Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2305.04035 · Physical review. B./Physical review. B

Supervised learning of an interacting two-dimensional hardcore boson model of a weak topological insulator using correlation functions

Amrita Ghosh, Mugdha Sarkar

Published Oct 15, 2024• 3 citations
arXivHTMLPDF

Abstract

We study a system of hard-core bosons on a two-dimensional periodic honeycomb lattice subjected to an on-site potential with alternating signs along $y$ direction, using machine learning (ML) techniques. The model hosts a rich phase diagram consisting of six different phases including a charge density wave, a superfluid phase and two dimer insulator phases, one of which is also a weak topological insulator with zero Chern number but a nontrivial Berry phase [Ghosh et al., SciPost Phys. 10, 059 (2021)]. Using two distinct correlation functions computed via quantum Monte Carlo method, a relatively simple ML model is able to learn information from the various phases simultaneously and accurately predict their phase boundaries. By employing our ML model trained on the data set obtained from the Hamiltonian without explicit interactions, we determine the phase structure of the system in the presence of nearest-neighbor interactions. Additionally, we investigate the robustness of the weak topological insulator phase against interactions by predicting the topological invariant, which is otherwise difficult to obtain.

DOI
10.1103/physrevb.110.165134

Related papers

Weak topological insulating phases of hard-core-bosons on the honeycomb lattice

Amrita Ghosh, Eytan Grosfeld

We study the phases of hard-core bosons on a two-dimensional periodic honeycomb lattice in the presence of an on-site potential with alternating sign along the different y-layers of the lattice. Using quantum Monte Carlo simulations supported by analytical calculations, we identify a weak topological insulator, characterized by a zero Chern number but non-zero Berry phase, which is manifested at either density 1/4 or 3/4, as determined by the potential pattern. Additionally, a charge-density-wave insulator is observed at 1/2-filling, whereas the phase diagram at intermediate densities is occupied by a superfluid phase. The weak topological insulator is further shown to be robust against any amount of nearest-neighbor repulsion, as well as weak next-nearest-neighbor repulsion. The experimental realization of our model is feasible in an optical lattice setup.

arXiv:2010.16126•SciPost Physics•Mar 9, 2021•9 citations

Topological bosonic states on ribbons of a honeycomb lattice

Yi‐Ping Wang, Xingchuan Zhu, Kefei Zou +2 more

The topological properties of hardcore bosons on ribbons of honeycomb lattice are studied using quantum Monte Carlo simulations. We map out a rich phase diagram with the superfluid and insulator phases at various fillings. Particularly, it is revealed that the insulator state at half filling is a topological bosonic state, which is characterized by a nontrivial Berry phase and a pair of end states. We provide an intuitive picture to understand this topological bosonic insulator state by showing that it can be adiabatically mapped to a topological fermionic model. It is also shown that the topological end states are robust against weak interactions beyond the hardcore repulsion. Our results can be simulated by using bosonic cold atoms trapped in designed optical lattices.

arXiv:1808.03025•Physical Review A•Oct 15, 2018•3 citations

Chiral bosonic topological insulator on the honeycomb lattice with anisotropic interactions

Amrita Ghosh, Eytan Grosfeld

We study hard-core bosons on the honeycomb lattice in the presence of anisotropic nearest-neighbor repulsive interactions. Using a quantum Monte Carlo (QMC) technique, we extract the phase diagram of the model in terms of the filling and the anisotropy. At half-filling we find a dimer insulator phase near maximum anisotropy that is characterized by a finite topological entanglement entropy $ln(2)/2$, indicative of a fractional quantum Hall state for bosons. We identify the presence of edge states and derive a QMC-based method to extract and verify their chirality. Remarkably, this phase arises in the absence of magnetic flux and without explicit lattice frustration.

arXiv:2012.11461•Physical review. B./Physical review. B•May 10, 2021•2 citations

Unsupervised and supervised learning of interacting topological phases from single-particle correlation functions

