Speeding up local correlation methods: System-inherent domains.

Area of Science:

• Computational chemistry
• Quantum mechanics
• Theoretical physics

Background:

• Correlated calculations require defining local virtual spaces.
• Standard methods for defining these spaces can be inefficient or problematic.

Purpose of the Study:

• To present a new, adaptive approach for determining local virtual spaces.
• To improve the efficiency and accuracy of correlated calculations.

Main Methods:

• Restricting virtual space in a pair-specific manner.
• Using a preceding approximate calculation to adapt to problem locality.
• Generating pair system-inherent domains.

Main Results:

• The new domains are significantly smaller than traditional ones.
• Accuracy is maintained with minimal loss.
• Integral transformations and residual evaluations are accelerated.
• Memory requirements are reduced.

Conclusions:

• The system-inherent domains offer a more efficient approach to correlated calculations.
• This method is particularly beneficial for high-accuracy requirements like pair-natural orbital generation.
• It addresses challenges in standard domain definitions for problems such as excited-state calculations.