Reproduce “Multiphase-Diff: Diffusion-Based Generative Modeling for High-Contrast Multiphase Physical Systems with Sharp Interfaces” 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
Yining Huang, Zhenyu Liang
This framework, Multiphase-Diff, generates physically consistent fields for high-contrast multiphase systems featuring sharp interfaces. It addresses difficulties in accurately modeling interface irregularities and low-amplitude phases by enforcing local conservation laws through novel representations and a preconditioned likelihood.
Key insights
- Conservative flux residuals provide a stable and physically consistent method for enforcing local conservation laws across sharp material interfaces without introducing nonphysical penalties.
- Transforming high-contrast data into a normalized, positive-constrained latent space is essential for generative models to accurately represent and resolve both high and low-amplitude physical features.
- Local normalization of physical residuals is critical for balanced supervision, preventing high-magnitude phases from dominating the training objective and ensuring fidelity across the entire field.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.