1 of 29

Large-scale simulation of�deep neural quantum states

Ao Chen

Supervisor: Markus Heyl

28.04.2025, Augsburg

2 of 29

Agenda

  • Why neural quantum states (NQS)?

  • From shallow to deep

  • NQS for quantum spin liquids

  • NQS for Fermi-Hubbard model

3 of 29

Why neural quantum states?

4 of 29

Image recognition

Input image

 

Model

0.98

0.12

5 of 29

Game of Go

Game status

 

Model

0.12

6 of 29

Quantum many-body problem

 

 

Model

 

 

 

 

7 of 29

From shallow to deep

8 of 29

One layer

G. Carleo & M. Troyer, Science (2017)

 

Restricted Boltzmann machine (RBM)

9 of 29

Deeper networks

10 of 29

More layers (Hidden)

Convolutional neural network (CNN)

K. Choo, T. Neupert & G. Carleo, Phys. Rev. B (2019)

C. Roth, A. Szabó & A. H. MacDonald, Phys. Rev. B (2023)

Group convolutional neural network (GCNN)

 

Method

Energy

MPS

RBM

CNN

GCNN

Fermionic MF + Lanczos

Ground state

M. Li et al., IEEE TPDS (2022)

 

 

11 of 29

Optimization

Stochastic gradient descent (SGD)

 

Stochastic reconfiguration (SR)

 

 

 

 

 

 

 

 

A. Chen & M. Heyl, Nat. Phys. (2024)

12 of 29

SR and MinSR (Hidden)

 

 

 

 

 

 

 

 

 

 

 

13 of 29

Benchmark

 

 

 

A. Chen & M. Heyl, Nat. Phys. (2024)

14 of 29

Comparison (Hidden)

Method

Cost

Accuracy

Availability

Exact diagonalization

(ED)

Exponential high

Exact

Small systems

Quantum Monte Carlo

(QMC)

Medium

Very high

Sign-problem free

Mean field

(MF/HF)

Low

Low

Weak correlation

Matrix product state

(MPS)

Medium

Very high

1D / Quasi-1D

Tensor network

(TN)

Very high

High

Most systems

Neural quantum state

(NQS)

High

High

Spin systems

Fermions in develop

15 of 29

NQS in�quantum spin liquids

16 of 29

Quantum spin liquids

Y. Nomura & M. Imada, Phys. Rev. X (2021)

Y. Iqbal, et al., Phys. Rev. B (2016)

 

 

 

A. Chen & M. Heyl, Nat. Phys. (2024)

17 of 29

Resonating valence bond (Hidden)

18 of 29

Structure factor

A. O. Scheie, ..., A. Chen, et al., arXiv:2406.17773 (2024)

 

19 of 29

Phase transition (Hidden)

20 of 29

NQS in�Fermi-Hubbard model

21 of 29

What’s the difficulty?

 

 

 

 

 

 

22 of 29

Determinant

Pfaffian

Hidden fermion

Determinant state

Hidden fermion

Pfaffian state

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

J. R. Moreno, et al., PNAS (2022)

A. Chen, et al., in preparation

23 of 29

Fermionic NQS

 

 

Hidden fermion determinant state (HFDS)

J. R. Moreno, et al., PNAS (2022)

Hidden fermion pfaffian state (HFPS)

A. Chen, et al., in preparation

24 of 29

Complexity (Hidden)

Mean field

Hidden fermion

Backflow

 

 

 

25 of 29

Order parameter

 

 

 

 

 

 

 

26 of 29

Superconducting order (Hidden)

 

d-wave pairing

 

pair-pair correlation

 

 

27 of 29

Summary

  • Motivation

  • Deeper networks

  • Quantum spin liquids

  • Fermi-Hubbard model

28 of 29

Publication

Ao Chen and Markus Heyl

Nat. Ph “Empowering deep neural quantum states through efficient optimization”

ys. 20, 1476 (2024)

Preprints

Ao Chen*, Vighnesh Naik*, and Markus Heyl

“Convolutional transformer wave functions”

arXiv:2503.10462 (2025)

AO Scheie, Minseong Lee, Kevin Wang, P Laurell, ES Choi, D Pajerowski, Qingming Zhang, Jie Ma, HD Zhou, Sangyun Lee, SM Thomas, MO Ajeesh, PFS Rosa, Ao Chen, Vivien S Zapf, M Heyl, CD Batista, E Dagotto,

JE Moore, and D Alan Tennant

“Spectrum and low-energy gap in triangular quantum spin liquid NaYbSe2”

arXiv:2406.17773 (2024)

Hongzheng Zhao, Ao Chen, Shu-Wei Liu, Marin Bukov, Markus Heyl, and Roderich Moessner

“Learning effective Hamiltonians for adaptive time-evolution quantum algorithms”

arXiv:2406.06198 (2024)

In preparation

Ao Chen, Anirvan Sengupta, Antoine Georges, and Christopher Roth

“Hidden fermion pfaffian state”

Vighnesh Naik, Ao Chen, and Markus Heyl

“Neural quantum dynamics assisted by basis rotation”

29 of 29

Thank you.