Publications

Learning mesh-free discrete differential operators with self-supervised graph neural networks

Calculation of kernel weights is costly for weights with polynomial consistency, necessary for accurate simulations. Here we introduce a new method for calculating discrete weights using graph neural networks.

L. Gerken Starepravo, G. Fourtakas, S. J. Lind, A. B. Harish, T. Tang and J. R. C. King

CMAME (2026)

Exploring neural network surrogates for high-order mesh-free interpolants

Calculation of kernel weights is costly for weights with polynomial consistency, necessary for accurate simulations. Can we design a neural network surrogate which bridges a gap between low- and high- order mesh-free interpolants?

L. Gerken Starepravo, G. Fourtakas, S. J. Lind, A. B. Harish and J. R. C. King

CPM (2026)

Compact LABFM - a framework for meshless methods with spectral-like resolving power

Resolving power is a key measure of the performance of a numerical method, with improved resolving power allowing more accurate simulations for a given cost. In this work we develop another framework for improving the resolving power of mesh-free methods.

H. Broadley, J. R. C. King and S. J. Lind

J. Comput. Phys. (2026)

Improving the accuracy of meshless methods via resolving power optimisation using multiple kernels

Resolving power is a key measure of the performance of a numerical method, with improved resolving power allowing more accurate simulations for a given cost. In this work we develop a framework for improving the resolving power of mesh-free methods.

H. Broadley, J. R. C. King and S. J. Lind

J. Comput. Phys. (2026)

Elasto-inertial transitions in viscoelastic flows through cylinder arrays

Cylinder arrays make a great prototype for porous media and in this paper we explore how we might trigger elasto-inertial turbulence to improve mixing or increase heat dissipation.

J. R. C. King, H. Broadley, and M. Beneitez

arXiv (2026)

Subyield Dynamics in Yield-Stress Materials

The paper shows that, once residual slip is accounted for, yield-stress materials below yield do not undergo plastic flow, but instead exhibit bounded nonlinear viscoelastic dynamics, challenging models that allow pre-yield flow and motivating improved constitutive descriptions.

A. Woodbridge, K. Amini, F. Lundell, O. Tammisola, A. Juel, R. J. Poole, and C. P. Fonte

Phys. Rev. Lett. (2026)

Network modelling of yield-stress fluid flow in randomly disordered porous media

This paper develops and validates a physics-based pore-network model for Herschel-Bulkley yield-stress fluids in disordered porous media, showing that near-yield transport is controlled by constriction-scale throat geometry and that wall slip lowers pressure losses by reopening otherwise blocked pathways.

C. P. Fonte, E. Sutton, K. Ohie, E. Doman, Y. Tasaka, and A. Juel

Appl. Phys. Lett. (2026)