Search papers, labs, and topics across Lattice.
This paper introduces Neural Boltzmann Equations (NBEs) to address the computational challenges of studying particle dynamics in the early universe, which are traditionally limited by fixed momentum grids and quadrature integration. By integrating physics-inspired neural distribution functions with Monte Carlo methods and the natural gradient approach, NBEs facilitate efficient parameter scans and enhance the evaluation of high-dimensional phase-space integrals. The framework is validated through a precision calculation of the effective number of relativistic neutrino degrees of freedom, showcasing its potential for complex astrophysical processes.
Neural Boltzmann Equations enable unprecedented precision in modeling early universe particle dynamics, overcoming traditional computational bottlenecks.
The dynamics of particles in the early universe are described by Boltzmann equations, which involve high-dimensional phase-space integrals. Classical approaches use quadrature integration and evolve the system on a fixed momentum grid, which scales poorly to complicated systems and parameter scans, severely limiting the complexity of processes that can be studied. We introduce Neural Boltzmann Equations (NBEs), which combine three coupled concepts to overcome these limitations. First, particle properties are encoded in physics-inspired neural distribution functions, with parameters that can be predicted using neural networks, enabling efficient parameter scans. Second, phase-space integrals are evaluated with Monte Carlo, using importance sampling tools from collider physics. Third, we use the natural gradient method to evolve the system. After demonstrating the individual benefits of NBEs, we use the framework to perform a precision calculation of the effective number of relativistic neutrino degrees of freedom in the early universe.