Simone Tibaldi, Giuseppe Magnifico, Davide Vodola +1 more

The recent advances in machine learning algorithms have boosted the application of these techniques to the field of condensed matter physics, in order e.g. to classify the phases of matter at equilibrium or to predict the real-time dynamics of a large class of physical models. Typically in these works, a machine learning algorithm is trained and tested on data coming from the same physical model. Here we demonstrate that unsupervised and supervised machine learning techniques are able to predict phases of a non-exactly solvable model when trained on data of a solvable model. In particular, we employ a training set made by single-particle correlation functions of a non-interacting quantum wire and by using principal component analysis, k-means clustering, t-distributed stochastic neighbor embedding and convolutional neural networks we reconstruct the phase diagram of an interacting superconductor. We show that both the principal component analysis and the convolutional neural networks trained on the data of the non-interacting model can identify the topological phases of the interacting model. Our findings indicate that non-trivial phases of matter emerging from the presence of interactions can be identified by means of unsupervised and supervised techniques applied to data of non-interacting systems.

arXiv:2202.09281•SciPost Physics•Jan 19, 2023•31 citations

Strong to weak topological interacting phase transition of bosons on a lattice

Amrita Ghosh, Eytan Grosfeld

We study hard-core bosons on the honeycomb lattice subjected to anisotropic nearest-neighbor hopping along with anisotropic nearest-neighbor repulsion, using a quantum Monte Carlo technique. At half filling, we find a transition from strong topological interacting order to weak topological interacting order as a function of the hopping anisotropy. The strong topological phase is characterized by a finite topological entanglement entropy, while the weak topological order is identified with a nontrivial value of the bipartite entanglement entropy. Some of the order parameters and their derivatives demonstrate abrupt changes when varying the parameters controlling the lattice anisotropies, thus revealing the nature of this topological interacting phase transition.

arXiv:2203.07177•Physical review. B./Physical review. B•Jul 13, 2023•2 citations

Quantum phase recognition via unsupervised machine learning

Peter Broecker, Fakher F. Assaad, Simon Trebst

The application of state-of-the-art machine learning techniques to statistical physic problems has seen a surge of interest for their ability to discriminate phases of matter by extracting essential features in the many-body wavefunction or the ensemble of correlators sampled in Monte Carlo simulations. Here we introduce a gener- alization of supervised machine learning approaches that allows to accurately map out phase diagrams of inter- acting many-body systems without any prior knowledge, e.g. of their general topology or the number of distinct phases. To substantiate the versatility of this approach, which combines convolutional neural networks with quantum Monte Carlo sampling, we map out the phase diagrams of interacting boson and fermion models both at zero and finite temperatures and show that first-order, second-order, and Kosterlitz-Thouless phase transitions can all be identified. We explicitly demonstrate that our approach is capable of identifying the phase transition to non-trivial many-body phases such as superfluids or topologically ordered phases without supervision.

