Multi-scale Deep Neural Network (MscaleDNN) for Solving Poisson-Boltzmann Equation in Complex Domains
This addresses computational challenges in simulating electrostatics for biophysics and chemistry, offering an incremental improvement in efficiency for domain-specific applications.
The paper tackled solving Poisson-Boltzmann equations in complex domains by proposing multi-scale deep neural networks (MscaleDNNs), which achieved fast uniform convergence and outperformed traditional fully connected DNNs as a mesh-less numerical method.
In this paper, we propose multi-scale deep neural networks (MscaleDNNs) using the idea of radial scaling in frequency domain and activation functions with compact support. The radial scaling converts the problem of approximation of high frequency contents of PDEs' solutions to a problem of learning about lower frequency functions, and the compact support activation functions facilitate the separation of frequency contents of the target function to be approximated by corresponding DNNs. As a result, the MscaleDNNs achieve fast uniform convergence over multiple scales. The proposed MscaleDNNs are shown to be superior to traditional fully connected DNNs and be an effective mesh-less numerical method for Poisson-Boltzmann equations with ample frequency contents over complex and singular domains.