographic dualities. To advance this programme, we explore the ground states of infinite chains with large number $N$ of Majorana fermions on each site, which interact via on-site $q$-body Sachdev-Ye-Kitaev (SYK) couplings, as well as via additional inhomogeneous hopping terms between nearest-neighbour sites. The hopping parameters are either aperiodically or randomly distributed. Our approach unifies techniques to solve SYK-like models in the large $N$ limit with a real-space renormalization group method known as strong-disorder renormalization group (SDRG). We show that the SDRG decimation of SYK dots linked by a strong hopping induces an effective hopping interaction between their neighbouring sites. If two decimated sites are nearest neighbours, in the large $q$ limit their local ground states admit a holographic dual description in terms of eternal traversable wormholes. At the end of the SDRG procedure, we obtain a factorised ground state of the infinite inhomogeneous SYK chains that we consider, which has a spacetime description involving a sequence of wormholes. This amounts to a local near-boundary description of the bulk geometry in the context of discrete holography., Comment: 81 pages, 7 figures

Quantum circuits have become a powerful tool in the study of many-body quantum physics, providing insights into both fast-thermalizing chaotic and non-thermalizing integrable many-body dynamics. In this work, we explore a distinct intermediate class - glassy quantum systems - where thermalization occurs, but over very long timescales. We introduce and analyze a Floquet random quantum circuit that can be tuned between glassy and fully ergodic behavior through a single adjustable parameter. This circuit can be understood as the unitary analog of the block Rosenzweig-Porter model, which is defined by a Hamiltonian. Using an effective field theory for random quantum circuits, we analyze the correlations between quasienergy eigenstates and thereby determine the time evolution of the disorder-averaged density matrix. In the intermediate regime the circuit displays a two-step thermalization process: an initial relaxation within weakly coupled sectors followed by a later, global thermalization. We also show that the ramp of the spectral form factor is enhanced by a factor of the number of sectors in the glassy regime, and at early times in the intermediate regime. These results indicate that quantum circuits provide an ideal platform for the exploration of nontrivial thermalization dynamics in many-body quantum systems, offering deeper insights into quantum thermalization., Comment: 7+8 pages, 3+2 figures

Solutions of the wave equations for time-independent disordered media can exhibit Anderson localization where instead of wave propagation we observe their localization around different points in space. Photonic time crystals are spatially homogeneous media in which the refractive index changes periodically in time, leading to the formation of bands in the wave number domain. By analogy to Anderson localization in space, one might expect that the presence of temporal disorder in photonic time crystals would lead to Anderson localization in the time domain. Here, we show that indeed periodic modulations of the refractive index with the addition of temporal disorder lead to Anderson localization in time, where an electromagnetic field can emerge from the temporally modulated medium at a certain moment in time and then decay exponentially over time. Thus, we are dealing with a situation where, in a fluctuating three-dimensional medium, the birth and death of waves can occur, and the mechanism of this phenomenon corresponds to Anderson localization., Comment: 4 pages + supplementary material, 3 figures

The winding number is the topological invariant that classifies chiral symmetric Hamiltonians with one-dimensional parametric dependence. In this work we complete our study of the winding number statistics in a random matrix model belonging to the chiral unitary class AIII. We show that in the limit of large matrix dimensions the winding number distribution becomes Gaussian. Our results include expressions for the statistical moments of the winding number and for the k-point correlation function of the winding number density.

The term sample dependence describes the troublesome tendency of nominally equivalent samples to exhibit different physical properties. High entropy oxides (HEOs) are a class of materials where sample dependence has the potential to be particularly profound due to their inherent chemical complexity. In this work, we prepare a spinel HEO of identical nominal composition by five distinct methods, spanning a range of thermodynamic and kinetic conditions: solid state, high pressure, hydrothermal, molten salt, and combustion syntheses. By structurally characterizing these five samples across all length scales with a variety of x-ray methods, we find that while the average structure is unaltered, the samples vary significantly in their local structures and their microstructures. The most profound differences are observed at intermediate length scales, both in terms of crystallite morphology and cation homogeneity. As revealed by x-ray fluorescence microscopy ideal cation homogeneity is achieved only in the case of combustion synthesis. These structural differences in turn significantly alter the observed functional properties, which we demonstrate via characterization of their magnetic response. While ferrimagnetic order is retained across all five samples, the sharpness of the transition, the size of the saturated moment, and the coercivity all show marked variations with synthesis method. We conclude that the chemical flexibility inherent to HEOs is complemented by strong synthesis method dependence, providing another axis along which to optimize these materials for a wide range of applications., Comment: 14 pages, 5 figures

