Response properties in phaseless auxiliary field quantum Monte Carlo.

Area of Science:

• Quantum chemistry
• Computational physics
• Materials science

Background:

• Calculating molecular properties is crucial for understanding chemical reactions and material behavior.
• Traditional methods for calculating response properties can be computationally expensive.
• Phaseless auxiliary field quantum Monte Carlo (AFQMC) is a powerful method for electronic structure calculations.

Purpose of the Study:

• To develop a computationally efficient method for calculating first-order response properties in phaseless AFQMC.
• To assess the biases and statistical efficiency of the new method.
• To investigate the impact of self-consistency and trial orbital choice on property calculations.

Main Methods:

• Application of automatic differentiation (AD) to phaseless AFQMC.
• Calculation of reduced density matrices using AD.
• Comparison of AD-based property calculations with self-consistently optimized orbitals and density functional theory (DFT) orbitals.

Main Results:

• AD enables calculation of reduced density matrices with the same computational cost scaling per sample as energy calculations.
• The cost prefactor for property calculations using AD is less than four in numerical tests.
• DFT-optimized orbitals provide accurate dipole moments for selected molecules, comparable to self-consistently optimized orbitals.

Conclusions:

• Automatic differentiation provides an efficient and accurate way to compute first-order response properties in phaseless AFQMC.
• The method significantly reduces the computational overhead for property calculations.
• DFT orbitals are a viable and efficient choice for property calculations in many cases, simplifying the computational workflow.