Disordered Systems and Neural Networks
See recent articles
Showing new listings for Monday, 18 November 2024
- [1] arXiv:2411.10247 [pdf, html, other]
-
Title: Anderson localization induced by structural disorderComments: 8 pages, 6 figuresSubjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Quantum Physics (quant-ph)
We examine the onset of Anderson localization in three-dimensional systems with structural disorder in the form of lattice irregularities and in the absence of any on-site disordered potential. Analyzing two models with distinct types of lattice regularities, we show that the Anderson localization transition occurs when the strength of the structural disorder is smoothly increased. Performing finite-size scaling analysis of the results, we show that the transition belongs to the same universality class as regular Anderson localization induced by onsite disorder. Our work identifies a new class of structurally disordered lattice models in which destructive interference of matter waves may inhibit transport and lead to a transition between metallic and localized phases.
- [2] arXiv:2411.10277 [pdf, html, other]
-
Title: Bound states in one-dimensional systems with colored noiseComments: 10 pages, 12 figures, To be published in Physical Review BSubjects: Disordered Systems and Neural Networks (cond-mat.dis-nn)
We investigate the phase transitions in a one-dimensional system with colored noise. Previous studies indicated that the phase diagram of this system included extended and disorder-induced localized phases. However, by studying the properties of wave functions, we find that this phase diagram can be further refined, revealing the existence of a bound phase for the large potential amplitude $W$ and noise control parameter $\alpha$. In the bound phase, the wave function cannot extend throughout the entire chain, tails decay faster than exponentially and its distribution expands as the system size increases. By adjusting the potential amplitude to induce a transition from the extended phase to the bound phase, we find that bound states coexist with extended states in the spectrum. In contrast, when the system transitions from the Anderson localized phase to the bound phase, we do not observe the obvious coexistence of Anderson localized and bound states. Finally, by performing the time evolution, we find that the dynamic transition point of the bound phase is inconsistent with the static one for large $\alpha$.
- [3] arXiv:2411.10445 [pdf, html, other]
-
Title: Inverse Melting of Polar Order in a Ferroelectric OxideSubjects: Disordered Systems and Neural Networks (cond-mat.dis-nn)
In many condensed matter systems, long range order emerges at low temperatures as thermal fluctuations subside. In the presence of competing interactions or quenched disorder, however, some systems can show unusual configurations that become more disordered at low temperature, a rare phenomenon known as "inverse melting". Here, we discover an inverse melting of the polar order in a ferroelectric oxide with quenched chemical disorder (BaTi1-xZrxO3) through direct atomic-scale visualization using in situ scanning transmission electron microscopy. In contrast to the clean BaTiO3 parent system in which long range order tracks lower temperatures, we observe in the doped system BaTi1-xZrxO3 that thermally driven fluctuations at high temperature give way to a more ordered state and then to a re-entrant disordered configuration at even lower temperature. Such an inverse melting of the polar order is likely linked to the random field generated by Zr dopants, which modulates the energy landscape arising from the competition between thermal fluctuations and random field pinning potential. These visualizations highlight a rich landscape of order and disorder in materials with quenched disorder, which may be key to understanding their advanced functionalities.
New submissions (showing 3 of 3 entries)
- [4] arXiv:2411.09754 (cross-list from cond-mat.mes-hall) [pdf, html, other]
-
Title: Coherent Magneto-Conductance Oscillations in Amorphous Topological Insulator NanowiresSubjects: Mesoscale and Nanoscale Physics (cond-mat.mes-hall); Disordered Systems and Neural Networks (cond-mat.dis-nn)
Recent experiments on amorphous materials have established the existence of surface states similar to those of crystalline three-dimensional topological insulators (TIs). Amorphous topological insulators are also independently of interest for thermo-electric and other properties. To develop an understanding of transport in these systems, we carry out quantum transport calculations for a tight-binding model of an amorphous nano-wire pierced by an axial magnetic flux, then compare the results to known features in the case of crystalline models with disorder. Our calculations complement previous studies in the crystalline case that studied the surface or used a Green's function method. We find that the periodicity of the conductance signal with varying magnetic flux is comparable to the crystalline case, with maxima occurring at odd multiples of magnetic flux quanta. However, the expected amplitude of the oscillation decreases with increasing amorphousness, as defined and described in the main text. We characterize this deviation from the crystalline case by taking ensemble averages of the conductance signatures for various wires with measurements simulated at finite temperatures. This striking transport phenomenon offers a metric to characterize amorphous TIs and stimulate further experiments on this class of materials.