The surface code, one of the leading candidates for quantum error correction, is known to protect encoded quantum information against stochastic, i.e., incoherent errors. The protection against coherent errors, such as from unwanted gate rotations, is however understood only for special cases, such as rotations about the $X$ or $Z$ axes. Here we consider generic single-qubit coherent errors in the surface code, i.e., rotations by angle $\alpha$ about an axis that can be chosen arbitrarily. We develop a statistical mechanical mapping for such errors and perform entanglement analysis in transfer matrix space to numerically establish the existence of an error-correcting phase, which we chart in a subspace of rotation axes to estimate the corresponding maximum-likelihood thresholds $\alpha_\text{th}$. The classical statistical mechanics model we derive is a random-bond Ising model with complex couplings and four-spin interactions (i.e., a complex-coupled Ashkin-Teller model). The error correcting phase, $\alpha<\alpha_\text{th}$, where the logical error rate decreases exponentially with code distance, is shown to correspond in transfer matrix space to a gapped one-dimensional quantum Hamiltonian exhibiting spontaneous breaking of a $\mathbb{Z}_2$ symmetry. Our numerical results rest on two key ingredients: (i) we show that the state evolution under the transfer matrix -- a non-unitary (1+1)-dimensional quantum circuit -- can be efficiently numerically simulated using matrix product states. Based on this approach, (ii) we also develop an algorithm to (approximately) sample syndromes based on their Born probability. The $\alpha_\text{th}$ values we find show that the maximum likelihood thresholds for coherent errors are larger than those for the corresponding incoherent errors (from the Pauli twirl), and significantly exceed the values found using minimum weight perfect matching., Comment: 16 pages, 6 figures

The jamming transition of frictionless athermal particles is a paradigm to understand the mechanics of amorphous materials at the atomic scale. Concepts related to the jamming transition and the mechanical response of jammed packings have cross-fertilized into other areas such as atomistic descriptions of the elasticity and plasticity of glasses. In this perspective article, the microscopic mathematical theory of the jamming transition is reviewed from first-principles. The starting point of the derivation is a microscopically-reversible particle-bath Hamiltonian from which the governing equation of motion for the grains under an external deformation is derived. From this equation of motion, microscopic expressions are obtained for both the shear modulus and the viscosity as a function of the distance from the jamming transition (respectively, above and below the transition). Regarding the vanishing of the shear modulus at the unjamming transition, this theory, as originally demonstrated in [Zaccone \& Scossa-Romano, Phys. Rev. B 83, 184205 (2011)], is currently the only quantitative microscopic theory in parameter-free agreement with numerical simulations of [O'Hern et al. Phys. Rev. E 68, 011306 (2003)] for jammed packings. The divergence of the viscosity upon approaching the jamming transition from below is derived here, for the first time, from the same microscopic Hamiltonian. The quantitative microscopic prediction of the diverging viscosity is shown to be in fair agreement with numerical results of sheared 2D soft disks from [Olsson \& Teitel, Phys. Rev. Lett. 99, 178001 (2007)]., Comment: Invited Perspective

Recent experimental work indicates that the Kitaev spin liquid may be realizable close to zero magnetic field in exfoliated $\alpha$-RuCl$_3$ flakes, thus providing a more versatile setting for studying non-Abelian anyons. Here, we propose a robust nanoscale signature of the Kitaev spin liquid that results from its edge states and manifests in the low-energy spin dynamics. In particular, we highlight a singular peak in the dynamical spin structure factor of a zigzag edge whose energy scales linearly with only one component of the magnetic field. This sharp feature in the local spin dynamics directly reflects the bond-directional Kitaev spin interactions and, more generally, the projective symmetries of the Kitaev spin liquid. We demonstrate that our proposed signature survives in the presence of edge disorder as well as non-Kitaev interactions, and provide detailed guidelines for experimentally observing it using inelastic electron tunneling spectroscopy and color-center relaxometry., Comment: 6+2 pages, 4 figures

To analyse how diffusion models learn correlations beyond Gaussian ones, we study the behaviour of higher-order cumulants, or connected n-point functions, under both the forward and backward process. We derive explicit expressions for the moment- and cumulant-generating functionals, in terms of the distribution of the initial data and properties of forward process. It is shown analytically that during the forward process higher-order cumulants are conserved in models without a drift, such as the variance-expanding scheme, and that therefore the endpoint of the forward process maintains nontrivial correlations. We demonstrate that since these correlations are encoded in the score function, higher-order cumulants are learnt in the backward process, also when starting from a normal prior. We confirm our analytical results in an exactly solvable toy model with nonzero cumulants and in scalar lattice field theory., Comment: 21 pages, many figures. Extended version of contribution accepted in the NeurIPS 2024 workshop "Machine Learning and the Physical Sciences"

