Improving ParModAuto for Equation Based Mathematical Modeling & Simulation: Metaheuristic Approach

Despite their widespread use, modeling and simulation tools still face limitations in scalability, particularly when executed in distributed computing environments. In our prior research, we introduced a task-based library named ParModAuto, designed to enable the automatic parallelization of Modelica simulation models by supporting efficient representation, clustering, scheduling, profiling, and execution of complex, dependency-heavy equation systems, which employs heuristic-driven algorithms to address computationally challenging, NP-complete optimization problems. This paper investigates whether further enhancements to the parallelization library can lead to greater speedups in simulation performance, as benchmark results indicate that additional gains are achievable.