Recommended Citation Contexts

Scalable Neural Quantum States Architecture for Quantum Chemistry

Tianchen Zhao, James Stokes, Shravan Veerapaneni

Machine Learning: Science and Technology, 2023

Citations: 65

Links

Abstract

Variational optimization of neural-network representations of quantum states has been successfully applied to solve interacting fermionic problems. Despite rapid developments, significant scalability challenges arise when considering molecules of large scale, which correspond to non-locally interacting quantum spin Hamiltonians consisting of sums of thousands or even millions of Pauli operators. In this work, we introduce scalable parallelization strategies to improve neural-network-based variational quantum Monte Carlo calculations for ab-initio quantum chemistry applications. We establish GPU-supported local energy parallelism to compute the optimization objective for Hamiltonians of potentially complex molecules. Using autoregressive sampling techniques, we demonstrate systematic improvement in wall-clock timings required to achieve CCSD baseline target energies. The performance is further enhanced by accommodating the structure of resultant spin Hamiltonians into the autoregressive sampling ordering. The algorithm achieves promising performance in comparison with the classical approximate methods and exhibits both running time and scalability advantages over existing neural-network based methods.

One-Sentence Summary

Introduces GPU-scalable parallelization strategies for neural-network variational Monte Carlo applied to ab-initio quantum chemistry, achieving systematic speedups over existing neural-network methods.

Key Contributions

Cite This Paper If...

You are working on scalable neural quantum states, neural-network variational Monte Carlo for quantum chemistry, autoregressive neural wavefunctions, or GPU-scaled quantum chemistry. This paper is relevant whenever you need efficient parallelization strategies for neural-network-based ab-initio electronic structure calculations or are benchmarking neural quantum state methods against classical quantum chemistry baselines like CCSD.

Frequently Asked Questions

What is the best method for scaling neural quantum states to large molecules?
This paper proposes GPU-supported local energy parallelism combined with autoregressive sampling, enabling neural-network VMC to handle Hamiltonians with thousands to millions of Pauli operators for complex molecules.
How does autoregressive sampling compare to MCMC for variational quantum Monte Carlo?
This paper shows that autoregressive sampling enables exact, parallelizable sampling without the bottlenecks of Markov Chain Monte Carlo, leading to systematic wall-clock improvements for reaching CCSD-level target energies.
Can neural quantum states achieve chemical accuracy for quantum chemistry?
This paper demonstrates that the scalable neural quantum states architecture achieves promising performance compared to classical approximate methods (CCSD), with both runtime and scalability advantages over prior neural-network approaches.
What is the role of Hamiltonian structure in neural quantum state optimization?
This paper finds that accommodating the structure of spin Hamiltonians into the autoregressive sampling ordering further enhances performance, suggesting that problem-aware architectures are critical for efficiency.
How do GPU parallelization strategies improve neural VMC?
This paper establishes local energy parallelism on GPUs that enables efficient computation of the optimization objective for large Hamiltonians, overcoming the primary scalability bottleneck in neural-network-based quantum chemistry.

Keywords

scalable neural quantum states neural quantum states quantum chemistry variational Monte Carlo quantum chemistry autoregressive neural quantum states GPU-scaled quantum chemistry ab initio neural quantum states

Related Papers

BibTeX

% TODO: BibTeX entry
@article{zhao2023scalablenqs,
  title={Scalable Neural Quantum States Architecture for Quantum Chemistry},
  author={Zhao, Tianchen and Stokes, James and Veerapaneni, Shravan},
  journal={Machine Learning: Science and Technology},
  year={2023}
}