Research on time crystals concerns the spontaneous breaking of translational symmetry in time, as well as the realization of phenomena and phases known from solid-state physics in the time domain. Periodically driven systems of massive particles are widely used in these studies. In the present work, we consider a photonic system and demonstrate that stable nonlinear propagation of a strong optical wave in a fiber leads to the establishment of quasi-periodic oscillations in the electromagnetic field intensity. A second, weaker signal optical wave propagating in the fiber senses these oscillations and, as a result, undergoes exponential localization in time. This is a temporal analogue of Aubry-Andre localization. If an optical detector is placed at a certain position in the fiber, the temporal localization of the probe wave will be observed in the form of the signal which emerges and then decays as a function of time.

We consider the injectivity property of the ReLU networks layers. Determining the ReLU injectivity capacity (ratio of the number of layer's inputs and outputs) is established as isomorphic to determining the capacity of the so-called $\ell_0$ spherical perceptron. Employing \emph{fully lifted random duality theory} (fl RDT) a powerful program is developed and utilized to handle the $\ell_0$ spherical perceptron and implicitly the ReLU layers injectivity. To put the entire fl RDT machinery in practical use, a sizeable set of numerical evaluations is conducted as well. The lifting mechanism is observed to converge remarkably fast with relative corrections in the estimated quantities not exceeding $\sim 0.1\%$ already on the third level of lifting. Closed form explicit analytical relations among key lifting parameters are uncovered as well. In addition to being of incredible importance in handling all the required numerical work, these relations also shed a new light on beautiful parametric interconnections within the lifting structure. Finally, the obtained results are also shown to fairly closely match the replica predictions from [40].

We show that the notion of critical droplets is central to an understanding of the nature of ground states in the Edwards-Anderson Ising model of a spin glass in arbitrary dimension. Given a specific ground state, suppose the coupling value for a given edge is varied with all other couplings held fixed. Beyond some specific value of the coupling, a droplet will flip leading to a new ground state; we refer to this as the critical droplet for that edge and ground state. We show that the distribution of sizes and energies over all edges for a specific ground state can be used to determine which of the leading scenarios for the spin glass phase is correct. In particular, the existence of low-energy interfaces between incongruent ground states as predicted by replica symmetry breaking is equivalent to the presence of critical droplets whose boundaries comprise a positive fraction of edges in the infinite lattice., Comment: 12 pages, no figures

Recently, a universal relation between the thermal expansion coefficient of glasses $\alpha_g$, their glass-transition temperature Tg, and the so-called fragility index m of the corresponding supercooled liquid state was found to be valid for more than 200 glass formers, namely $\alpha_g/m\propto 1/T_g$ [P. Lunkenheimer et al., Nat. Phys. 19, 694 (2023)]. Here we show that this could also have far-reaching consequences for our understanding of crystal melting. Namely, when considering the empirically founded 2/3-rule, stating that the ratio of Tg and the melting temperature Tm is about 2/3 for almost all materials, for crystals a similar relation, $\alpha_c/m\propto 1/T_m$, should apply. Indeed, we find that the available experimental data are well consistent with such a relation. This implies that the melting of a crystal into an ordinary (non-supercooled) liquid is influenced by the fragility, a property quantifying the non-Arrhenius dynamics in the supercooled-liquid state of the material. We argue that this can be explained by a significant enhancement of the "ideal" (non-cooperative) melting temperature arising from the cooperativity of the particle motion in the liquid state above Tm. Therefore, a reassessment of the currently widely accepted microscopic understanding of crystal melting, still founded on the general ideas that lead to the time-honored Lindemann melting criterion, may be necessary., Comment: 9 pages, 3 figures + 6 pages Supplemental Material

We introduce a machine learning prediction workflow to study the impact of defects on the Raman response of 2D materials. By combining the use of machine-learned interatomic potentials, the Raman-active $\Gamma$-weighted density of states method and splitting configurations in independant patches, we are able to reach simulation sizes in the tens of thousands of atoms, with diagonalization now being the main bottleneck of the simulation. We apply the method to two systems, isotopic graphene and defective hexagonal boron nitride, and compare our predicted Raman response to experimental results, with good agreement. Our method opens up many possibilities for future studies of Raman response in solid-state physics., Comment: 9 pages, 6 figures

