Deterministic optimization of Jastrow factors

Area of Science:

• Quantum Chemistry
• Computational Physics
• Electronic Structure Theory

Background:

• Jastrow factors are essential in variational Monte Carlo (VMC) and transcorrelated (TC) methods for accurate electronic structure.
• Optimizing Jastrow factors with VMC can introduce noise, hindering calculations of weak interactions.
• Current TC methods achieve high accuracy but face challenges with VMC-derived Jastrow noise.

Purpose of the Study:

• To develop a deterministic alternative to VMC for Jastrow factor optimization.
• To eliminate noise in Jastrow factors for improved accuracy in TC and VMC calculations.
• To provide a reproducible method for optimizing Jastrow functions.

Main Methods:

• Minimizing the "variance of the TC reference energy" in a standard basis set.
• Deriving and implementing analytic expressions for the derivatives of TC Hamiltonian matrix elements.
• Applying the method to first-row atoms and molecules.

Main Results:

• The deterministic method produces noise-free Jastrow factors.
• Optimized Jastrow functions yield Slater-Jastrow wavefunctions with low variance.
• The method achieves lower energies compared to VMC energy-minimization, comparable to TC approaches.
• Results are reproducible and comparable to VMC variance optimization.

Conclusions:

• The proposed deterministic method offers a viable, noise-free alternative for Jastrow optimization.
• This approach improves accuracy in transcorrelated methods and standard VMC.
• It provides a reproducible and effective way to refine Jastrow functions for electronic structure calculations.