Hybrid Cloud-Edge Systems for Computational Physics: Enhancing Large-Scale Simulations Through Distributed Models

Abstract

The growing complexity of large-scale simulations in computational physics necessitates innovative computing paradigms that balance computational power, scalability, and efficiency. This study introduces a hybrid cloud-edge system architecture designed to optimize performance in distributed physics simulations. By integrating the high computational capacity of cloud infrastructure with the low-latency benefits of edge computing, the proposed system accelerates data processing and model execution for large-scale simulations. Experimental evaluations were conducted on a molecular dynamics simulation dataset, comprising over 10 billion particle interactions, distributed across 1,000 edge nodes and a centralized cloud platform. Results demonstrated a 45% reduction in processing time compared to cloud-only systems, with edge nodes handling 60% of preliminary computations locally, thereby minimizing data transmission overhead. Energy efficiency tests revealed a 28% reduction in power consumption due to intelligent workload partitioning between cloud and edge resources. Scalability assessments confirmed the architecture's ability to seamlessly support up to 50,000 concurrent simulation tasks without significant degradation in performance or latency. Additionally, the hybrid model improved fault tolerance, achieving a 96% recovery rate during node failures, ensuring reliability in critical computations. These findings underscore the potential of hybrid cloud-edge systems to revolutionize computational physics by enabling real-time, large-scale simulations while addressing energy and scalability challenges. The proposed architecture provides a robust framework for advancing distributed computing solutions in physics and other data-intensive scientific domains.