- [5] arXiv:2411.09784 (cross-list from quant-ph) [pdf, html, other]
-
Title: Reinforced Disentanglers on Random Unitary CircuitsComments: 9 pages, 7 figures, 1 table. Submitted to QIP 2025Subjects: Quantum Physics (quant-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Machine Learning (cs.LG)
We search for efficient disentanglers on random Clifford circuits of two-qubit gates arranged in a brick-wall pattern, using the proximal policy optimization (PPO) algorithm \cite{schulman2017proximalpolicyoptimizationalgorithms}. Disentanglers are defined as a set of projective measurements inserted between consecutive entangling layers. An efficient disentangler is a set of projective measurements that minimize the averaged von Neumann entropy of the final state with the least number of total projections possible. The problem is naturally amenable to reinforcement learning techniques by taking the binary matrix representing the projective measurements along the circuit as our state, and actions as bit flipping operations on this binary matrix that add or delete measurements at specified locations. We give rewards to our agent dependent on the averaged von Neumann entropy of the final state and the configuration of measurements, such that the agent learns the optimal policy that will take him from the initial state of no measurements to the optimal measurement state that minimizes the entanglement entropy. Our results indicate that the number of measurements required to disentangle a random quantum circuit is drastically less than the numerical results of measurement-induced phase transition papers. Additionally, the reinforcement learning procedure enables us to characterize the pattern of optimal disentanglers, which is not possible in the works of measurement-induced phase transitions.
- [6] arXiv:2411.10355 (cross-list from math-ph) [pdf, html, other]
-
Title: Transmission eigenvalue distribution in disordered media from anisotropic field theoryComments: 25 pages, 7 figuresSubjects: Mathematical Physics (math-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Optics (physics.optics)
A field theory for the distribution of transmission eigenvalues through a disordered medium is developed in details. The central equation of this theory is a self-consistent transport equation for a $2\times 2$ matrix radiance. This equation determines the transmission eigenvalue distribution not only in the diffusive regime but also in the quasiballistic regime, where the radiance is expected to be significantly anisotropic in the material bulk. We show that the matrix transport equation can be solved analytically in this regime. We also show that, in addition to the finite-width waveguide, our equation is able to predict the transmission eigenvalue distribution through an infinite slab.
- [7] arXiv:2411.10360 (cross-list from math-ph) [pdf, html, other]
-
Title: Anisotropic Field Theory of Wave Transmission Statistics in Disordered MediaComments: 6 pages, 3 figuresSubjects: Mathematical Physics (math-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Optics (physics.optics)
We present a field theory to characterize the mean distribution of transmission eigenvalues for coherent wave propagation through a disordered medium. Unlike the Dorokhov-Mello-Pereyra-Kumar (DMPK) theory, our approach does not rely on the isotropy hypothesis, which assumes uniform field intensity across scattering channels. As a result, it accurately predicts the transmission eigenvalue distribution both in the diffusive and quasiballistic regime. In addition, it proves to be more versatile than the DMPK theory, allowing for the incorporation of physical effects such as absorption or partial channel control with minimal adjustments to the formalism. Although these effects are commonly encountered in experiments on complex media, until now there has been no ab initio theory addressing them. Our predictions are numerically validated against transmission eigenvalue distributions computed from the microscopic wave equation.
