VesNet: Neural network accelerated solver for simulating Stokesian vesicle suspensions

2026-06-24

AI Summary

Simulating suspensions of deformable vesicles in Stokes flow is computationally demanding due to strong nonlinear fluid–structure coupling, interfacial evolution, and multiscale effects. This work proposes VesNet, a hybrid framework that introduces neural networks into two-dimensional vesicle simulations for the first time, efficiently approximating vesicle self-interactions—including coupling with background flow and short-range lubrication forces—while retaining conventional methods for boundary reparameterization and far-field hydrodynamics. The approach achieves substantial computational speedups without compromising physical accuracy: it outperforms a multithreaded MATLAB CPU solver by over 100× and accelerates computations approximately 5× compared to its GPU counterpart. VesNet successfully reproduces the dynamics of single and double vesicles and captures collective behaviors of thousands of vesicles in Taylor–Green and Poiseuille flows.

Abstract

Numerical simulation of deformable particle suspensions in Stokes flow is computationally expensive due to nonlinear fluid-structure interactions, evolving interfaces, and multiscale hydrodynamics. We present VesNet, a hybrid framework that accelerates two-dimensional vesicle suspension simulations by approximating vesicle self interactions, including background flow coupling and short-range lubrication forces, while retaining conventional modules for boundary reparameterization and far-field hydrodynamics. A GPU-accelerated implementation achieves over 100x speedup compared to a multithreaded MATLAB CPU boundary integral solver and about 5x relative to its GPU counterpart. VesNet accurately captures key dynamics, including single-vesicle phase behavior, pair interactions, and large-scale suspensions in Taylor-Green and Poiseuille flows, enabling efficient simulations of thousands of vesicles on modest computational resources.

Problem

• Stokes flow
• vesicle suspension
• fluid-structure interaction
• computational cost
• deformable particles

Innovation

• VesNet
• neural network acceleration
• Stokesian vesicle suspensions
• GPU-accelerated simulation
• hybrid numerical framework