Active space selection with self-healing diffusion Monte Carlo algorithms for periodic solids

Area of Science:

• Quantum Chemistry
• Computational Materials Science
• Solid-State Physics

Background:

• Multideterminant Diffusion Monte Carlo (DMC) offers higher accuracy than single-determinant approaches.
• Self-Healing Diffusion Monte Carlo (SHDMC) iteratively refines multideterminant wavefunctions.
• Traditional methods like Configuration Interaction and Complete Active Space (CAS) are limited for solid systems.

Purpose of the Study:

• To develop and evaluate self-contained active space selection algorithms specifically for SHDMC.
• To enable accurate quantum mechanical calculations for solid materials using SHDMC.
• To overcome limitations of existing methods in applying accurate electronic structure calculations to solids.

Main Methods:

• Development of novel active space selection algorithms integrated directly with SHDMC.
• Benchmarking against established methods, including selected CI and reference SHDMC trajectories.
• Application to calculate the ground state energy of a graphene unit cell.

Main Results:

• The proposed SHDMC-based algorithms provide accurate results for solid systems.
• An 'auto-branching' algorithm was identified as optimal for balancing accuracy and computational cost.
• Demonstrated the first completely self-contained DMC-based active space selection without external dependencies.

Conclusions:

• SHDMC, coupled with novel self-contained active space selection, is a viable and accurate method for solids.
• The 'auto-branching' algorithm offers an efficient approach for practical applications.
• This work establishes a new paradigm for electronic structure calculations in condensed matter physics.