Excited state mean-field theory without automatic differentiation.

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Theoretical Chemistry

Background:

• Accurate prediction of excited states is crucial for understanding chemical processes.
• Existing methods for excited state calculations can be computationally expensive.
• Charge transfer excitations involve significant electron redistribution, posing a challenge for theoretical models.

Purpose of the Study:

• To develop a more computationally efficient formulation of excited state mean-field theory (ESMFT).
• To improve the prediction of charge redistribution during charge transfer excitations.
• To reduce the computational cost associated with calculating excited state properties.

Main Methods:

• Analytical expression of wave function optimization derivatives using Fock-like matrices.
• Grouping Fock builds to minimize access to two-electron integrals.
• Implementation of shell-pair screening strategy for cubic cost scaling.
• Comparison with coupled cluster benchmark calculations.

Main Results:

• Reduced demand for memory-intensive two-electron integrals.
• Achieved cubic overall cost scaling for ESMFT calculations.
• Demonstrated accurate prediction of charge density changes in charge transfer excitations.
• ESMFT captures orbital relaxation effects and avoids self-interaction errors.

Conclusions:

• The new ESMFT formulation offers a significant reduction in computational cost.
• ESMFT provides a more accurate description of charge transfer excitations compared to other low-cost methods.
• This work paves the way for more efficient and accurate theoretical studies of excited state phenomena.