Data-Driven Construction of a Generalized Kinetic Collision Operator from Molecular Dynamics

Area of Science:

• Plasma Physics
• Computational Physics
• Statistical Mechanics

Background:

• Conventional collision operators, like the Landau operator, often assume simplified collision dynamics.
• These models may not fully capture complex energy transfer mechanisms in plasmas, particularly those influenced by collective environmental interactions.

Purpose of the Study:

• To develop a generalized collision operator using a data-driven approach from molecular dynamics simulations.
• To incorporate symmetry-breaking terms dependent on both relative and average velocities to model heterogeneous energy transfer.

Main Methods:

• Utilized molecular dynamics to generate data for learning a novel collision operator.
• Developed a collision operator with a symmetry-breaking form, incorporating average velocity dependence.

Main Results:

• The novel operator accurately captures heterogeneous energy transfer arising from collective environmental interactions.
• Preserving the anisotropy introduced by the average velocity dependence is critical for accurate predictions.
• The proposed model outperforms the Landau model in the moderate coupling regime.

Conclusions:

• The data-driven, symmetry-breaking collision operator provides a more accurate description of plasma kinetics and transport coefficients.
• The inclusion of average velocity dependence is essential for modeling anisotropic energy transfer in plasmas.
• This approach offers a significant advancement over conventional models for specific plasma regimes.