Structural two level systems (TLSs) ubiquitous in amorphous solids are dramatically sensitive to thermal cycling to about $20$K and then back to low temperature, a process upon which the excitation energy of most TLSs is significantly changed. Using Monte Carlo simulations we demonstrate that this phenomenon is not contained within the standard tunneling model, but is well explained by a model that includes an additional set of TLSs that are pseudo-gapped at low energies, yet possess strong strain interaction through which they generate significant dynamical disorder upon thermal cycling. Our results give support for the general applicability of the Two-TLS model to amorphous solids at low temperatures, in that nearing our understanding of the low-temperature universality of phonon attenuation in amorphous solids. Further, we suggest thermal cycling as a unique protocol to distinguish TLS noise from other noise sources in quantum circuits., Comment: 8 pages, 5 figures (including Supplementary Material)

Mosaic lattice models have been recently introduced as a special class of disordered systems displaying resonance energies, multiple mobility edges and anomalous transport properties. In such systems on-site potential disorder, either uncorrelated or incommensurate, is introduced solely at every equally-spaced sites within the lattice, with a spacing $M \geq 2$. A remarkable property of disordered mosaic lattices is the persistence of extended states at some resonance frequencies that prevent complete Anderson localization, even in the strong disorder regime. Here we introduce a broader class of mosaic lattices and derive general expressions of mobility edges and localization length for incommensurate sinusoidal disorder, which generalize previous results [Y. Wang {\it et al.}, Phys. Rev. Lett. {\bf 125}, 196604 (2020)]. For both incommensurate and uncorrelated disorder, we prove that Anderson localization is protected by the open gaps of the disorder-free lattice, and derive some general criteria for complete Anderson localization. The results are illustrated by considering a few models, such as the mosaic Su-Schrieffer-Heeger (SSH) model and the trimer mosaic lattice., Comment: 13 pages, 6 figures, accepted for publication in the Phys. Rev. B

In this paper, we investigate a non-Hermitian Aubry-Andr\'e-Harper model characterized by power-law hoppings ($1/s^{a}$) and a quasi-periodic parameter $\beta$, where $a$ denotes the power-law index, $s$ represents the hopping distance, and $\beta$ belongs to the metallic mean family. In the intermediate phases, we find that ergodic states correspond to complex eigenvalues, multifractal states to real eigenvalues, and localized states may exhibit either complex or real eigenvalues. Moreover, both real and complex energy spectra emerge in the localized phase, with real spectra attributed to pseudo-Hermiticity. Under open boundary conditions, our analysis of fractal dimensions and eigenstates reveals that all ergodic states transform into skin states. Furthermore, we demonstrate that long-range hoppings weaken the skin effect, offering a new perspective for exploring non-Hermitian skin effects., Comment: 8 pages, 9 figures

We develop a theoretical model to investigate wave propagation in media with random time-varying properties, where temporal fluctuations lead to complex scattering dynamics. Focusing on the ensemble-averaged field, we derive an exact expression for the average Green's function in the presence of finite temporal disorder, and extend our analysis to the thermodynamic limit. In contrast to spatial disorder, causality prevents recurrent scattering, allowing us to achieve a non-perturbative solution. We identify two distinct propagation regimes: one where the field oscillates with constant amplitude, and another where it grows exponentially at long times. Our findings offer new insights into wave dynamics in temporally disordered media, with potential applications in time-varying metamaterials, dynamic sensing, and imaging in turbulent or chaotic environments.

Eigenstate multifractality, a hallmark of non-interacting disordered metals which can potentially be observed in many-body localized states as well, is characterized by anomalous slow dynamics and appears relevant for many areas of quantum physics from measurement-driven systems to superconductivity. We propose a novel approach to achieve non-ergodic multifractal (NEM) states without disorder by iteratively introducing defects into a crystal lattice, reshaping it from a plain structure into fractal geometry. By comprehensive analysis of the Sierpi\'nski gasket case, we find a robust evidence of the emergence of NEM states that go beyond the conventional classification of quantum states and designate new pathways for quantum transport studies. We discuss potential experimental signatures of these states., Comment: 19 pages, 18 figures

We have studied the diluted Heisenberg spin glass model in a 3-component random field for the commonly-used one-dimensional long-range model where the probability that two spins separated by a distance $r$ interact with one another falls as $1/r^{2 \sigma}$, for two values of $\sigma$, $0.75$ and $0.85$. No de Almeida-Thouless line is expected at these $\sigma$ values. The spin glass correlation length $\xi_{\text{SG}}$ varies with the random field as expected from the Imry-Ma argument and the droplet scaling picture of spin glasses. However, when $\xi_{\text{SG}}$ becomes comparable to the system size $L$, there are departures which we attribute to the features deriving from the TNT picture of spin glasses. For the case $\sigma =0.85$ these features go away for system sizes with $L >L^*$, where $L^*$ is large ($\approx 4000-8000$ lattice spacings). In the case of $\sigma = 0.75$ we have been unable to study large enough systems to determine its value of $L^*$. We sketch a renormalization group scenario to explain how these features could arise. On this scenario finite size effects on the droplet scaling picture in low-dimensional spin glasses produce TNT features and some aspects of Parisi's replica symmetry breaking theory of the Sherrington-Kirkpatrick model., Comment: 11 pages, 3 + 3 figures, 2 tables