- [8] arXiv:2411.10376 (cross-list from physics.flu-dyn) [pdf, html, other]
-
Title: Soliton Dynamics over a Disordered TopographyComments: in press in Physical Review Letters (2024)Journal-ref: Physical Review Letters (2024)Subjects: Fluid Dynamics (physics.flu-dyn); Disordered Systems and Neural Networks (cond-mat.dis-nn); Pattern Formation and Solitons (nlin.PS)
We report on the dynamics of a soliton propagating on the surface of a fluid in a 4-m-long canal with a random or periodic bottom topography. Using a full space-and-time resolved wavefield measurement, we evidence, for the first time experimentally, how the soliton is affected by the disorder, in the context of Anderson localization, and how localization depends on nonlinearity. For weak soliton amplitudes, the localization length is found in quantitative agreement with a linear shallow-water theory. For higher amplitudes, this spatial attenuation of the soliton amplitude is found to be enhanced. Behind the leading soliton slowed down by the topography, different experimentally unreported dynamics occur: Fission into backward and forward nondispersive pulses for the periodic case, and scattering into dispersive waves for the random case. Our findings open doors to potential applications regarding ocean coastal protection against large-amplitude waves.
- [9] arXiv:2411.10430 (cross-list from cond-mat.str-el) [pdf, html, other]
-
Title: Simulating the two-dimensional $t-J$ model at finite doping with neural quantum statesSubjects: Strongly Correlated Electrons (cond-mat.str-el); Disordered Systems and Neural Networks (cond-mat.dis-nn); Quantum Gases (cond-mat.quant-gas)
Simulating large, strongly interacting fermionic systems remains a major challenge for existing numerical methods. In this work, we present, for the first time, the application of neural quantum states - specifically, hidden fermion determinant states (HFDS) - to simulate the strongly interacting limit of the Fermi-Hubbard model, namely the $t-J$ model, across the entire doping regime. We demonstrate that HFDS achieve energies competitive with matrix product states (MPS) on lattices as large as $6 \times 10$ sites while using several orders of magnitude fewer parameters, suggesting the potential for efficient application to even larger system sizes. This remarkable efficiency enables us to probe low-energy physics across the full doping range, providing new insights into the competition between kinetic and magnetic interactions and the nature of emergent quasiparticles. Starting from the low-doping regime, where magnetic polarons dominate the low energy physics, we track their evolution with increasing doping through analyses of spin and polaron correlation functions. Our findings demonstrate the potential of determinant-based neural quantum states with inherent fermionic sign structure, opening the way for simulating large-scale fermionic systems at any particle filling.
Cross submissions (showing 6 of 6 entries)
- [10] arXiv:2311.09450 (replaced) [pdf, html, other]
-
Title: Unifying view of fermionic neural network quantum states: From neural network backflow to hidden fermion determinant statesComments: 19 pages, 11 figuresJournal-ref: Phys. Rev. B 110, 115124 (2024)Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Quantum Physics (quant-ph)
Among the variational wave functions for Fermionic Hamiltonians, neural network backflow (NNBF) and hidden fermion determinant states (HFDS) are two prominent classes to provide accurate approximations to the ground state. Here we develop a unifying view of fermionic neural quantum states casting them all in the framework of NNBF. NNBF wave-functions have configuration-dependent single-particle orbitals (SPO) which are parameterized by a neural network. We show that HFDS with $r$ hidden fermions can be written as a NNBF with an $r \times r$ determinant Jastrow and a restricted low-rank $r$ additive correction to the SPO. Furthermore, we show that in NNBF wave-functions, such determinant Jastrow's can generically be removed at the cost of further complicating the additive SPO correction increasing its rank by $r$. We numerically and analytically compare additive SPO corrections generated by the product of two matrices with inner dimension $r$. We find that larger $r$ wave-functions span a larger space and give evidence that simpler and more direct updates to the SPO's tend to be more expressive and better energetically. These suggest the standard NNBF approach is preferred amongst other related choices. Finally, we uncover that the row-selection used to select single-particle orbitals allows significant sign and amplitude modulation between nearby configurations and is partially responsible for the quality of NNBF and HFDS wave-functions.
