arXiv subjects

Attila Szabó

Publications and source records attributed to Attila Szabó.

At least 19 recordsLinked to original sources

arXiv2026

TeMFpy: a Python library for converting fermionic mean-field states into tensor networks

We introduce TeMFpy, a Python library for converting fermionic mean-field states to finite or infinite matrix product state (MPS) form. TeMFpy includes new, efficient, and easy-to-understand algorithms for both Slater determinants and Pfaffian states. Together with Gutzwiller projection, these also allow the user to build variational wave functions for various strongly correlated electron systems, such as quantum spin liquids. We present all implemented algorithms in detail and describe how they can be accessed through TeMFpy, including full example workflows. TeMFpy is built on top of TeNPy and, therefore, integrates seamlessly with existing MPS-based algorithms.

cond-mat.str-el↗

arXiv2026

Parafermionic and decoupled multicritical points in a frustrated $\mathbb{Z}_6$ clock chain

We introduce a generalised six-state clock chain that interpolates between the clock and Potts models via a multicritical point described by decoupled Ising and three-state Potts models. We find that this decoupling extends into stable phases that break only $\mathbb{Z}_2$ or $\mathbb{Z}_3$ symmetry. We also use boundary CFT analysis and level spectroscopy to conclusively identify a $\mathbb{Z}_6$ parafermion multicritical point terminating the clock model Luttinger-liquid phase. Our work shows that parafermions emerge far from integrability, even in systems with intertwined Ising and three-state Potts orders.

cond-mat.str-el↗

arXiv2025

Rainbow chains and numerical renormalisation group for accurate chiral conformal spectra

Based on the relationship between reduced and thermal density matrices in conformal field theory (CFT), we show that the entanglement spectrum of a conformal critical chain with exponentially decaying terms consists of conformal towers of the associated chiral CFT, with only weak finite-size effects. Through free-fermion and interacting examples, we show that these entanglement spectra present a reliable method to extract detailed CFT spectra from single wave functions without access to the parent Hamiltonian. We complement our method with a Wilsonian numerical renormalisation group algorithm for solving interacting, exponentially decaying chain Hamiltonians.

cond-mat.str-el↗

arXiv2025

Generalised Burnside and Dixon algorithms for irreducible projective representations

Based on the recently proposed character theory of projective representations of finite groups proposed, we generalise several algorithms for computing character tables and matrices of irreducible linear representations to projective representations. In particular, we present an algorithm based on that of Burnside to compute the characters of all irreducible projective representations of a finite group with a given Schur multiplier, and transpose it to exact integer arithmetic following Dixon's character table algorithm. We also describe an algorithm based on that of Dixon to split a projective representation into irreducible subspaces in floating-point arithmetic, and discuss how it can be used to compute matrices for all projective irreps with a given multiplier. Our algorithms bypass the construction of the representation group of the Schur multiplier, which makes them especially attractive for floating-point computations, where exact values of the multiplier are not necessarily available.

math.RT↗

arXiv2024

Noncoplanar and chiral spin states on the way towards Néel ordering in fullerene Heisenberg models

Using high-accuracy variational Monte Carlo based on group-convolutional neural networks (GCNNs), we obtain the symmetry-resolved low-energy spectrum of the spin-1/2 Heisenberg model on several highly symmetric fullerene geometries, including the famous C$_{60}$ buckminsterfullerene. We argue that as the degree of frustration is lowered in large fullerenes, they display characteristic features of incipient magnetic ordering: correlation functions show high-intensity Bragg peaks consistent with Néel-like ordering, while the low-energy spectrum is organised into a tower of states. Competition with frustration, however, turns the simple Néel order into a noncoplanar one. Remarkably, we find and predict chiral incipient ordering in a large number of fullerene structures.

