Solving the QCD effective kinetic theory with neural networks.

Area of Science:

• High Energy Physics
• Computational Physics
• Nuclear Physics

Background:

• Event-by-event simulations in Quantum Chromodynamics (QCD) kinetic theory are computationally intensive.
• The high dimensionality of the collision integral in the Boltzmann equation presents a significant bottleneck.

Purpose of the Study:

• To investigate the use of neural networks for accelerating QCD kinetic theory simulations.
• To develop a faster method for evaluating the collision integral.

Main Methods:

• Utilized a neural network to estimate the high-dimensional collision integral.
• Compared neural network predictions with traditional Monte Carlo evaluations.
• Verified accuracy by comparing moments of the distribution function for isotropic and anisotropic cases.

Main Results:

• The neural network accurately estimates the collision integral, reducing computation time significantly.
• The network effectively predicts the time evolution of distribution functions.
• Results were validated against traditional methods for various distribution functions.

Conclusions:

• Neural networks offer a computationally efficient alternative for evaluating collision integrals in QCD kinetic theory.
• This approach paves the way for detailed event-by-event modeling of pre-equilibrium stages in heavy-ion collisions.