Mechanism Learning: Prototype-Anchored Mechanism Inference for Scientific Forecasting

TL;DR

This paper introduces mechanism learning, a novel framework for scientific forecasting that estimates local mechanisms to improve robustness and accuracy, especially in data-scarce and complex regimes.

Contribution

It proposes a structured mechanism space with prototype anchors to enhance forecasting stability and performance over traditional direct prediction methods.

Findings

Mechanism spaces resist collapse and maintain local consistency.

Significant improvements in switching stability for Burgers dynamics.

State-of-the-art performance in WeatherBench2 and Lorenz96 benchmarks.

Abstract

Scientific forecasting typically relies on direct state prediction, an approach that grows brittle under data scarcity, extended horizons, non-stationary dynamics, or high-dimensional complexity. While raw state trajectories are highly sensitive in these regimes, underlying local evolution rules often exhibit robust reusability. We introduce mechanism learning, a framework that forecasts future states by estimating the currently active local mechanism. Our method compresses local spatiotemporal fragments into mechanism descriptors, forming a data-driven, structured mechanism space where proximity reflects similar local evolution rules. To ground these estimates in observed data, we utilize prototype anchors, a set of representative mechanisms that sparsely cover the space of local rules. We evaluate this approach on Burgers dynamics, WeatherBench2, and Lorenz96. Empirically, the learned mechanism spaces resist collapse and maintain strong local consistency. Compared to direct prediction and other models including FNO, NODE, LSTM, and reservoir-family methods, our framework demonstrates predictive gains in fragile regimes: it significantly improves switching stability in Burgers dynamics and achieves state-of-the-art performance both under the scarce-data fixed-horizon WeatherBench2 protocol and in intermediate-complexity Lorenz96. Ablation studies and drift diagnostics confirm that these improvements are driven by finite prototype anchoring rather than sheer latent capacity. Together, these results establish mechanism learning as a principled, robust alternative to direct state prediction in forecasting complex systems.