cond-mat.str-el↗

arXiv2023

Dynamics of the antiferromagnetic spin ice phase in pyrochlore spinels

Motivated by the classical spin-nematic state observed in the breathing pyrochlore spinel LiGa$_{0.95}$In$_{0.05}$Cr$_4$O$_8$, we theoretically discuss spin dynamics in models of spin-lattice coupling in these materials. Semiclassical dynamical simulations successfully recover the key features of inelastic neutron-scattering experiments on LiGa$_{0.95}$In$_{0.05}$Cr$_4$O$_8$: a broad finite-energy peak alongside a continuum of scattering near the (200) wave vector that extends from the elastic line to high energies. To interpret this result, we develop a small-fluctuation theory for the spin-ice-like nematic ground states, analogous to linear spin-wave theory for conventionally ordered magnets, which reproduces the numerical simulation results quantitatively. In particular, we find that the inelastic peak is well explained by collective modes confined to ferromagnetic loops of the underlying nematic order. In addition, we find a sharp, linearly dispersing mode in the dynamical structure factor, which originates in long-wavelength fluctuations of the nematic director: We believe identifying this mode will be an interesting target for future experiments on these materials. We also outline potential future applications of our methods to both pyrochlore spinels and other spin-nematic systems.

cond-mat.str-el↗

arXiv2023

High-accuracy variational Monte Carlo for frustrated magnets with deep neural networks

We show that neural quantum states based on very deep (4--16-layered) neural networks can outperform state-of-the-art variational approaches on highly frustrated quantum magnets, including quantum-spin-liquid candidates. We focus on group convolutional neural networks (GCNNs) that allow us to impose space-group symmetries on our ansätze. We achieve state-of-the-art ground-state energies for the $J_1-J_2$ Heisenberg models on the square and triangular lattices, in both ordered and spin-liquid phases, and discuss ways to access low-lying excited states in nontrivial symmetry sectors. We also compute spin and dimer correlation functions for the quantum paramagnetic phase on the triangular lattice, which do not indicate either conventional or valence-bond ordering.

cond-mat.str-el↗

arXiv2023

Fragmented spin ice and multi-$k$ ordering in rare-earth antiperovskites

We study near-neighbour and dipolar Ising models on a lattice of corner-sharing octahedra. In an extended parameter range of both models, frustration between antiferromagnetism and a spin-ice-like three-in-three-out rule stabilises a Coulomb phase with correlated dipolar and quadrupolar spin textures, both yielding distinctive neutron-scattering signatures. Strong further-neighbour perturbations cause the two components to order independently, resulting in unusual multi-$k$ orders. We propose experimental realisations of our model in rare-earth antiperovskites.

cond-mat.str-el↗

arXiv2022

Adopting Automated Bug Assignment in Practice: A Longitudinal Case Study at Ericsson

The continuous inflow of bug reports is a considerable challenge in large development projects. Inspired by contemporary work on mining software repositories, we designed a prototype bug assignment solution based on machine learning in 2011-2016. The prototype evolved into an internal Ericsson product, TRR, in 2017-2018. TRR's first bug assignment without human intervention happened in April 2019. Our study evaluates the adoption of TRR within its industrial context at Ericsson. Moreover, we investigate 1) how TRR performs in the field, 2) what value TRR provides to Ericsson, and 3) how TRR has influenced the ways of working. We conduct an industrial case study combining interviews with TRR stakeholders, minutes from sprint planning meetings, and bug tracking data. The data analysis includes thematic analysis, descriptive statistics, and Bayesian causal analysis. TRR is now an incorporated part of the bug assignment process. Considering the abstraction levels of the telecommunications stack, high-level modules are more positive while low-level modules experienced some drawbacks. On average, TRR automatically assigns 30% of the incoming bug reports with an accuracy of 75%. Auto-routed TRs are resolved around 21% faster within Ericsson, and TRR has saved highly seasoned engineers many hours of work. Indirect effects of adopting TRR include process improvements, process awareness, increased communication, and higher job satisfaction. TRR has saved time at Ericsson, but the adoption of automated bug assignment was more intricate compared to similar endeavors reported from other companies. We primarily attribute the difference to the very large size of the organization and the complex products. Key facilitators in the successful adoption include a gradual introduction, product champions, and careful stakeholder analysis.

cs.SE↗

arXiv2022

NetKet 3: Machine Learning Toolbox for Many-Body Quantum Systems

We introduce version 3 of NetKet, the machine learning toolbox for many-body quantum physics. NetKet is built around neural-network quantum states and provides efficient algorithms for their evaluation and optimization. This new version is built on top of JAX, a differentiable programming and accelerated linear algebra framework for the Python programming language. The most significant new feature is the possibility to define arbitrary neural network ansätze in pure Python code using the concise notation of machine-learning frameworks, which allows for just-in-time compilation as well as the implicit generation of gradients thanks to automatic differentiation. NetKet 3 also comes with support for GPU and TPU accelerators, advanced support for discrete symmetry groups, chunking to scale up to thousands of degrees of freedom, drivers for quantum dynamics applications, and improved modularity, allowing users to use only parts of the toolbox as a foundation for their own code.

