Parallel Execution Of Simulations Research Articles

An Open-Source Many-Scenario Approach for Power System Dynamic Simulation on HPC Clusters

In this paper we introduce an approach to accelerate many-scenario (i.e., hundreds to thousands) power system simulations which is based on a highly scalable and flexible open-source software environment. In this approach, the parallel execution of simulations follows the single program, multiple data (SPMD) paradigm, where the dynamic simulation program is executed in parallel and takes different inputs to generate different scenarios. The power system is modeled using an existing Modelica library and compiled to a simulation executable using the OpenModelica Compiler. Furthermore, the parallel simulation is performed with the aid of a message-passing interface (MPI) and the approach includes dynamic workload balancing. Finally, benchmarks with the simulation environment are performed on high-performance computing (HPC) clusters with four test cases. The results show high scalability and a considerable parallel speedup of the proposed approach in the simulation of all scenarios.

A framework for hybrid simulation with online model updating suitable for hard real‐time computing

The hybrid simulation method is used to test one or some components of a prototype structure subjected to a plausible loading history, accounting for their interaction with the untested ones, which are simulated numerically. If tested components have similar numerical counterparts, a possible approach to reduce simulation errors is to update the parameters of numerical substructures based on tested physical substructures. For this reason, online parameter estimation has gained the attention of the hybrid simulation community in the last decade. The term online indicates that the parameters of the identification model of the physical substructure are updated during the experiment. Main state-of-the-art middleware tools (e.g., OpenFresco and UI-SIMCOR) have been extended to support online model updating for the pseudodynamic hybrid simulation method. In this case, both numerical substructures and dynamic identification models are implemented on existing finite-element analysis software, which communicates with the middleware using a data exchange protocol with non-deterministic time schedule (e.g., TCP/IP). On the other hand, fast- and real-time hybrid simulation methods require a deterministic data exchange schedule between substructures, which imposes the adoption of hard real-time implementations. In this context, partitioned time integration is proposed to coordinate the parallel execution of simulation and model updating processes with heterogeneous sampling rates. As a result, the allocation of computational resources can leverage parallelization capabilities of multi-core CPUs.

A hierarchical composite framework of parallel discrete event simulation for modelling complex adaptive systems

As complex adaptive systems(CAS) continue to grow in scale and complexity, and the need for system adaptability increases, systems modelling has become an essential concern. Parallel discrete event simulation became a preferred choice as logical process world view, which bridges complex system modelling and high-performance computing. To resolve the shortcoming of this world view identified with respect to modularity and scalability. A hierarchical composite modelling framework was proposed, which is a three-level architecture intended to support the composition and integration of sub-models. The bottom layer is simulation model component(SMC), which is not a model but implement some simulation-specific support functionality. The middle layer is logical process model(LP), which describes an agent which can react to the current situation by executing a sequence of SMCs. The top layer is CAS system model, which defines a CAS model consist of several LPs and also the interactions between these LPs. The hierarchical composite modelling process and parallel simulation execution strategy are discussed to support the modelling and simulation of a CAS. In order to verify its effectiveness, a complex social opinion system model is proposed based on this hierarchical composite modelling framework. The experimental results confirms the viability of utilizing multi-level architecture for simulating large scale complex adaptive systems.

A high performance framework for modeling and simulation of large-scale complex systems

Due to the quick advances in the scale of problem domain of complex systems under investigation, the complexity of multi-input component models used to construct logical processes (LP) has significantly increased. High-performance computing technologies have therefore been extensively used to enable parallel simulation execution. However, the traditional multi-process parallel method (MPM) executes LPs in parallel on multi-core platforms, which ignores the intrinsic parallel capabilities of multi-input component models. In this study, a vectorized component model (VCM) framework has been proposed. The design aims to better utilize the parallelism of multi-input component models. A two-level composite parallel method (CPM) has then been constructed within the framework, which can sustain complex system simulation applications consisting of multi-input component models. CPM first employs MPM to dispatch LPs onto a multi-core computing platform. It then maps VCMs to the multiple-core platform for parallel execution. Experimental results indicate that (1) the proposed VCM framework can better utilize the parallelism of multi-input component models, and (2) CPM can significantly improve the performance comparing to the traditional MPM. The results also show that CPM can effectively cope with the size and complexity of complex simulation applications with multi-input component models.

Massive exploration of perturbed conditions of the blood coagulation cascade through GPU parallelization.

The introduction of general-purpose Graphics Processing Units (GPUs) is boosting scientific applications in Bioinformatics, Systems Biology, and Computational Biology. In these fields, the use of high-performance computing solutions is motivated by the need of performing large numbers of in silico analysis to study the behavior of biological systems in different conditions, which necessitate a computing power that usually overtakes the capability of standard desktop computers. In this work we present coagSODA, a CUDA-powered computational tool that was purposely developed for the analysis of a large mechanistic model of the blood coagulation cascade (BCC), defined according to both mass-action kinetics and Hill functions. coagSODA allows the execution of parallel simulations of the dynamics of the BCC by automatically deriving the system of ordinary differential equations and then exploiting the numerical integration algorithm LSODA. We present the biological results achieved with a massive exploration of perturbed conditions of the BCC, carried out with one-dimensional and bi-dimensional parameter sweep analysis, and show that GPU-accelerated parallel simulations of this model can increase the computational performances up to a 181× speedup compared to the corresponding sequential simulations.

