Abstract for:Executing system dynamics models on GPU

The parameter optimization task is very useful and popular use case of system dynamics model usage among researches. Solving of this kind of task can be used in many different areas for decision making and improving efficient of production. However, solving of this task requires many computation resources. For example, when simple model with small count of input parameters was used for parameter optimization task can produce many combinations, more than 10’000. Each of those input parameters group should be used for executing model. Executing model on this set of input data can consumption many computation resources of local environment. In general, resources of local environment have restriction and limitations. One of a possible ways to solve issue with computation resources – use cloud-based environments, like sdCloud. However, this way is not solve issue with resources, just relocate issue to another side. Answer for this issue is using another computation node that specially designed for computing large blocks of data in parallel. CPU is not well designed for computing large blocks of data. So we should looks to other devices, like GPU. In this paper we are describes, why GPU is a better device to improve performance of parameters optimization