quant-ph↗

arXiv2021

Learning to Deblur and Rotate Motion-Blurred Faces

We propose a solution to the novel task of rendering sharp videos from new viewpoints from a single motion-blurred image of a face. Our method handles the complexity of face blur by implicitly learning the geometry and motion of faces through the joint training on three large datasets: FFHQ and 300VW, which are publicly available, and a new Bern Multi-View Face Dataset (BMFD) that we built. The first two datasets provide a large variety of faces and allow our model to generalize better. BMFD instead allows us to introduce multi-view constraints, which are crucial to synthesizing sharp videos from a new camera view. It consists of high frame rate synchronized videos from multiple views of several subjects displaying a wide range of facial expressions. We use the high frame rate videos to simulate realistic motion blur through averaging. Thanks to this dataset, we train a neural network to reconstruct a 3D video representation from a single image and the corresponding face gaze. We then provide a camera viewpoint relative to the estimated gaze and the blurry image as input to an encoder-decoder network to generate a video of sharp frames with a novel camera viewpoint. We demonstrate our approach on test subjects of our multi-view dataset and VIDTIMIT.

cs.CV↗

arXiv2021

High-temperature transport and polaron speciation in the anharmonic Holstein model

We study the finite-temperature transport of electrons coupled to anharmonic local phonons. Our focus is on the high-temperature incoherent regime, where controlled calculations are possible both for weak and strong electron--phonon coupling. At strong coupling, the dynamics is described in terms of a multiple-species gas of small polarons: this emergent `speciation' is driven by energy conservation. We explicitly compute the dc and ac response in this regime. We discuss the breakdown of the polaron picture, and the onset of localization, in the limit where phonons become quasistatic.

cond-mat.str-el↗

arXiv2020

Neural network wave functions and the sign problem

Neural quantum states (NQS) are a promising approach to study many-body quantum physics. However, they face a major challenge when applied to lattice models: Convolutional networks struggle to converge to ground states with a nontrivial sign structure. We tackle this problem by proposing a neural network architecture with a simple, explicit, and interpretable phase ansatz, which can robustly represent such states and achieve state-of-the-art variational energies for both conventional and frustrated antiferromagnets. In the latter case, our approach uncovers low-energy states that exhibit the Marshall sign rule and are therefore inconsistent with the expected ground state. Such states are the likely cause of the obstruction for NQS-based variational Monte Carlo to access the true ground states of these systems. We discuss the implications of this observation and suggest potential strategies to overcome the problem.

cond-mat.str-el↗

arXiv2020

Mixed spectra and partially extended states in a two-dimensional quasiperiodic model

We introduce a two-dimensional generalisation of the quasiperiodic Aubry-André model. Even though this model exhibits the same duality relation as the one-dimensional version, its localisation properties are found to be substantially more complex. In particular, partially extended single-particle states appear for arbitrarily strong quasiperiodic modulation. They are concentrated on a network of low-disorder lattice lines, while the rest of the lattice hosts localised states. This spatial separation protects the localised states from delocalisation, so no mobility edge emerges in the spectrum. Instead, localised and partially extended states are interspersed, giving rise to an unusual type of mixed spectrum and enabling complex dynamics even in the absence of interactions. A striking example is ballistic transport across the low-disorder lines while the rest of the system remains localised. This behaviour is robust against disorder and other weak perturbations. Our model is thus directly amenable to experimental studies and promises fascinating many-body localisation properties.

cond-mat.dis-nn↗

arXiv2019

Unsupervised Generative 3D Shape Learning from Natural Images

In this paper we present, to the best of our knowledge, the first method to learn a generative model of 3D shapes from natural images in a fully unsupervised way. For example, we do not use any ground truth 3D or 2D annotations, stereo video, and ego-motion during the training. Our approach follows the general strategy of Generative Adversarial Networks, where an image generator network learns to create image samples that are realistic enough to fool a discriminator network into believing that they are natural images. In contrast, in our approach the image generation is split into 2 stages. In the first stage a generator network outputs 3D objects. In the second, a differentiable renderer produces an image of the 3D objects from random viewpoints. The key observation is that a realistic 3D object should yield a realistic rendering from any plausible viewpoint. Thus, by randomizing the choice of the viewpoint our proposed training forces the generator network to learn an interpretable 3D representation disentangled from the viewpoint. In this work, a 3D representation consists of a triangle mesh and a texture map that is used to color the triangle surface by using the UV-mapping technique. We provide analysis of our learning approach, expose its ambiguities and show how to overcome them. Experimentally, we demonstrate that our method can learn realistic 3D shapes of faces by using only the natural images of the FFHQ dataset.

