Fast semistochastic heat-bath configuration interaction

Area of Science:

• Quantum Chemistry
• Computational Physics
• Theoretical Chemistry

Background:

• Solving the many-body Schrödinger equation is computationally intensive.
• Accurate quantum chemical calculations require large configuration interaction expansions.
• Existing selected configuration interaction methods face scalability challenges.

Purpose of the Study:

• To present a fast semistochastic heat-bath configuration interaction (SHCI) method.
• To improve computational efficiency by eliminating bottlenecks in variational and perturbative steps.
• To enable calculations with a significantly larger number of determinants.

Main Methods:

• Developed a fast semistochastic heat-bath configuration interaction (SHCI) algorithm.
• Optimized variational and perturbative steps of the SHCI method.
• Implemented parallelization and efficient data structures like distributed hash tables.

Main Results:

• The improved SHCI method includes two orders of magnitude more determinants than previous methods.
• Calculated benchmark energy for chromium dimer using X2C relativistic Hamiltonian.
• Largest calculation involved two billion Slater determinants and trillions of perturbative determinants with <10⁻⁵ Ha uncertainty.

Conclusions:

• The enhanced SHCI method offers a significant advancement in solving the many-body Schrödinger equation.
• This approach enables highly accurate quantum chemical calculations for complex systems.
• The method provides a scalable and efficient tool for electronic structure calculations.