Energy-based truncation of multi-determinant wavefunctions in quantum Monte Carlo.

Area of Science:

• Computational chemistry
• Quantum mechanics
• Electronic structure theory

Background:

• Diffusion Monte Carlo (DMC) calculations often require truncating large multi-determinant expansions.
• Existing truncation methods rely on wavefunction-based criteria, which may not always be optimal.

Purpose of the Study:

• To introduce a novel, energy-based criterion for truncating multi-determinant expansions in DMC calculations.
• To demonstrate the effectiveness of this energy-based approach across diverse chemical systems.

Main Methods:

• The proposed method truncates determinants based on their contribution to the total energy, rather than wavefunction amplitudes.
• This approach utilizes quantities readily available from standard configuration interaction (CI) calculations.

Main Results:

• The energy-based truncation shows consistent behavior across systems with varying electronic correlation.
• Accurate calculations of the electron affinity of oxygen and the atomization energy of the carbon dimer validate the method.
• Effective error cancellation is achieved in energy differences.

Conclusions:

• The energy-based truncation method offers a more intuitive and robust alternative to wavefunction-based criteria.
• Its simplicity and ease of implementation make it a valuable tool for computational chemists.
• The method enables accurate and efficient calculations of key chemical properties.