Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
FlowForge is introduced as a staged local rollout engine for predicting CFD flow-fields, addressing the limitations of large, complex deep learning surrogates. It compiles a locality-preserving update schedule and executes it with a shared lightweight local predictor, updating spatial sites stage by stage based on bounded local context. Experiments across PDEBench, CFDBench, and BubbleML demonstrate that FlowForge achieves comparable or superior pointwise accuracy, improved robustness to noise and missing data, and stable multi-step rollout behavior with reduced latency compared to strong baselines.
Unlock faster and more robust CFD flow-field predictions with FlowForge, a staged local rollout engine that outperforms global models while slashing latency.
Deep learning surrogates for CFD flow-field prediction often rely on large, complex models, which can be slow and fragile when data are noisy or incomplete. We introduce FlowForge, a staged local rollout engine that predicts future flow fields by compiling a locality-preserving update schedule and executing it with a shared lightweight local predictor. Rather than producing the next frame in a single global pass, FlowForge rewrites spatial sites stage by stage so that each update conditions only on bounded local context exposed by earlier stages. This compile-execute design aligns inference with short-range physical dependence, keeps latency predictable, and limits error amplification from global mixing. Across PDEBench, CFDBench, and BubbleML, FlowForge matches or improves upon strong baselines in pointwise accuracy, delivers consistently better robustness to noise and missing observations, and maintains stable multi-step rollout behavior while reducing per-step latency.