Quantum Monte Carlo Methods in Many-Body Systems

Quantum Monte Carlo (QMC) methods constitute a family of stochastic techniques for solving the many-body Schrödinger equation with direct treatment of electron correlation. By sampling high-dimensional wavefunction amplitudes, approaches such as variational Monte Carlo (VMC) and diffusion Monte Carlo (DMC) can achieve near‐exact ground-state energies within controlled approximations. Central to these methods is the construction of a trial wavefunction, often enhanced by Jastrow factors to capture dynamic correlation or constrained by nodal surfaces to enforce fermionic antisymmetry. Recent progress in neural-network-based ansätze and hybrid algorithms has dramatically improved the flexibility of trial states, enabling accurate treatment of molecules, solids and model systems across diverse regimes of electronic correlation. Applications range from benchmarks of molecular binding and phase stability to predictions of quantum phase transitions and excited-state properties. QMC’s favourable scaling with system size and its amenability to high-performance computing make it a cornerstone for predictive many-body simulations in chemistry, condensed-matter physics and materials science.

Research from Nature Portfolio

Advances in neural-network trial functions have revitalised diffusion Monte Carlo for complex systems. One study integrates deep architectures within fixed-node DMC to optimise nodal surfaces for a broad range of atomic and molecular systems, yielding benchmark-quality ground-state energies and refined binding-energy estimates via an empirical VMC–DMC extrapolation. A parallel effort extends molecular neural networks to periodic solids by incorporating boundary conditions that enable ab initio QMC calculations of cohesive energies and electron densities for systems from one-dimensional hydrogen chains to two-dimensional graphene, attaining accuracy competitive with established methods. Finally, deep variational Monte Carlo has been used to compute low-lying excited states directly with a neural-network ansatz, achieving high-precision results for small molecules and tackling larger targets such as benzene and ethylene conical intersections, thereby broadening QMC’s scope beyond ground-state properties.

Quantum Monte Carlo Methods in Many-Body Systems publication trend

The graph below shows the total number of articles in quantum monte carlo methods in many-body systems across all publications each year (not limited to Nature Index journals).

Technical terms

Variational Monte Carlo (VMC): A stochastic method that optimises a trial wavefunction by minimising its energy expectation value via Monte Carlo sampling.

Diffusion Monte Carlo (DMC): A projector QMC technique that refines a trial wavefunction towards the ground state by simulating imaginary-time diffusion under fixed-node constraints.

Fixed-node approximation: A constraint enforcing the zeroes of the trial wavefunction to maintain antisymmetry, mitigating the sign problem at the expense of nodal bias.

Jastrow factor: A multiplicative function in the trial wavefunction that accounts for electron–electron correlation by enforcing cusp conditions and short-range correlations.

Fermionic neural network: A deep-learning-based wavefunction ansatz that embeds antisymmetry and correlation patterns, enhancing variational flexibility for many-electron systems.

Auxiliary-field QMC (AFQMC): A method that transforms interacting electron problems into fluctuating one-body problems via auxiliary fields, solved by Monte Carlo sampling.

Nodal surface: The hypersurface across which the wavefunction changes sign; its accurate representation is critical for controlling fermionic sign errors in QMC.

References

1. Towards the ground state of molecules via diffusion Monte Carlo on neural networks. Nature Communications (2023).
2. Ab initio calculation of real solids via neural network ansatz. Nature Communications (2022).
3. Electronic excited states in deep variational Monte Carlo. Nature Communications (2023).
4. Discovering Quantum Phase Transitions with Fermionic Neural Networks. Physical Review Letters (2023).
5. QMCPACK: Advances in the development, efficiency, and application of auxiliary field and real-space variational and diffusion quantum Monte Carlo. The Journal of Chemical Physics (2020).