We gratefully acknowledge support from
the Simons Foundation and member institutions.

Disordered Systems and Neural Networks

New submissions

[ total of 10 entries: 1-10 ]
[ showing up to 2000 entries per page: fewer | more ]

New submissions for Thu, 25 Nov 21

[1]  arXiv:2111.12398 [pdf, other]
Title: Adaptive-weighted Tree Tensor Networks for disordered quantum many-body systems
Comments: 8 pages, 6 figures
Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Quantum Gases (cond-mat.quant-gas); Statistical Mechanics (cond-mat.stat-mech); Strongly Correlated Electrons (cond-mat.str-el); Quantum Physics (quant-ph)

We introduce an adaptive-weighted Tree Tensor Network, for the study of disordered and inhomogeneous quantum many-body systems. This ansatz is assembled on the basis of the random couplings of the physical system with a procedure that considers a tunable weight parameter to prevent completely unbalanced trees. Using this approach, we compute the ground state of the two-dimensional quantum Ising model in the presence of quenched random disorder and frustration, with lattice size up to $32 \times 32$. We compare the results with the ones obtained using the standard homogeneous Tree Tensor Networks and the completely self-assembled Tree Tensor Networks, demonstrating a clear improvement of numerical precision as a function of the weight parameter, especially for large system sizes.

[2]  arXiv:2111.12569 [pdf, ps, other]
Title: Rigorous approaches for spin glass and Gaussian spin glass with P-wise interactions
Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Mathematical Physics (math-ph)

Purpose of this paper is to face up to P-spin glass and Gaussian P-spin model, i.e. spin glasses with polynomial interactions of degree P > 2. We consider the replica symmetry and first step of replica simmetry breaking assumptions and we solve the models via transport equation and Guerra's interpolating technique, showing that we reach the same results. \\ Thus, using rigorous approaches, we recover the same expression for quenched statistical pressure and self-consistency equation in both assumption found with other techniques, including the well-known \textit{replica trick} technique. \\ At the end, we show that for $P=2$ the Gaussian P-spin glass model is intrinsecally RS.

[3]  arXiv:2111.12644 [pdf, other]
Title: A new Phenomenon: Glass Paramagnetism. Further Experimental and Theoretical Details
Comments: 29 pages, 24 figures, submitted for publication
Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Mesoscale and Nanoscale Physics (cond-mat.mes-hall); Materials Science (cond-mat.mtrl-sci)

In a recent manuscript, the discovery of a new phenomenon in glasses has been reported: {\it glass paramagnetism}, that is the intrinsic magnetisation developed by a substance that has failed to crystallise in a temperature quench when placed in an external magnetic field. The field- ($H$) and especially the temperature- ($T$) dependence of the intrinsic magnetisation $M_{intr}=M_{intr}(H,T)$ is very unexpected, with broad peaks developing for high $H/T$ and marked $T$- oscillations at fixed $H$ for intermediate-to-high temperatures. In this work we present details on the samples, new data for other glassy systems and especially the theoretical background that is capable of explaining most of the experimental data. New phenomena are however emerging from these data, for example a deviation from the Curie-like behaviour of the magnetic susceptibility ($M_{intr}/H$ at low fields) due to the glassy structure alone as well as evidence for new and intriguing quantum coherence effects at the lowest temperatures.

[4]  arXiv:2111.12687 [pdf, other]
Title: Many-body localization transition in a frustrated XY chain
Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Strongly Correlated Electrons (cond-mat.str-el)

We study many-body localization (MBL) transition in a one-dimensional isotropic XY chain with a weak next-nearest-neighbor frustration in a random magnetic field. We perform finite-size exact diagonalization calculations of level-spacing statistics and fractal dimensions to demonstrate the MBL transition with increasing the random field amplitude. An equivalent representation of the model in terms of spinless fermions explains the presence of the delocalized phase by the appearance of an effective non-local interaction between the fermions. This interaction appears due to frustration provided by the next-nearest-neighbor hopping.

Cross-lists for Thu, 25 Nov 21

[5]  arXiv:2111.12143 (cross-list from cs.LG) [pdf, other]
Title: Critical initialization of wide and deep neural networks through partial Jacobians: general theory and applications to LayerNorm
Comments: 28 pages, 8 figures
Subjects: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); High Energy Physics - Theory (hep-th); Machine Learning (stat.ML)

Deep neural networks are notorious for defying theoretical treatment. However, when the number of parameters in each layer tends to infinity the network function is a Gaussian process (GP) and quantitatively predictive description is possible. Gaussian approximation allows to formulate criteria for selecting hyperparameters, such as variances of weights and biases, as well as the learning rate. These criteria rely on the notion of criticality defined for deep neural networks. In this work we describe a new way to diagnose (both theoretically and empirically) this criticality. To that end, we introduce partial Jacobians of a network, defined as derivatives of preactivations in layer $l$ with respect to preactivations in layer $l_0<l$. These quantities are particularly useful when the network architecture involves many different layers. We discuss various properties of the partial Jacobians such as their scaling with depth and relation to the neural tangent kernel (NTK). We derive the recurrence relations for the partial Jacobians and utilize them to analyze criticality of deep MLP networks with (and without) LayerNorm. We find that the normalization layer changes the optimal values of hyperparameters and critical exponents. We argue that LayerNorm is more stable when applied to preactivations, rather than activations due to larger correlation depth.

