cc 16(10): e2

Research Article

Dynamic State Space Partitioning for Adaptive Simulation Algorithms

Download235 downloads
  • @ARTICLE{10.4108/eai.14-12-2015.2262710,
        author={Tobias Helms and Steffen Mentel and Adelinde Uhrmacher},
        title={Dynamic State Space Partitioning for Adaptive Simulation Algorithms},
        journal={EAI Endorsed Transactions on Collaborative Computing},
        volume={2},
        number={10},
        publisher={ACM},
        journal_a={CC},
        year={2016},
        month={1},
        keywords={adaptive algorithms, reinforcement learning, component-based simulation software, dynamic state space representations},
        doi={10.4108/eai.14-12-2015.2262710}
    }
    
  • Tobias Helms
    Steffen Mentel
    Adelinde Uhrmacher
    Year: 2016
    Dynamic State Space Partitioning for Adaptive Simulation Algorithms
    CC
    EAI
    DOI: 10.4108/eai.14-12-2015.2262710
Tobias Helms1,*, Steffen Mentel1, Adelinde Uhrmacher1
  • 1: University of Rostock
*Contact email: tobias.helms@uni-rostock.de

Abstract

Adaptive simulation algorithms can automatically change their configuration during runtime to adapt to changing computational demands of a simulation, e.g., triggered by a changing number of model entities or the execution of a rare event. These algorithms can improve the performance of simulations. They can also reduce the configuration effort of the user. By using such algorithms with machine learning techniques, the advantages come with a cost, i.e., the algorithm needs time to learn good adaptation policies and it must be equipped with the ability to observe its environment. An important challenge is to partition the observations to suitable macro states to improve the effectiveness and efficiency of the learning algorithm. Typically, aggregation algorithms, e.g., the adaptive vector quantization algorithm (AVQ), that dynamically partition the state space during runtime are preferred here. In this paper, we integrate the AVQ into an adaptive simulation algorithm.