Abstract
In the last few years Agent Based Models (ABMs) have attracted growing interest in the field of computational simulation thanks to their applicability in very heterogeneous landscapes, usability for fine-grained descriptions and comprehensibility for application domain experts. However, the lack of a well-defined semantics for specifying how agents behave and how they get coupled and scheduled may lead to inconsistent results. To fill this gap we proposed a well defined ABMs semantics that, using Extended Stochastic Symmetric Nets for model description, allows the modeller to automatically derive the corresponding ABM simulator that is directly executable in the NetLogo ABM framework. In the present paper we propose an improvement that exploits locality of state change effects to avoid recomputing the rates of the enabled events at each state change. This is achieved by exploiting structural properties of the ESSN model to generate optimized NetLogo code (semi)automatically. The results obtained for an example case-study demonstrate a relevant improvement in terms of execution time when structural optimizations are employed to reduce rates calculations.
Lingua originale | Inglese |
---|---|
Pagine | 353-368 |
Numero di pagine | 16 |
DOI | |
Stato di pubblicazione | Pubblicato - 2023 |
Evento | 19th European Workshop on Performance Engineering European Workshop on Computer Performance Engineering and Stochastic Modelling - EPEW 2023 and 27th International Conference on Analytical and Stochastic Modelling Techniques and Applications ASMTA 2023 - Firenze, Italy Durata: 1 gen 2023 → … |
???event.eventtypes.event.conference???
???event.eventtypes.event.conference??? | 19th European Workshop on Performance Engineering European Workshop on Computer Performance Engineering and Stochastic Modelling - EPEW 2023 and 27th International Conference on Analytical and Stochastic Modelling Techniques and Applications ASMTA 2023 |
---|---|
Città | Firenze, Italy |
Periodo | 1/01/23 → … |
Keywords
- Agent Based Modeling and Simulation