The edge of a quantum critical system can exhibit multiple distinct types of boundary criticality. We use a numerical real-space renormalization group (RSRG) to study the boundary criticality of a 2d quantum Ising model with random exchange couplings and transverse fields, whose bulk exhibits an infinite randomness critical point. This approach enables an asymptotically numerically exact extraction of universal scaling data from very large systems with many thousands of spins that cannot be efficiently simulated directly. We identify three distinct classes of boundary criticality, and extract key scaling exponents governing boundary-boundary and boundary-bulk correlations and dynamics. We anticipate that this approach can be generalized to studying a broad class of (disordered) boundary criticality, including symmetry-enriched criticality and edge modes of gapless symmetry-protected topological states, in contexts were other numerical methods are restricted to one-dimensional chains.

Current progress in artificial intelligence is centered around so-called large language models that consist of neural networks processing long sequences of high-dimensional vectors called tokens. Statistical physics provides powerful tools to study the functioning of learning with neural networks and has played a recognized role in the development of modern machine learning. The statistical physics approach relies on simplified and analytically tractable models of data. However, simple tractable models for long sequences of high-dimensional tokens are largely underexplored. Inspired by the crucial role models such as the single-layer teacher-student perceptron (aka generalized linear regression) played in the theory of fully connected neural networks, in this paper, we introduce and study the bilinear sequence regression (BSR) as one of the most basic models for sequences of tokens. We note that modern architectures naturally subsume the BSR model due to the skip connections. Building on recent methodological progress, we compute the Bayes-optimal generalization error for the model in the limit of long sequences of high-dimensional tokens, and provide a message-passing algorithm that matches this performance. We quantify the improvement that optimal learning brings with respect to vectorizing the sequence of tokens and learning via simple linear regression. We also unveil surprising properties of the gradient descent algorithms in the BSR model.

The brain's synaptic network, characterized by parallel connections and feedback loops, drives information flow between neurons through a large system with infinitely many degrees of freedom. This system is best modeled by the graph $C^*$-algebra of the underlying directed graph, the Toeplitz-Cuntz-Krieger algebra, which captures the diversity of potential information pathways. Coupled with the gauge action, this graph algebra defines an {\em algebraic quantum system}, and here we demonstrate that its thermodynamic properties provide a natural framework for describing the dynamic mappings of information flow within the network. Specifically, we show that the KMS states of this system yield global statistical measures of neuronal interactions, with computational illustrations based on the {\em C. elegans} synaptic network., Comment: 6 page, 4 figures

We study a stroboscopic quantum Ising model with Fibonacci dynamics. Focusing on boundary spin correlation functions in long but finite chains, our simulations as well as analytical arguments reveal a self-similar phase diagram exhibiting regions with Majorana zero modes (MZM) as well as Majorana golden-ratio modes (MGM). We identify the self-similarity transform which governs the evolution of the phase diagram with increasing simulation time. Integrability-breaking perturbations lead to a temporal decay of the boundary spin correlations, ultimaltely limiting the self-similarity of the phase diagram. Our predictions are testable with current quantum information processors., Comment: 5+4 pages, 5 figures

We propose the first learning scheme for functional differential equations (FDEs). FDEs play a fundamental role in physics, mathematics, and optimal control. However, the numerical analysis of FDEs has faced challenges due to its unrealistic computational costs and has been a long standing problem over decades. Thus, numerical approximations of FDEs have been developed, but they often oversimplify the solutions. To tackle these two issues, we propose a hybrid approach combining physics-informed neural networks (PINNs) with the \textit{cylindrical approximation}. The cylindrical approximation expands functions and functional derivatives with an orthonormal basis and transforms FDEs into high-dimensional PDEs. To validate the reliability of the cylindrical approximation for FDE applications, we prove the convergence theorems of approximated functional derivatives and solutions. Then, the derived high-dimensional PDEs are numerically solved with PINNs. Through the capabilities of PINNs, our approach can handle a broader class of functional derivatives more efficiently than conventional discretization-based methods, improving the scalability of the cylindrical approximation. As a proof of concept, we conduct experiments on two FDEs and demonstrate that our model can successfully achieve typical $L^1$ relative error orders of PINNs $\sim 10^{-3}$. Overall, our work provides a strong backbone for physicists, mathematicians, and machine learning experts to analyze previously challenging FDEs, thereby democratizing their numerical analysis, which has received limited attention. Code is available at \url{https://github.com/TaikiMiyagawa/FunctionalPINN}., Comment: Accepted at NeurIPS 2024. Both authors contributed equally. Some contents are omitted due to arXiv's storage limit. Please refer to the full paper at OpenReview (NeurIPS 2024) or https://github.com/TaikiMiyagawa/FunctionalPINN

