Deep learning approaches to surrogates for solving the diffusion equation for mechanistic real-world simulations

TL;DR

This paper demonstrates that convolutional neural networks can efficiently approximate solutions to the diffusion equation, significantly speeding up complex simulations in biological and physical models, with customizable accuracy measures.

Contribution

It introduces a CNN-based surrogate model for the stationary diffusion equation and discusses training strategies and accuracy metrics for practical application.

Findings

Neural network surrogate is about 1000 times faster than direct calculation.

Training with roll-back improves convergence.

Various loss functions help tailor accuracy to specific needs.

Abstract

In many mechanistic medical, biological, physical and engineered spatiotemporal dynamic models the numerical solution of partial differential equations (PDEs) can make simulations impractically slow. Biological models require the simultaneous calculation of the spatial variation of concentration of dozens of diffusing chemical species. Machine learning surrogates, neural networks trained to provide approximate solutions to such complicated numerical problems, can often provide speed-ups of several orders of magnitude compared to direct calculation. PDE surrogates enable use of larger models than are possible with direct calculation and can make including such simulations in real-time or near-real time workflows practical. Creating a surrogate requires running the direct calculation tens of thousands of times to generate training data and then training the neural network, both of which are computationally expensive. We use a Convolutional Neural Network to approximate the stationary solution to the diffusion equation in the case of two equal-diameter, circular, constant-value sources located at random positions in a two-dimensional square domain with absorbing boundary conditions. To improve convergence during training, we apply a training approach that uses roll-back to reject stochastic changes to the network that increase the loss function. The trained neural network approximation is about 1e3 times faster than the direct calculation for individual replicas. Because different applications will have different criteria for acceptable approximation accuracy, we discuss a variety of loss functions and accuracy estimators that can help select the best network for a particular application.