Physics-informed deep generative models (Mathematical Institute seminar) [developing]
Key Questions
What are physics-informed deep generative models?
Physics-informed deep generative models integrate physical laws like PDEs into generative AI, often using PINNs (Physics-Informed Neural Networks) with Bayesian approaches, dual PINNs, GNPs, and AIMS. They were discussed in a Mathematical Institute seminar on March 18, 2026.
What applications of PINNs are highlighted?
PINNs are applied to Riccati equations, tumor modeling, Srivastava problems, plasma simulations, fractional dynamics, finance LOB/HJB-RL, power systems, PT NLSE, and metasurface LaSt-QGAN. They serve as surrogates for complex systems.
How is AlphaFold used as a prior in structure determination?
AlphaFold acts as a prior for experimental structure determination, conditioning predictions with machine learning advances. It transforms structural biology by enabling swift and accurate protein predictions.
What is the role of fractional calculus in enzymatic competition?
Fractional calculus models RT-IN enzymatic competition in biology, investigating biochemical interplay between reverse transcriptase and integrase. It provides a novel approach to competitive dynamics.
What are AI-driven reduced order models for power converters?
AI-ROMs for resonant power converters offer practical insights into modeling, as shared in a presentation. They leverage AI to simplify simulations for power electronics.
How do robot swarms achieve self-organization?
Harvard PNAS research on robot swarms finds a 'Goldilocks' level of randomness prevents stagnation, enabling effective movement. Too much uniformity or randomness hinders performance in tasks.
What is the inverse problem for wave propagation?
The inverse problem identifies the source function in wave equations, starting with direct problem solutions. It applies numerical methods for source reconstruction in wave propagation.
What are post-quantum fractional quadrature inequalities?
Multiparameter post-quantum fractional quadrature inequalities involve simulations for advanced numerical analysis. They extend classical inequalities to quantum and fractional settings.
2026-03-18 Bayesian/PDE/dual PINN + GNPs + AIMS. New: Hakim plasma DG; Harry Dym frac; Hatch plasma; PINNs Riccati/tumor/Srivastava; coral/Boolean/frac bio; finance LOB/HJB-RL; power/PT NLSE; metasurface LaSt-QGAN; greenhouse; non-Markovian; hp-DG Burgers-Huxley; M1 kinetic; frac PDEs/Riesz/tsunami; NLS Hartree; SVM diabetes; evo bio; GP molecular; Spivak cat bio; Agram BSDE Volterra; AlphaFold priors; AI-ROMs power; Chakraverty ANN; frac enzymatic; wave inverse; elliptic sine-Gordon; RN BH GR; PINNs surrogates/complex systems; RG ODEs; ground reaction PINNs; neutral DE oscillation; post-quantum frac quadrature inequalities; Harvard PNAS robot swarms self-org.