The discovery of a charge density wave (CDW) in the antiferromagnetic kagome metal FeGe has prompted interest in the interplay between kagome physics, CDW, and magnetism. However, a crucial aspect for understanding these emergent phenomena-the precise CDW structure-remains ambiguous. Recent studies have assumed uniformly distributed Ge dimers, but this assumption is problematic. The predicted band structure based on this model exhibits an abrupt disappearance of a Ge-$p$ band in the Fermi surface, contradicting experimental observations from angle-resolved photoemission spectroscopy (ARPES). In this study, we propose that a CDW phase with disordered Ge dimers can reconcile theoretical predictions with ARPES results. This model reproduces the observed CDW gaps while preserving the Ge-$p$ band. Depending on experimental conditions, Ge dimers can be randomly distributed or exhibit phase separation from pristine regions. Our findings reveal the crucial role of Ge dimer disorder in the FeGe CDW and suggest potential implications of this disorder for other properties, such as magnetism and transport, in this system., Comment: Supplementary included

We compute the distribution of triplets of stationary points in the energy landscape of the spherical p-spin model, by evaluating the quenched three-point complexity by means of the Kac-Rice formalism. We show the occurrence of transitions in the organization of stationary points in the landscape, identifying regions where local minima and saddles accumulate and cluster around other stationary points, thus displaying the presence of correlations in the landscape. We discuss the implications of these findings for the dynamical exploration of the energy landscape in the activated regime, specifying conditions under which transitions between local minima are expected to exhibit correlated rates and when, conversely, activated jumps are likely to be memoryless.

In all van der Waals layered antiferromagnetic semiconductors investigated so far a negative magnetoresistance has been observed in vertical transport measurements, with characteristic trends that do not depend on applied bias. Here we report vertical transport measurements on layered antiferromagnetic semiconductor CrPS$_4$ that exhibit a drastically different behavior, namely a strongly bias dependent, positive magnetoresistance that is accompanied by pronounced oscillations for devices whose thickness is smaller than 10 nm. We establish that this unexpected behavior originates from transport being space-charge limited, and not injection limited as for layered antiferromagetic semiconductors studid earlier. Our analysis indicates that the positive magnetoresistance and the oscillations only occur when electrons are injected into in-gap defect states, whereas when electrons are injected into the conduction band the magnetoresistance vanishes. We propose a microscopic explanation for the observed phenomena that combines concepts typical of transport through disordered semiconductors with known properties of the CrPS$_4$ magnetic state, which captures all basic experimental observations. Our results illustrate the need to understand in detail the nature of transport through vdW magnets, to extract information about the nature of the order magnetic states and its microscopic properties., Comment: 10pages, 7 figures, under evaluation

Given a sequence of $d \times d$ symmetric matrices $\{\mathbf{W}_i\}_{i=1}^n$, and a margin $\Delta > 0$, we investigate whether it is possible to find signs $(\epsilon_1, \dots, \epsilon_n) \in \{\pm 1\}^n$ such that the operator norm of the signed sum satisfies $\|\sum_{i=1}^n \epsilon_i \mathbf{W}_i\|_\mathrm{op} \leq \Delta$. Kunisky and Zhang (2023) recently introduced a random version of this problem, where the matrices $\{\mathbf{W}_i\}_{i=1}^n$ are drawn from the Gaussian orthogonal ensemble. This model can be seen as a random variant of the celebrated Matrix Spencer conjecture and as a matrix-valued analog of the symmetric binary perceptron in statistical physics. In this work, we establish a satisfiability transition in this problem as $n, d \to \infty$ with $n / d^2 \to \tau > 0$. Our contributions are twofold. First, we prove that the expected number of solutions with margin $\Delta=\kappa \sqrt{n}$ has a sharp threshold at a critical $\tau_1(\kappa)$: for $\tau < \tau_1(\kappa)$ the problem is typically unsatisfiable, while for $\tau > \tau_1(\kappa)$ the average number of solutions becomes exponentially large. Second, combining a second-moment method with recent results from Altschuler (2023) on margin concentration in perceptron-type problems, we identify a second threshold $\tau_2(\kappa)$, such that for $\tau>\tau_2(\kappa)$ the problem admits solutions with high probability. In particular, we establish for the first time that a system of $n = \Theta(d^2)$ Gaussian random matrices can be balanced so that the spectrum of the resulting matrix macroscopically shrinks compared to the typical semicircle law. Our proofs rely on deriving concentration inequalities and large deviation properties for the law of correlated Gaussian matrices under spectral norm constraints, extending results in the literature and offering insights of independent interest., Comment: 29 pages, 2 figures

