Predicting material properties of soft condensed matter is often difficult because of its strong tendency to self-organise into mesoscopic structures. These structures are much larger than the molecular scale, but their morphology and interactions determine the overall behaviour of the material on the macroscopic scale. The modelling of flow poses a particular challenge and specific mesoscopic simulation methods like the lattice Boltzmann method have emerged which permit an accurate description of fluid-structure interactions.

After a brief introduction into the lattice Boltzmann method I will present recent results on novel composite materials that consist of nanoparticles embedded in a liquid crystalline host material. These systems have tunable optical and elastic properties and show promising potential for optical switch gear, multistable displays or e-papers [1].

A good performance on modern computing architectures is often indispensable to resolve the relevant time and length scales. Top supercomputers today use either many-core processors or graphical processing units (GPUs) to achieve their leading status, but are notoriously difficult to program. I will show how good performance can be achieved for grid-based applications by introducing a new abstraction layer called targetDP, which targets data parallel hardware in a performance-portable manner [2].

[1] K. Stratford, O. Henrich, J. Lintuvuori, M.E. Cates, D. Marenduzzo, Nat. Comm. 5, 3954 (2014).
[2] A. Gray, A. Hart, O. Henrich, K. Stratford, Int. J. High Perform. Comput. Appl. 29, 274 (2015).