[6]  arXiv:2111.12524 (cross-list from physics.soc-ph) [pdf, other]
Title: Hidden transition in multiplex networks
Comments: 12 pages, 6 figures
Subjects: Physics and Society (physics.soc-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn)

Weak multiplex percolation generalizes percolation to multi-layer networks, represented as networks with a common set of nodes linked by multiple types (colors) of edges. We report a novel discontinuous phase transition in this problem. This anomalous transition occurs in networks of three or more layers without unconnected nodes, $P(0)=0$. Above a critical value of a control parameter, the removal of a tiny fraction $\Delta$ of nodes or edges triggers a failure cascade which ends either with the total collapse of the network, or a return to stability with the system essentially intact. The discontinuity is not accompanied by any singularity of the giant component, in contrast to the discontinuous hybrid transition which usually appears in such problems. The control parameter is the fraction of nodes in each layer with a single connection, $\Pi=P(1)$. We obtain asymptotic expressions for the collapse time and relaxation time, above and below the critical point $\Pi_c$, respectively. In the limit $\Delta\to0$ the total collapse for $\Pi>\Pi_\text{c}$ takes a time $T \propto 1/(\Pi-\Pi_\text{c})$, while there is an exponential relaxation below $\Pi_\text{c}$ with a relaxation time $\tau \propto 1/[\Pi_\text{c}-\Pi]$.

[7]  arXiv:2111.12568 (cross-list from cond-mat.stat-mech) [pdf, other]
Title: Percolation thresholds of randomly rotating patchy particles on Archimedean lattices
Comments: 13 pages for the main text, 23 pages for the supplemental material
Subjects: Statistical Mechanics (cond-mat.stat-mech); Disordered Systems and Neural Networks (cond-mat.dis-nn); Soft Condensed Matter (cond-mat.soft)

We study the percolation of randomly rotating patchy particles on $11$ Archimedean lattices in two dimensions. Each vertex of the lattice is occupied by a particle, and in each model the patch size and number are monodisperse. When there are more than one patches on the surface of a particle, they are symmetrically decorated. As the proportion $\chi$ of the particle surface covered by the patches increases, the clusters connected by the patches grow and the system percolates at the threshold $\chi_c$. We combine Monte Carlo simulations and the critical polynomial method to give precise estimates of $\chi_c$ for disks with one to six patches and spheres with one to two patches on the $11$ lattices. For one-patch particles, we find that the order of $\chi_c$ values for particles on different lattices is the same as that of threshold values $p_c$ for site percolation on same lattices, which implies that $\chi_c$ for one-patch particles mainly depends on the geometry of lattices. For particles with more patches, symmetry and shapes of particles also become important in determining $\chi_c$. We observe that $\chi_c$ values on different lattices do not follow the same order as those for one-patch particles, and get some rules, e.g., some patchy particle models are equivalent with site percolation on same lattices, and different patchy particles can share the same $\chi_c$ value on a given lattice. By analytically calculating probabilities of different patch-covering structures of a single particle as a function of $\chi$ near $\chi_c$, we provide some understanding of these results, and furthermore, we obtain $\chi_c$ for disks with an arbitrary number of patches on five lattices. For these lattices, when the number of patches increases, $\chi_c$ values for patchy disks repeat in periodic ways.

Replacements for Thu, 25 Nov 21

[8]  arXiv:2105.13889 (replaced) [pdf, other]
Title: Equilibrium and non-Equilibrium regimes in the learning of Restricted Boltzmann Machines
Comments: 12 page, 4 figures. accepted to Neurips 2021 Supplementary Material
Journal-ref: NeurIps 2021
Subjects: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech)
[9]  arXiv:2109.15169 (replaced) [pdf, other]
Title: Variational learning of quantum ground states on spiking neuromorphic hardware
Comments: 14 pages, 7 figures
Subjects: Quantum Physics (quant-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Emerging Technologies (cs.ET); Neural and Evolutionary Computing (cs.NE)
[10]  arXiv:2111.07460 (replaced) [pdf, other]
Title: Effect of disorder on density of states and conductivity in higher order van Hove singularities in two dimensional bands
Comments: 15 pages, 7 figures and 36 references
Subjects: Strongly Correlated Electrons (cond-mat.str-el); Disordered Systems and Neural Networks (cond-mat.dis-nn); Materials Science (cond-mat.mtrl-sci)
[ total of 10 entries: 1-10 ]
[ showing up to 2000 entries per page: fewer | more ]

Disable MathJax (What is MathJax?)

Links to: arXiv, form interface, find, cond-mat, recent, 2111, contact, help  (Access key information)