We investigate analytically and numerically the Anderson localization of quasiparticles in binary Bose mixtures in the presence of the Lee-Huang-Yang quantum and thermal corrections subjected to correlated disordered potentials. We calculate the density profiles, the Bogoliubov quasiparticles modes, and the localization length in both the mixture and droplet phases. We show that for one-dimensional speckle potentials, the peculiar interplay of disorder and the Lee-Huang-Yang fluctuations may enhance localization in one component while inducing delocalization in the other. Our results reveal also that thermal fluctuations lead to the emergence of two and multiple localization maxima. In the droplet state, our findings uncover that the Anderson localization of quasiparticles is weak in the flat-top region since its excitation modes are restricted to those below the particle-emission threshold., Comment: Accepted for publication in Physical Review A,13 pages,9 figures

As large language models (LLMs) become central to AI applications, gaining a deeper understanding of their inner workings is increasingly important. In this work, we analyze the weight matrices of pretrained transformer models -- specifically BERT and Llama -- using random matrix theory (RMT) as a zero-information hypothesis. While randomly initialized weights perfectly agree with RMT predictions, deviations emerge after training, allowing us to locate learned structures within the models. We identify layer-type specific behaviors that are consistent across all blocks and architectures considered. By pinpointing regions that deviate from RMT predictions, we highlight areas of feature learning and confirm this through comparisons with the activation covariance matrices of the corresponding layers. Our method provides a diagnostic tool for identifying relevant regions in transformer weights using only the trained matrices. Additionally, we address the ongoing debate regarding the significance of small singular values in the context of fine-tuning and alignment in LLMs. Our findings reveal that, after fine-tuning, small singular values play a crucial role in the models' capabilities, suggesting that removing them in an already aligned transformer can be detrimental, as it may compromise model alignment., Comment: 17 pages, 14 figures

Motivated by the recent appearance of the trillium lattice in the search for materials hosting spin liquids, we study the ground state of the classical Heisenberg model on its linegraph, the trilline lattice. We find that this network realises the recently proposed notion of a fragile spin liquid in three dimensions. Additionally, we analyze the Ising case and argue for a possible $\mathbb{Z}_2$ quantum spin liquid phase in the corresponding quantum dimer model. Like the well-known $U(1)$ spin liquids, the classical phase hosts moment fractionalisation evidenced in the diluted lattice, but unlike these, it exhibits exponential decay both in spin correlations and interactions between fractionalised moments. This provides the first instance of a purely short-range correlated classical Heisenberg spin liquid in three dimensions.

Extended and critical states are two common phenomena in Fibonacci quasicrystals. In this paper, we first reveal the difference between the extended phase and the critical phase in the extended-critical Fibonacci quasicrystal from the perspectives of quantum transport and Wigner distribution. The transport conductance indicates that the extended-critical transition resembles a metallic-insulating transition. Moreover, the Wigner distributions show that the Wigner distribution of the extended wave function is localized in the momentum direction of the phase space, while that of the critical wave function is sub-extended in the momentum direction of the phase space. Based on the results of entanglement entropy, the extended-critical transition is a thermodynamic phase transition because it is accompanied by decreasing entropy. We engineer a quantum heat cycle engine with the extended-critical quasicrystal as the working medium, and find that there are rich working modes in the engine, such as quantum accelerator, quantum heater and quantum heat engine. Importantly, the extended quasicrystals are more conducive to the realization of quantum heat engines, while the critical quasicrystals are more conducive to the realization of quantum heaters. Our work is an important step toward exploring the rich thermodynamic applications of Fibonacci quasicrystals., Comment: 6 pages, 3 figures

Design of photonic crystals having large bandgaps above a prescribed band is a well-known physical problem with many applications. A connection to an interesting mathematical construction was pointed out some time ago: it had been conjectured that optimal structures for gaps between bands n and n+1 correspond, in case of transverse magnetic polarisation, to rods located at the generators of centroidal Voronoi tessellation (CVT), and in case of transverse electric polarisation to the walls of this tessellation. We discover another mathematical receipt which produces even better solutions: optimal packing of discs in square and triangular tori. It provides solutions qualitatively different from CVT, sometimes increasing the resulting bandgap size in several times. We therefore introduce two new classes of periodic structures with remarkable properties which may find applications in many other areas of modern solid state physics: arrays of particles located at the centers of optimally packed discs on tori, and nets corresponding to the walls of their Voronoi tessellations., Comment: 11 pages. Suggestions/comments/ collaboration ideas are welcomed

