Open source Differentiable ODE Solving Infrastructure
This work provides an incremental improvement by making existing ODE solving methods more accessible and efficient for researchers in fields like physics, chemistry, and biology.
The authors integrated GPU-accelerated, differentiable ODE solvers into the open-source DeepChem framework to make these tools accessible for modeling dynamic systems, achieving high accuracy with mean squared errors from 10^-4 to 10^-6 and scalability for systems with up to 100 compartments.
Ordinary Differential Equations (ODEs) are widely used in physics, chemistry, and biology to model dynamic systems, including reaction kinetics, population dynamics, and biological processes. In this work, we integrate GPU-accelerated ODE solvers into the open-source DeepChem framework, making these tools easily accessible. These solvers support multiple numerical methods and are fully differentiable, enabling easy integration into more complex differentiable programs. We demonstrate the capabilities of our implementation through experiments on Lotka-Volterra predator-prey dynamics, pharmacokinetic compartment models, neural ODEs, and solving PDEs using reaction-diffusion equations. Our solvers achieved high accuracy with mean squared errors ranging from $10^{-4}$ to $10^{-6}$ and showed scalability in solving large systems with up to 100 compartments.