arXiv:1707.00663•Jul 3, 2017•101 citations
We study a system of hard-core bosons on a two-dimensional periodic honeycomb lattice subjected to an on-site potential with alternating signs along yyy direction, using machine learning (ML) techniques. The model hosts a rich phase diagram consisting of six different phases including a charge density wave, a superfluid phase and two dimer insulator phases, one of which is also a weak topological insulator with zero Chern number but a nontrivial Berry phase [Ghosh et al., SciPost Phys. 10, 059 (2021)]. Using two distinct correlation functions computed via quantum Monte Carlo method, a relatively simple ML model is able to learn information from the various phases simultaneously and accurately predict their phase boundaries. By employing our ML model trained on the data set obtained from the Hamiltonian without explicit interactions, we determine the phase structure of the system in the presence of nearest-neighbor interactions. Additionally, we investigate the robustness of the weak topological insulator phase against interactions by predicting the topological invariant, which is otherwise difficult to obtain.
We study the phases of hard-core bosons on a two-dimensional periodic honeycomb lattice in the presence of an on-site potential with alternating sign along the different y-layers of the lattice. Using quantum Monte Carlo simulations supported by analytical calculations, we identify a weak topological insulator, characterized by a zero Chern number but non-zero Berry phase, which is manifested at either density 1/4 or 3/4, as determined by the potential pattern. Additionally, a charge-density-wave insulator is observed at 1/2-filling, whereas the phase diagram at intermediate densities is occupied by a superfluid phase. The weak topological insulator is further shown to be robust against any amount of nearest-neighbor repulsion, as well as weak next-nearest-neighbor repulsion. The experimental realization of our model is feasible in an optical lattice setup.
The topological properties of hardcore bosons on ribbons of honeycomb lattice are studied using quantum Monte Carlo simulations. We map out a rich phase diagram with the superfluid and insulator phases at various fillings. Particularly, it is revealed that the insulator state at half filling is a topological bosonic state, which is characterized by a nontrivial Berry phase and a pair of end states. We provide an intuitive picture to understand this topological bosonic insulator state by showing that it can be adiabatically mapped to a topological fermionic model. It is also shown that the topological end states are robust against weak interactions beyond the hardcore repulsion. Our results can be simulated by using bosonic cold atoms trapped in designed optical lattices.
We study hard-core bosons on the honeycomb lattice in the presence of anisotropic nearest-neighbor repulsive interactions. Using a quantum Monte Carlo (QMC) technique, we extract the phase diagram of the model in terms of the filling and the anisotropy. At half-filling we find a dimer insulator phase near maximum anisotropy that is characterized by a finite topological entanglement entropy ln(2)/2ln(2)/2ln(2)/2, indicative of a fractional quantum Hall state for bosons. We identify the presence of edge states and derive a QMC-based method to extract and verify their chirality. Remarkably, this phase arises in the absence of magnetic flux and without explicit lattice frustration.
The recent advances in machine learning algorithms have boosted the application of these techniques to the field of condensed matter physics, in order e.g. to classify the phases of matter at equilibrium or to predict the real-time dynamics of a large class of physical models. Typically in these works, a machine learning algorithm is trained and tested on data coming from the same physical model. Here we demonstrate that unsupervised and supervised machine learning techniques are able to predict phases of a non-exactly solvable model when trained on data of a solvable model. In particular, we employ a training set made by single-particle correlation functions of a non-interacting quantum wire and by using principal component analysis, k-means clustering, t-distributed stochastic neighbor embedding and convolutional neural networks we reconstruct the phase diagram of an interacting superconductor. We show that both the principal component analysis and the convolutional neural networks trained on the data of the non-interacting model can identify the topological phases of the interacting model. Our findings indicate that non-trivial phases of matter emerging from the presence of interactions can be identified by means of unsupervised and supervised techniques applied to data of non-interacting systems.
We study hard-core bosons on the honeycomb lattice subjected to anisotropic nearest-neighbor hopping along with anisotropic nearest-neighbor repulsion, using a quantum Monte Carlo technique. At half filling, we find a transition from strong topological interacting order to weak topological interacting order as a function of the hopping anisotropy. The strong topological phase is characterized by a finite topological entanglement entropy, while the weak topological order is identified with a nontrivial value of the bipartite entanglement entropy. Some of the order parameters and their derivatives demonstrate abrupt changes when varying the parameters controlling the lattice anisotropies, thus revealing the nature of this topological interacting phase transition.
The application of state-of-the-art machine learning techniques to statistical physic problems has seen a surge of interest for their ability to discriminate phases of matter by extracting essential features in the many-body wavefunction or the ensemble of correlators sampled in Monte Carlo simulations. Here we introduce a gener- alization of supervised machine learning approaches that allows to accurately map out phase diagrams of inter- acting many-body systems without any prior knowledge, e.g. of their general topology or the number of distinct phases. To substantiate the versatility of this approach, which combines convolutional neural networks with quantum Monte Carlo sampling, we map out the phase diagrams of interacting boson and fermion models both at zero and finite temperatures and show that first-order, second-order, and Kosterlitz-Thouless phase transitions can all be identified. We explicitly demonstrate that our approach is capable of identifying the phase transition to non-trivial many-body phases such as superfluids or topologically ordered phases without supervision.