Reproduce “OPINE: Orthogonal Projection in Invertible Nonlinear Embeddings for Turbulent Flows” 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
Anqiao Ouyang, Hongyi Ke, Qi Wang
OPINE compresses flow fields by applying an invertible learned transformation and projecting onto leading POD directions. The projection discards information, while variance-ordered directions provide nested reductions. At D = 32 on forced Burgers turbulence, median test NMSE is 2.2 × 10⁻⁶, versus 7.5 × 10⁻⁴ for a CNN autoencoder, 2.5 × 10⁻³ for an MLP autoencoder, and 1.2 × 10⁻³ for POD; this reconstruction result does not establish universal superiority or predictive dynamics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.