Hybrid Neural World Models

2026-05-27

AI Summary

This work addresses the silent failure of neural surrogates in simulating physical systems with discontinuities such as shocks and fronts. The authors propose a hybrid neural world model that employs a single network to directly predict arbitrary future states in continuous time, implicitly encoding discontinuity locations without requiring explicit supervision of jump points. By leveraging forward propagation to generate trajectory-level error maps, the method precisely localizes anomalous regions and integrates a fallback mechanism based on a reference solver alongside unsupervised uncertainty estimation, enabling efficient inference and adaptive correction. Evaluated on reaction–diffusion, compressible Euler, and rigid-body collision systems, the model achieves 26–72× CPU speedup; incorporating error-map-based fallback reduces approximation error by approximately 50%, substantially outperforming existing unsupervised baselines.

Abstract

Neural surrogates promise large speedups over classical solvers for physical dynamics but fail silently at sharp dynamical events such as shocks, fronts, and contact. We present hybrid neural world models for physical dynamics: a recipe for training and deploying multi-horizon surrogates in physical state space, where a single network with continuous horizon conditioning is trained with direct supervision against textbook reference solvers to predict any future state at horizon T in one forward pass. Although no part of the training data, loss function, or architecture supervises discontinuity location, the trained surrogate encodes it implicitly, recoverable from its forward passes alone as a per-trajectory error map that concentrates on shocks, fronts, and contacts, and stays small elsewhere. The map is competitive with or better than standard label-free baselines including deep ensembles, learned error heads, gradient-magnitude indicators, and locally-adaptive conformal prediction, while using only a single trained network and requiring no calibration set or governing-equation knowledge. The recipe supports two operating points. Mode 1 runs the surrogate alone for maximum throughput, with same-hardware CPU speedups of 26x to 72x against textbook solvers on the PDE environments. Mode 2 uses the error map to gate a reference-solver fallback, deferring uncertain trajectories and roughly halving the surrogate's residual error at the default operating point. The recipe applies without modification across reaction-diffusion, compressible Euler, and rigid-body collision dynamics.

Problem

• neural surrogates
• physical dynamics
• discontinuities
• shocks
• fronts

Innovation

• hybrid neural world models
• neural surrogates
• discontinuity detection
• multi-horizon prediction
• physics-informed machine learning