GPU Computing to Speed-Up the Resolution of Microrheology Models

Abstract

Active microrheology is a technique to obtain rheological properties in soft matter from the microscopic motion of colloidal tracers used as probes and subjected to external forces. This technique extends the measurement of the friction coefficient to the nonlinear-response regime of strongly driven probes. Active microrheology can be described starting from microscopic equations of motion for the whole system including both the host-fluid particles and the tracer. While the main observable is the effective friction coefficient with the bath, tracer position correlation functions describe the tracer motion, and reveal the underlying dynamics of the host bath. On the other hand, pulling the tracer provokes a non-linear non-affine strain field in the host bath, what requires a deep understanding of the dynamics of the system. Different theoretical approaches have been proposed to deal with this problem.

In this work, we present simulations of a tracer dragged by a constant force through a dense bath of hard colloids. The size of the system has been varied, keeping the bath density constant, approaching the hydrodynamic limit. In order to calculate the tracer’s position, iterative methods have to be used. These methods are computationally highly demanding, specially when the number of colloidal particles is high. Therefore, it is necessary to use HPC in order to develop and validate this kind of models. The present work shows the results of the considered microrheology method varying the number of colloidal tracers and using GPU computing in order to solve problems of interest.

This work has been partially supported by the Spanish Ministry of Science throughout projects TIN15-66680 and FIS-2015-69022-P and CAPAP-H5 network TIN2014-53522, by J. Andalucía through projects P12-TIC-301 and P11-TIC7176, and by the European Regional Development Fund (ERDF).