Determinant and Derivative-Free Quantum Monte Carlo Within the Stochastic Representation of Wavefunctions

Liam Bernheimer,Hristiana Atanasova,Guy Cohen

2024-02-10

Abstract:Describing the ground states of continuous, real-space quantum many-body systems, like atoms and molecules, is a significant computational challenge with applications throughout the physical sciences. Recent progress was made by variational methods based on machine learning (ML) ansatzes. However, since these approaches are based on energy minimization, ansatzes must be twice differentiable. This (a) precludes the use of many powerful classes of ML models; and (b) makes the enforcement of bosonic, fermionic, and other symmetries costly. Furthermore, (c) the optimization procedure is often unstable unless it is done by imaginary time propagation, which is often impractically expensive in modern ML models with many parameters. The stochastic representation of wavefunctions (SRW), introduced in Nat Commun 14, 3601 (2023), is a recent approach to overcoming (c). SRW enables imaginary time propagation at scale, and makes some headway towards the solution of problem (b), but remains limited by problem (a). Here, we argue that combining SRW with path integral techniques leads to a new formulation that overcomes all three problems simultaneously. As a demonstration, we apply the approach to generalized ``Hooke's atoms'': interacting particles in harmonic wells. We benchmark our results against state-of-the-art data where possible, and use it to investigate the crossover between the Fermi liquid and the Wigner molecule within closed-shell systems. Our results shed new light on the competition between interaction-driven symmetry breaking and kinetic-energy-driven delocalization.

Strongly Correlated Electrons,Mesoscale and Nanoscale Physics,Quantum Physics

What problem does this paper attempt to address?

This paper primarily explores how to use the Stochastic Representation of Wavefunctions (SRW) in quantum Monte Carlo methods to address the computational challenges of describing the ground state of continuous real-space quantum many-body systems. Traditional methods such as Variational Monte Carlo (VMC) rely on twice-differentiable wave function ansatz, which limits the use of machine learning models and incurs high costs in handling exchange symmetries (such as fermions or bosons). The paper proposes a new framework that combines path integral techniques, allowing the ansatz to be non-differentiable or discontinuous, thus overcoming these three issues.

Specifically, the new method includes the following steps:
1. Provide an initial wave function guess.
2. Perform imaginary-time evolution through sampling points.
3. Use path integrals to obtain an approximate propagation value of the wave function.
4. Find a new wave function estimate through regression, while considering symmetry constraints.
5. Iterate these steps until the energy estimate converges.
6. A variational upper bound on the energy can be obtained.

By using path integrals, researchers are able to enforce exchange symmetries at a lower cost without sacrificing the general expressiveness. The paper validates this method by simulating "Hooke's atoms" with harmonic oscillator potentials and Coulomb interactions, studying the transition states from non-interacting to classical Wigner molecules, and revealing the competition between interaction-driven symmetry breaking and kinetic-energy-driven localization.

In summary, the new method proposed in the paper, enhanced SRW with path integrals, resolves the limitations of applying deep learning models in quantum many-body systems, improves efficiency and accuracy, and allows for efficient handling of symmetries.