TurboRVB: A many-body toolkit for ab initio electronic simulations by quantum Monte Carlo

Area of Science:

• Computational Physics
• Quantum Chemistry
• Materials Science

Background:

• Standard mean-field approaches like Density Functional Theory (DFT) often fail for strongly correlated systems.
• Accurate electronic structure calculations are crucial for understanding material properties.
• Quantum Monte Carlo (QMC) methods offer a powerful alternative for complex electronic systems.

Purpose of the Study:

• To introduce TurboRVB, a computational package for ab initio Quantum Monte Carlo (QMC) simulations.
• To enable accurate simulations of molecular and bulk electronic systems, especially those with strong correlations.
• To provide a computationally feasible tool for large-scale electronic structure calculations.

Main Methods:

• Implementation of Variational Monte Carlo (VMC) and lattice regularized diffusion Monte Carlo algorithms.
• Utilization of strongly correlated many-body wave functions (WFs) incorporating Jastrow factors and Pfaffians/antisymmetrized geminal powers.
• Application of adjoint algorithmic differentiation for efficient energy derivative and ionic force calculations.
• Parallelization using hybrid MPI-OpenMP protocols for efficient computation on modern hardware, including GPUs.

Main Results:

• TurboRVB successfully implements advanced QMC algorithms and strongly correlated wave functions.
• The code allows for accurate material descriptions beyond the capabilities of standard DFT methods.
• Efficient evaluation of energy derivatives and ionic forces enables structural optimizations and molecular dynamics.
• Full wave function optimization is achieved through advanced stochastic algorithms.

Conclusions:

• TurboRVB provides a computationally efficient and accurate tool for ab initio QMC simulations.
• The package is capable of handling complex electronic systems and large-scale problems.
• Its features facilitate detailed studies of material properties and dynamics.