SPIN-ODE: Stiff Physics-Informed Neural ODE for Chemical Reaction Rate Estimation

2025-05-08

AI Summary

Estimating rate constants for stiff chemical kinetic systems in atmospheric chemistry remains challenging due to training instability and poor convergence. Method: We propose the first Physics-Informed Neural ODE framework tailored for stiff chemical reactions, featuring a three-stage co-optimization strategy that decouples concentration trajectory modeling from rate coefficient inversion. Our approach integrates an explicit Chemical Reaction Neural Network (CRNN), stiff ODE solvers (Tsit5/Cash–Karp), and embedded physical constraints. Results: Evaluated on synthetic and real atmospheric datasets, our method achieves a 3.2× speedup in convergence and reduces relative error in rate constant estimation by 57%, significantly improving both accuracy and training robustness. Contribution: This work represents the first systematic adaptation of physics-informed Neural ODEs to stiff chemical kinetics, enabling interpretable and robust end-to-end rate constant inversion grounded in mechanistic chemical principles.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Search and Optimization: Learning to Search
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Estimating rate constants from complex chemical reactions is essential for advancing detailed chemistry. However, the stiffness inherent in real-world atmospheric chemistry systems poses severe challenges, leading to training instability and poor convergence that hinder effective rate constant estimation using learning-based approaches. To address this, we propose a Stiff Physics-Informed Neural ODE framework (SPIN-ODE) for chemical reaction modelling. Our method introduces a three-stage optimisation process: first, a latent neural ODE learns the continuous and differentiable trajectory between chemical concentrations and their time derivatives; second, an explicit Chemical Reaction Neural Network (CRNN) extracts the underlying rate coefficients based on the learned dynamics; and third, fine-tune CRNN using a neural ODE solver to further improve rate coefficient estimation. Extensive experiments on both synthetic and newly proposed real-world datasets validate the effectiveness and robustness of our approach. As the first work on stiff Neural ODEs for chemical rate coefficient discovery, our study opens promising directions for integrating neural networks with detailed chemistry.

Problem

• Estimating rate constants from complex chemical reactions
• Addressing stiffness in atmospheric chemistry systems
• Improving convergence in learning-based rate constant estimation

Innovation

• Three-stage optimization for stiff chemical systems
• Latent neural ODE learns differentiable trajectories
• CRNN extracts rate coefficients from dynamics