We apply a variational Ansatz based on neural networks to the problem of spin-$1/2$ fermions in a harmonic trap interacting through a short distance potential. We showed that standard machine learning techniques lead to a quick convergence to the ground state, especially in weakly coupled cases. Higher couplings can be handled efficiently by increasing the strength of interactions during "training"., Comment: 8 pages, 5 figures

Conformal prediction has emerged as a powerful tool for building prediction intervals that are valid in a distribution-free way. However, its evaluation may be computationally costly, especially in the high-dimensional setting where the dimensionality and sample sizes are both large and of comparable magnitudes. To address this challenge in the context of generalized linear regression, we propose a novel algorithm based on Approximate Message Passing (AMP) to accelerate the computation of prediction intervals using full conformal prediction, by approximating the computation of conformity scores. Our work bridges a gap between modern uncertainty quantification techniques and tools for high-dimensional problems involving the AMP algorithm. We evaluate our method on both synthetic and real data, and show that it produces prediction intervals that are close to the baseline methods, while being orders of magnitude faster. Additionally, in the high-dimensional limit and under assumptions on the data distribution, the conformity scores computed by AMP converge to the one computed exactly, which allows theoretical study and benchmarking of conformal methods in high dimensions.

Density functional theory (DFT) offers a desirable balance between quantitative accuracy and computational efficiency in practical many-electron calculations. Its central component, the exchange-correlation energy functional, has been approximated with increasing levels of complexity ranging from strictly local approximations to nonlocal and orbital-dependent expressions with many tuned parameters. In this work, we formulate a general way of rewriting complex density functionals using deep neural networks in a way that allows for simplified computation of Kohn-Sham potentials as well as higher functional derivatives through automatic differentiation, enabling access to highly nonlinear response functions and forces. These goals are achieved by using a recently developed class of robust neural network models capable of modeling functionals, as opposed to functions, with explicitly enforced spatial symmetries. Functionals treated in this way are then called global density approximations and can be seamlessly integrated with existing DFT workflows. Tests are performed for a dataset featuring a large variety of molecular structures and popular meta-GGA density functionals, where we successfully eliminate orbital dependencies coming from the kinetic energy density, and discover a high degree of transferability to a variety of physical systems. The presented framework is general and could be extended to more complex orbital and energy dependent functionals as well as refined with specialized datasets., Comment: 5 pages, 3 figures (supplementary material 8 pages, 2 figures)

The interplay between quantum and thermal fluctuations in the presence of quenched random disorder is a long-standing open theoretical problem which has been made more urgent by advances in modern experimental techniques. The fragility of charge density wave order to impurities makes this problem of particular interest in understanding a host of real materials, including the cuprate high-temperature superconductors. To address this question, we consider the quantum version of an exactly solvable classical model of two-dimensional randomly pinned incommensurate charge density waves first introduced by us in a recent work, and use the large-$N$ technique to obtain the phase diagram and order parameter correlations. Our theory considers quantum and thermal fluctuations and disorder on equal footing by accounting for all effects non-perturbatively, which reveals a novel crossover between under-damped and over-damped dynamics of the fluctuations of the charge density wave order parameter., Comment: 46 pages, 7 figures

Understanding the origins of complexity is a fundamental challenge with implications for biological and technological systems. Network theory emerges as a powerful tool to model complex systems. Networks are an intuitive framework to represent inter-dependencies among many system components, facilitating the study of both local and global properties. However, it is unclear whether we can define a universal theoretical framework for evolving networks. While basic growth mechanisms, like preferential attachment, recapitulate common properties such as the power-law degree distribution, they fall short in capturing other system-specific properties. Tinkering, on the other hand, has shown to be very successful in generating modular or nested structures "for-free", highlighting the role of internal, non-adaptive mechanisms in the evolution of complexity. Different network extensions, like hypergraphs, have been recently developed to integrate exogenous factors in evolutionary models, as pairwise interactions are insufficient to capture environmentally-mediated species associations. As we confront global societal and climatic challenges, the study of network and hypergraphs provides valuable insights, emphasizing the importance of scientific exploration in understanding and managing complexity., Comment: Contributed chapter to "Nonlinear Dynamics for Biological Systems", M. Stich, J. Carballido-Landeira (Eds), Springer, Switzerland (submitted, 2024)