ABT, short for Agent-Based Template, is a self developed modular JAVA framework for high.performance and reproducible simulation of individual-based models. Thanks to its modular structure, it enables a wide variety of real complex systems to be modelled quickly and flexibly in silico; individuals can be humans but also vehicles or even viruses.
While the programmer has still complete freedom when implementing the model's logic, ABT offers a huge library of useful tools including a discrete event simulator, a random number generator, a state-variable recorder, input-output-routines, visualisation tools, and highly efficient methods for dealing with model-time. The availability of these features spare the modeller valuable time for focusing on what is really important: the model-specific program logic.
The ABT framework is currently used in the occupancy simulation of hospitals and in the delay prediction within railway networks as well as of the Corona-virus epidemic.
For the occupancy simulation of hospitals, patient flows are generated on the basis of historical data and forecasts. These patients go through certain processes in the hospital. This enables the estimation of capacities of individual wards and resources within the hospital. The results are used in the planning of new hospitals.
In order to predict delays within railway networks, the effects of certain initiatal delays on the entire timetable are simulated. For this purpose, the existing timetable is mapped and each section is assigned a certain probability of delay based on historical data. By limiting certain resources (locomotives, train paths), these delays spread within the network. The results of the simulations are used to evaluate locomotive circulations.
Authors: Matthias Rössler, Martin Bicher, Štefan Emrich