Applications of High Performance Algorithms to Large Scale Cellular Automata Frameworks Used in Pharmaceutical Modelling
Marija Bezbradica, Martin Crane and Heather J. Ruskin
Cellular automata (CA) have been gaining momentum as a promising tool in the field of in silico Drug Dissolution System (DDS) modelling due to their bottom-up approach and the ability to simulate broad range of physico-chemical reactions at small domain level. In pharmaceutical applications, CA models use a combination of discrete-event rules, probabilistic distributions and fundamental physical laws to predict the structural behaviour of DDS over time. Practical algorithmic solutions in this field are lacking in the context of large-scale, high-precision, high-fidelity simulations. We consider different parallelisation strategies for a CA framework used in modelling coated drug spheres. Here, one of the main factors influencing drug dissolution is sensitivity to the physical coating thickness, orders of magnitude smaller compared to the core sphere of drug, requiring large 3D models to accommodate the measurement range. We analyse individual and hybrid applicability of two APIs, OpenMP and MPI, to the problem. Speed-up potential and communication overhead are explored. The findings have important implications, enabling incorporation of diverse interactions at small scale within the global framework, while allowing for solution scalability.
Keywords: Parallelisation strategies; hybrid models; scatter-gather algorithms; complex drug dissolution systems.