Technology Category
Machine Learning: Life-Long and Continual LearningComputer Vision: Diffusion Models for VisionSearch and Optimization: Learning to Search
Adaptive recurrent flow map operator learning for reaction diffusion dynamics
🤖 AI Summary
This work addresses the challenges of autoregressive error accumulation and poor out-of-distribution generalization in long-term modeling of reaction–diffusion systems by proposing DDOL-ART, a purely data-driven operator learning method. DDOL-ART employs an adaptive recurrent training strategy to learn single-step neural operators from short time-series data and incorporates a lightweight validation mechanism that dynamically terminates unreliable predictions and redirects the optimization trajectory. This approach enables stable long-term forecasting and zero-shot generalization under significant morphological changes—without relying on physics-based residual terms. Experiments on the FitzHugh–Nagumo, Gray–Scott, and Lambda–Omega systems demonstrate that DDOL-ART achieves several-fold faster training than physics-residual-based methods while exhibiting superior in-distribution stability and out-of-distribution robustness.
Technology Category
Application Category
📝 Abstract
Reaction-diffusion (RD) equations underpin pattern formation across chemistry, biology, and physics, yet learning stable operators that forecast their long-term dynamics from data remains challenging. Neural-operator surrogates provide resolution-robust prediction, but autoregressive rollouts can drift due to the accumulation of error, and out-of-distribution (OOD) initial conditions often degrade accuracy. Physics-based numerical residual objectives can regularize operator learning, although they introduce additional assumptions, sensitivity to discretization and loss design, and higher training cost. Here we develop a purely data-driven operator learner with adaptive recurrent training (DDOL-ART) using a robust recurrent strategy with lightweight validation milestones that early-exit unproductive rollout segments and redirect optimization. Trained only on a single in-distribution toroidal Gaussian family over short horizons, DDOL-ART learns one-step operators that remain stable under long rollouts and generalize zero-shot to strong morphology shifts across FitzHugh-Nagumo (FN), Gray-Scott (GS), and Lambda-Omega (LO) systems. Across these benchmarks, DDOL-ART delivers a strong accuracy and cost trade-off. It is several-fold faster than a physics-based numerical-loss operator learner (NLOL) under matched settings, and it remains competitive on both in-distribution stability and OOD robustness. Training-dynamics diagnostics show that adaptivity strengthens the correlation between validation error and OOD test error performance, acting as a feedback controller that limits optimization drift. Our results indicate that feedback-controlled recurrent training of DDOL-ART generates robust flow-map surrogates without PDE residuals, while simultaneously maintaining competitiveness with NLOL at significantly reduced training costs.
Problem
Research questions and friction points this paper is trying to address.
reaction-diffusion dynamics
operator learning
long-term stability
out-of-distribution generalization
autoregressive rollout
Innovation
Methods, ideas, or system contributions that make the work stand out.
adaptive recurrent training
data-driven operator learning
reaction-diffusion dynamics
zero-shot generalization
validation milestone
🔎 Similar Papers
No similar papers found.