cs.CV↗

arXiv2019

Seeing beyond the light: Vison and photon electrodynamics in quantum spin ice

Understanding the nature and behaviour of excitations in quantum spin liquids, and in topological phases of matter in general, is of fundamental importance and has proven crucial for experimental detection and characterisation of candidate materials. Current theoretical and numerical techniques, however, have limited capabilities, especially when it comes to studying gapped excitations. Here, we propose a semiclassical numerical method to study systems whose spin liquid behaviour is underpinned by perturbative ring-exchange Hamiltonians. Our method can readily access both thermodynamic and spectral properties. We focus in particular on quantum spin ice and its photon and vison excitations. After benchmarking the method against existing results on photons, we use it to characterise visons and their thermodynamic behaviour, which remained hitherto largely unexplored. We find that visons, in contrast to spinons in classical spin ice, form a weak electrolyte: vison pairs are the dominant population at low temperatures. This is reflected in the behaviour of thermodynamic quantities, such as pinch point motifs in the relevant correlators. Visons also appear to strongly hybridise with the photon background, a phenomenon that affects the way these quasiparticles may show up in inelastic response measurements. Our results demonstrate that the method, and generalisations thereof, can substantially help our understanding of quasiparticles and their interplay in quantum spin ice and other quantum spin liquids, quantum dimer models, and lattice gauge theories in general.

cond-mat.str-el↗

arXiv2019

Generalised route to effective field theories for quantum systems with local constraints

Some of the exciting phenomena uncovered in strongly correlated systems in recent years - for instance quantum topological order, deconfined quantum criticality, and emergent gauge symmetries -- appear in systems in which the Hilbert space is effectively projected at low energies in a way that imposes local constraints on the original degrees of freedom. Cases in point include spin liquids, valence bond systems, dimer models, and vertex models. In this work, we use a slave boson description coupled to a large-$S$ path integral formulation to devise a generalised route to obtain effective field theories for such systems. We demonstrate the validity and capability of our approach by studying quantum dimer models and by comparing our results with the existing literature. Field-theoretic approaches to date are limited to bipartite lattices, they depend on a gauge-symmetric understanding of the constraint, and they lack generic quantitative predictive power for the coefficients of the terms that appear in the Lagrangians of these systems. Our method overcomes all these shortcomings and we show how the results up to quadratic order compare with the known height description of the square lattice quantum dimer model, as well as with the numerical estimate of the speed of light of the photon excitations on the diamond lattice. Finally, instanton considerations allow us to infer properties of the finite-temperature behaviour in two dimensions.

cond-mat.str-el↗

arXiv2018

Unsupervised 3D Shape Learning from Image Collections in the Wild

We present a method to learn the 3D surface of objects directly from a collection of images. Previous work achieved this capability by exploiting additional manual annotation, such as object pose, 3D surface templates, temporal continuity of videos, manually selected landmarks, and foreground/background masks. In contrast, our method does not make use of any such annotation. Rather, it builds a generative model, a convolutional neural network, which, given a noise vector sample, outputs the 3D surface and texture of an object and a background image. These 3 components combined with an additional random viewpoint vector are then fed to a differential renderer to produce a view of the sampled object and background. Our general principle is that if the output of the renderer, the generated image, is realistic, then its input, the generated 3D and texture, should also be realistic. To achieve realism, the generative model is trained adversarially against a discriminator that tries to distinguish between the output of the renderer and real images from the given data set. Moreover, our generative model can be paired with an encoder and trained as an autoencoder, to automatically extract the 3D shape, texture and pose of the object in an image. Our trained generative model and encoder show promising results both on real and synthetic data, which demonstrate for the first time that fully unsupervised 3D learning from image collections is possible.

cs.CV↗