New Developments on Generalized Simulated Annealing Applied to ab-initio Protein Structure Prediction

Proteins are the building blocks of cells and the executioners of nearly all cellular functions. Their activity directly depends on their specific three dimensional structure, determined by the folding of its amino acid chain. The folding process ultimately creates a stable structure balancing internal contacts between amino acids and their occlusion to create the protein surface and the hydrophobic core. In this work we explore a new design for applying Generalized Simulated Annealing (GSA) on protein structure prediction, based on previous software developed by our group. The GSA is a stochastic search algorithm employed in energy minimization and used in global optimization problems, such as gravity models, fitting of numerical data and conformation optimization of small molecules. The software deploys a new way of updating the protein structure at each step of the simulation, a different potential energy calculation function based on NAMD and parallel execution of simulations, granting a new take on ab-initio protein structure prediction. The design of the software also allows for the inclusion of data derived from large scale analysis of protein structures from the PDB, as the Solvation Free Energy, allowing us to use information already gathered by experimental structure determinations. We present results on the 20 residue trp-cage mini protein and mastoparan-X, a 13 amino acid peptide. Both chains fold with RMSD of 0,2 nm and 0,1 nm respectively after 10000 GSA steps and a molecular dinamics optimization of 1 ms for trp-cage and 200 ns for mastoparan-X. Structure prediction softwares allow us to study protein structures that cannot be experimentally determined, by using data on chemical bonds, non-bonded interactions and protein solvation. Once approximately predicted, the three dimensional structure can be refined by molecular dynamics simulations.

Fine-Grained Parallel and Distributed Spatial Stochastic Simulation of Biological Reactions

Fine-Grained Parallel and Distributed Spatial Stochastic Simulation of Biological Reactions

Sim-PETEK: A Parallel Simulation Execution Framework for Grid Environments

In this paper we introduce a framework for parallel and distributed execution of simulations (Sim-PETEK), a middleware for minimizing the total run time of batch runs and Monte Carlo trials. Sim-PETEK proposes a generic solution for applications in the simulation domain, which improves on our previous work done to parallelize simulation runs in a single node, multiple central processing unit (CPU) setting. Our new framework aims at managing a heterogeneous computational resource pool consisting of multiple CPU nodes distributed on a potentially geographically dispersed network, through a service-oriented middleware layer that is compliant to Web Services Resource Framework standard, thereby providing a scalable and flexible architecture for simulation software developers. What differentiates Sim-PETEK from a general-purpose, Grid-based job-distribution middleware is a number of simulation-specific aspects regarding the specification, distribution, monitoring, result collection and aggregation of simulation runs. These aspects are prevalent in the structure of the messages and in the protocol of interaction both among the constituent services of the framework and within the interfaces exposed to the external clients.

Performance estimation for parallel discrete event simulation using a synchronous conservative algorithm

When parallel discrete event simulation is applied to actual problems, the virtual time synchronization algorithm which has been studied for a long period and mapping the simulation model onto a parallel computer pose serious problems. For these problems, since accurate performance measurements in the execution of parallel simulation are not available, few effective solutions have been proposed. If we can estimate the performance of a parallel simulator precisely, then virtual time synchronization algorithms can be implemented and mapping can be done efficiently. This paper studies the synchronous conservative algorithm, in which parallel simulation is performed using barrier synchronizations and a mathematical model is constructed by assuming that complicated events are generated with intervals that follow an exponential distribution. As a result of comparing the estimated execution time obtained in this model with the execution time of a parallel computer simulation, we found that the execution time can be estimated with high accuracy. Using this estimation model, the overhead of the algorithm is analyzed and important points concerning algorithm implementation and mapping are discussed. © 1998 Scripta Technica, Syst Comp Jpn, 28(10): 40–49, 1997

Dynamic load balancing of a multi-cluster simulator on a network of workstations

Performance of Time Warp simulation systems are often measured on exclusively available parallel computing resources. In distributed systems exclusive use is normally not feasible. Instead, due to the multi-tasking operating systems, many users share the workstations and their availability for parallel simulation purposes varies extensively. Time Warp has been found to be very sensitive to variations in available processing power. This paper presents two methods for a Time Warp VLSI simulation system to reduce the negative effect of a non-ideal environment on the execution of parallel simulations. A dynamic load balancing algorithm which adapts to the change of available processing power is presented. This mechanism, together with a multi-cluster partitioning technique significantly improve the performance of Time Warp based simulation systems on heterogeneous computing resource.