Fast Simulation Without Randomness: A Simulation Tool Combining Proxels and Discrete Phases

The simulation of discrete stochastic models using state space-based methods has recently become practical through the proxel-based simulation algorithm. Proxels implement the method of supplementary variables and generate a DTMC of the model’s state space. However, due to state space explosion the method does not yet work efficiently on larger models. By using discrete phase-type approximations, which are also DTMCs, for some distributions the state space can be reduced significantly and larger models can be simulated with comparable accuracy. The tool presented here enables a user to selectively replace general distribution functions by phase approximations and simulate the resulting model. This is one step further towards a general purpose proxel and phase-based simulator.