Deep Neural Networks for Modeling Astrophysical Nuclear Reacting Flows
arXiv:2504.14180 · doi:10.3847/1538-4357/adf331
Abstract
In astrophysical simulations, nuclear reacting flows pose computational challenges due to the stiffness of reaction networks. We introduce neural network-based surrogate models using the DeePODE framework to enhance simulation efficiency while maintaining accuracy and robustness. Our method replaces conventional stiff ODE solvers with deep learning models trained through evolutionary Monte Carlo sampling from zero-dimensional simulation data, ensuring generalization across varied thermonuclear and hydrodynamic conditions. Tested on 3-species and 13-species reaction networks, the models achieve accuracy relative to semi-implicit numerical solutions and deliver a speedup on CPUs. A temperature-thresholded deployment strategy ensures stability in extreme conditions, sustaining neural network utilization above 75\% in multi-dimensional simulations. These data-driven surrogates effectively mitigate stiffness constraints, offering a scalable approach for high-fidelity modeling of astrophysical nuclear reacting flows.
References in corpus (19)
- A Multi-Code Analysis Toolkit for Astrophysical Simulation Data
- CASTRO: A New Compressible Astrophysical Solver. I. Hydrodynamics and Self-Gravity
- Head-on collisions of white dwarfs in triple systems could explain type Ia supernova
- Three-Dimensional Delayed-Detonation Model of Type Ia Supernova
- Deflagrations and Detonations in Thermonuclear Supernovae
- Double Detonations with Thin, Modestly Enriched Helium Layers Can Make Normal Type Ia Supernovae
- Double detonations of sub-Chandrasekhar mass CO white dwarfs: Can different core and He shell masses explain variations of Type Ia supernovae?
- High resolution simulations of the head-on collision of white dwarfs
- Magneto-Hydrodynamical Effects on Nuclear Deflagration Fronts in Type Ia Supernovae
- MAESTROeX: A Massively Parallel Low Mach Number Astrophysical Solver
- Improved Coupling of Hydrodynamics and Nuclear Reactions via Spectral Deferred Corrections
- An accurate and efficient numerical calculation of detonation waves in multidimensional supernova simulations using a burning limiter and adaptive quasi-statistical equilibrium
- pynucastro: an interface to nuclear reaction rates and code generator for reaction network equations
- Neural Networks for Nuclear Reactions in MAESTROeX
- The Impact of Resolution on Double-Detonation Models for Type Ia Supernovae
- Confronting double-detonation sub-Chandrasekhar models with the low-luminosity suppression of Type Ia supernovae
- Comparing Early Evolution of Flames in X-ray Bursts in Two and Three Dimensions
- Hydrodynamical simulations of proton ingestion flashes in Type I X-ray Bursts
- Simulating Lateral H/He Flame Propagation in Type I X-ray Bursts