Reproduce “Wiener Chaos Expansion based Neural Operator for Singular Stochastic Partial Differential Equations” 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
Dai Shi, Luke Thompson, Andi Han, Peiyan Hu, Junbin Gao, José Miguel Hernández-Lobato
Researchers from the University of Cambridge and University of Sydney developed WCE-FiLM-NO, a neural operator specifically for singular stochastic partial differential equations like the dynamic phi^4_2 model. This model achieved lower relative L2 errors compared to baseline methods and notably performed well without requiring explicit renormalization factors, demonstrating improved generalizability.
Key insights
- Solutions to semi-linear SPDEs can be expressed via Wiener Chaos Expansion (WCE), decomposing them into deterministic propagators and stochastic Wick-Hermite features learnable by neural networks.
- The Da Prato-Debussche decomposition for singular SPDEs allows separating the solution into a smooth remainder (solving a 'Shift Equation') and explicitly representable low-order stochastic convolution terms.
- Feature-wise Linear Modulation (FiLM) provides an effective architectural mechanism to embed the structural decomposition of singular SPDE solutions, enhancing the neural operator's ability to model complex dependencies.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.