Development and application of real-time and interactive software for complex system

Abstract

Soft materials have attracted considerable interest in recent years for predicting the characteristics of phase separation and self-assembly in nanoscale structures. A popular method for demonstrating and simulating the dynamic behaviour of particles (e.g. particle tracking) and to consider effects of simulation parameters is cell dynamic simulation (CDS). This is a cellular computerisation technique that can be used to investigate different aspects of morphological topographies of soft material systems. The acquisition of quantitative data from particles is a critical requirement in order to obtain a better understanding and of characterising their dynamic behaviour. To achieve this objective particle tracking methods considering quantitative data and focusing on different properties and components of particles is essential. Despite the availability of various types of particle tracking used in experimental work, there is no method available to consider uniform computational data. In order to achieve accurate and efficient computational results for cell dynamic simulation method and particle tracking, two factors are essential: computing/calculating time-scale and simulation system size. Consequently, finding available computing algorithms and resources such as sequential algorithm for implementing a complex technique and achieving precise results is critical and rather expensive. Therefore, it is highly desirable to consider a parallel algorithm and programming model to solve time-consuming and massive computational processing issues. Hence, the gaps between the experimental and computational works and solving time consuming for expensive computational calculations need to be filled in order to investigate a uniform computational technique for particle tracking and significant enhancements in speed and execution times.
The work presented in this thesis details a new particle tracking method for integrating diblock copolymers in the form of spheres with a shear flow and a novel designed GPU-based parallel acceleration approach to cell dynamic simulation (CDS). In addition, the evaluation of parallel models and architectures (CPUs and GPUs) utilising the mixtures of application program interface, OpenMP and programming model, CUDA were developed. Finally, this study presents the performance enhancements achieved with GPU-CUDA of approximately ~2 times faster than multi-threading implementation and 13~14 times quicker than optimised sequential processing for the CDS computations/workloads respectively.