Xinuk

Xinuk is an implementation of a highly scalable agent-based spatial simulation framework, written in Scala. It uses sbt build tool.

Models and scenarios

This repository contains several subprojects. xinuk-core contains the core of the framework, while the remaining subprojects contain implementation of a different simulation model and scenario:

• mock: randomly moving agents, an implementation serving as an introduction to the framework
• fortwist: habitat of benthic creatures feeding on the algae growing on the seabed
• rabbits: predator-prey scenario of stationary, rapidly growing lettuce and mobile rabbits feeding on the lettuce
• torch: simplified scenario of fire emergency evacuation
• urban: people navigating urban environment, with optional rules of proximity avoidance

Each implemented scenario follows a similar structure. A Main object (e.g. MockMain) can be found in the main project package, along with the reference.conf file in the corresponding resources directory. These files provide all the necessary tools to run the simulation, e.g. using

sbt mock/run

If necessary, the values can be edited directly in the corresponding .conf file or overridden with commandline argument, by adding -Dpath.in.config.file=newValue, e.g. -Dmock.config.mockInitialSignal=2 (see mock config file).

Distributed execution

The framework supports execution in distributed environment:

• exactly one instance of the framework must be manually elected as supervisor by setting <model-name>.config.isSupervisor=true,
• each instance must be provided with its own and supervisor's host by setting clustering.ip and clustering.supervisor.ip (and must be able to reach each other),
• each instance must be provided with the desired total number of instances by setting akka.cluster.min-nr-of-members,
• all other simulation parameters should be identical for all instances.

Related publications

Bujas, J., Dworak, D., Turek, W., & Byrski, A. (2019).
High-performance computing framework with desynchronized information propagation for large-scale simulations.

Renc, P., Bielech, M., Pęcak, T., Morawiecki, P., Paciorek, M., Turek, W., Byrski, A., & Wąs, J. (2020).
HPC large-scale pedestrian simulation based on proxemics rules.

Paciorek, M., Bogacz, A., & Turek, W. (2020).
Scalable signal-based simulation of autonomous beings in complex environments.

Paciorek, M., Bujas, J., Dworak, D., Turek, W., & Byrski, A. (2021).
Validation of signal propagation modeling for highly scalable simulations.

Paciorek, M., & Turek, W. (2021).
Agent-based modeling of social phenomena for high performance distributed simulations.

Paciorek, M., Poklewski-Koziełł, D., Racoń-Leja, K., Byrski, A., Gyurkovich, M., & Turek, W. (2021).
Microscopic simulation of pedestrian traffic in urban environment under epidemic conditions.