Efficient Langevin dynamics for "noisy" forces.

Area of Science:

• Computational Physics
• Materials Science
• Quantum Chemistry

Background:

• Efficient Boltzmann-sampling is crucial for materials simulations but computationally expensive for large systems.
• Stochastic methods offer scalability but introduce noisy forces, limiting sampling efficiency and stability.
• First-order Langevin dynamics (FOLD) is efficient for deterministic forces but struggles with noisy ones.

Purpose of the Study:

• To develop a novel, general, optimal, and stable sampling approach for FOLD under noisy forces.
• To improve the efficiency and applicability of Boltzmann-sampling in large-scale materials simulations.
• To address the limitations of existing methods when dealing with stochastic electronic structure calculations.

Main Methods:

• Developed a new, general, optimal, and stable sampling strategy for FOLD with noisy forces.
• Applied the enhanced FOLD approach to silicon nanocrystals.
• Utilized stochastic density functional theory for electronic structure calculations.

Main Results:

• Achieved an order-of-magnitude improvement in sampling efficiency.
• Demonstrated enhanced stability of the FOLD method under noisy force conditions.
• Successfully applied the method to complex systems like silicon nanocrystals.

Conclusions:

• The developed approach significantly advances efficient Boltzmann-sampling for large systems with noisy forces.
• This method provides a more stable and efficient alternative for stochastic first-principles simulations.
• Opens new avenues for accurate and scalable simulations in computational materials science.