- [11] arXiv:2301.04514 (replaced) [pdf, html, other]
-
Title: Multi-scale Laplacian community detection in heterogeneous networksComments: 19 pages, 14 figuresSubjects: Physics and Society (physics.soc-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Biological Physics (physics.bio-ph)
Heterogeneous and complex networks represent intertwined interactions between real-world elements or agents. Determining the multi-scale mesoscopic organization of clusters and intertwined structures is still a fundamental and open problem of complex network theory. By taking advantage of the recent Laplacian Renormalization Group, we scrutinize information diffusion pathways throughout networks to shed further light on this issue. Based on inter-node communicability, our definition provides a clear-cut framework for resolving the multi-scale mesh of structures in complex networks, disentangling their intrinsic arboreal architecture. As it does not consider any topological null-model assumption, the LRG naturally permits the introduction of scale-dependent optimal partitions. Moreover, we demonstrate the existence of a particular class of nodes, called 'metastable' nodes, that switching regions to which they belong at different scales, likely playing a pivotal role in cross-regional communication and, therefore, in managing macroscopic effects of the whole network.
- [12] arXiv:2310.12535 (replaced) [pdf, html, other]
-
Title: Scalable molecular simulation of electrolyte solutions with quantum chemical accuracySubjects: Chemical Physics (physics.chem-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Soft Condensed Matter (cond-mat.soft); Statistical Mechanics (cond-mat.stat-mech); Computational Physics (physics.comp-ph)
Electrolyte solutions play critical role in a vast range of important applications, yet an accurate and scalable method of predicting their properties without fitting to experiment has remained out of reach, despite over a century of effort. Here, we combine state-of-the-art density functional theory and equivariant neural network potentials to demonstrate this capability, reproducing key structural, thermodynamic, and kinetic properties. We show that neural network potentials (NNPs) can be recursively trained on a subset of their own output to enable coarse-grained/continuum-solvent molecular simulations that can access much longer timescales than possible with all atom simulations. We observe the surprising formation of Li cation dimers along with identical anion-anion pairing of chloride and bromide anions. Finally, we reproduce simulate the crystal phase and infinite dilution pairing free energies despite being trained only on moderate concentration solutions. This approach should be scaled to build a greatly expanded database of electrolyte solution properties than currently exists.
- [13] arXiv:2406.02269 (replaced) [pdf, html, other]
-
Title: Graph Neural Networks Do Not Always OversmoothSubjects: Machine Learning (stat.ML); Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG)
Graph neural networks (GNNs) have emerged as powerful tools for processing relational data in applications. However, GNNs suffer from the problem of oversmoothing, the property that the features of all nodes exponentially converge to the same vector over layers, prohibiting the design of deep GNNs. In this work we study oversmoothing in graph convolutional networks (GCNs) by using their Gaussian process (GP) equivalence in the limit of infinitely many hidden features. By generalizing methods from conventional deep neural networks (DNNs), we can describe the distribution of features at the output layer of deep GCNs in terms of a GP: as expected, we find that typical parameter choices from the literature lead to oversmoothing. The theory, however, allows us to identify a new, non-oversmoothing phase: if the initial weights of the network have sufficiently large variance, GCNs do not oversmooth, and node features remain informative even at large depth. We demonstrate the validity of this prediction in finite-size GCNs by training a linear classifier on their output. Moreover, using the linearization of the GCN GP, we generalize the concept of propagation depth of information from DNNs to GCNs. This propagation depth diverges at the transition between the oversmoothing and non-oversmoothing phase. We test the predictions of our approach and find good agreement with finite-size GCNs. Initializing GCNs near the transition to the non-oversmoothing phase, we obtain networks which are both deep and expressive.