Stochastic Difference-Dedicated Configuration Interaction for Magnetic Exchange in Large Active Spaces

Area of Science:

• Computational chemistry
• Quantum mechanics
• Materials science

Background:

• Simulating magnetic properties in strongly correlated systems is a key challenge in electronic structure theory.
• The Difference-Dedicated Configuration Interaction (DDCI) method is accurate but computationally expensive for large systems.
• Existing methods struggle with the steep growth of computational space as system size increases.

Purpose of the Study:

• To develop a computationally tractable formulation of DDCI for larger magnetic systems.
• To overcome the limitations of conventional DDCI in simulating magnetic exchange couplings.
• To enable accurate electronic structure calculations for complex magnetic materials.

Main Methods:

• Introduced a stochastic formulation of DDCI using Full Configuration Interaction Quantum Monte Carlo (FCIQMC).
• Integrated DDCI with the Generalized Active Space framework to manage computational complexity.
• Validated the Stochastic-DDCI approach against conventional DDCI for a trinuclear manganese cluster.

Main Results:

• Conventional DDCI with a small active space failed to reproduce experimental magnetic coupling data.
• Stochastic-DDCI, using a significantly larger active space, accurately reproduced the experimental spin ladder.
• The new method achieved remarkable accuracy, with deviations below 33 cm⁻¹ compared to experimental data.

Conclusions:

• Stochastic-DDCI significantly expands the applicability of DDCI methodologies to larger active spaces.
• This advancement allows for the study of more complex magnetic systems previously inaccessible.
• The developed method offers a powerful tool for accurate simulation of magnetic properties in correlated materials.