Reproduce “Enhancing predictive capabilities in fusion burning plasmas through surrogate-based optimization in core transport solvers” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
P. Rodriguez-Fernandez, N.T. Howard, A. Saltzman, S. Kantamneni, J. Candy, C. Holland, M. Balandat, S. Ament and 1 more
Researchers at MIT Plasma Science and Fusion Center developed PORTALS, a surrogate-based optimization framework, to accelerate core transport solvers using nonlinear gyrokinetic simulations for fusion burning plasmas. This framework enabled a 5-channel DIII-D plasma prediction with 26 profile evaluations, demonstrating feasibility for high-fidelity, multi-channel predictions while also improving efficiency for parameter scans with an average of 5 extra evaluations per scenario.
Key insights
- Reformulating flux-matching as a surrogate-based optimization problem significantly reduces the computational burden of high-fidelity nonlinear gyrokinetic simulations for core transport.
- Bayesian optimization, combined with physics-informed Gaussian processes and noise handling, improves robustness to stiff transport behavior and inherent noise in near-marginal stability conditions.
- Strategic initialization with simple relaxation and subsequent surrogate training provides realistic samples and guides the optimization effectively, accelerating convergence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.