Taming the fixed-node error in diffusion Monte Carlo via range separation

Area of Science:

• Quantum Chemistry
• Computational Physics

Background:

• Accurate electronic structure calculations are crucial for understanding chemical systems.
• Combining density-functional theory (DFT) and wave function theory (WFT) offers a promising route to improve accuracy.
• Fixed-node diffusion Monte Carlo (FN-DMC) is a powerful WFT method but often requires complex trial wave functions.

Purpose of the Study:

• To develop a novel computational scheme, RS-DFT-CIPSI, by combining DFT and WFT.
• To obtain accurate FN-DMC energies using compact multi-determinant trial wave functions.
• To assess the efficiency and accuracy of the RS-DFT-CIPSI method for chemical applications.

Main Methods:

• Range separation (RS) of the interelectronic Coulomb operator.
• Combining short-range exchange-correlation functionals with Configuration Interaction using a Perturbative Selection made Iteratively (CIPSI).
• Fixed-node diffusion Monte Carlo (FN-DMC) energy calculations using RS-DFT-CIPSI trial wave functions.

Main Results:

• RS-DFT-CIPSI yields lower FN-DMC energies with more compact multi-determinant expansions compared to conventional CIPSI.
• The number of determinants required for a given accuracy is significantly reduced.
• The RS-DFT scheme effectively mimics short-range correlation effects, simplifying the process.

Conclusions:

• RS-DFT-CIPSI provides an efficient and accurate method for electronic structure calculations.
• The method demonstrates effective error cancellation and produces compact trial wave functions.
• This approach is well-suited for the accurate description of large chemical systems.