Features

Simple and intuitive API for simulators and control strategies:

• Only four types of calls from mosaik to a simulator:

  • init – Initialize the simulator.
  • create – Create model instances.
  • step – Perform a simulation step based on some input data for a time interval.
  • get_data – Get data from the simulator.

• Control strategies can make asynchronous requests to mosaik during a step:

  • get_progress – Get the current simulation progress in percent.
  • get_related_entities – Get a list of entities related to another entity (to allow traversing the simulated topology).
  • get_data – Retrieve data from another simulator.
  • set_data – Set input data for another simulator.

• Based on plain network sockets with JSON encoded messages.

• High-level API implementations that implement the network event loop and provide a simple base class that you can inherit from. Implementations are currently available for:

  • Python
  • Java
  • Matlab
  • C#

  Implementations for other languages will be added on demand.

Simple but flexible scenario creation:

• Scenario description in pure Python leveraging its simple syntax and great flexibility to support from a few up to tens of thousands entities.
• You just start a simulator and get a model factory. You use that factory to create model instances (entities). You connect the entities with others. You run simulation. Done.
• Connection entities is based on a single primitive called connect. It only connects two entities with each other but can be used as a building block to implement any connection strategy for sets of entities that you can imagine.

Management of external simulator processes:

• Simulators written in Python 3 can be executed in process.
• Simulators can be started as external processes.
• Mosaik can connect to an already running instance of a simulator.

Event-based simulation execution:

• Based on the established Asyncio simulation library.
• Simulators can have different and varying step sizes. They only need to agree on a common time resolution, which can be defined in the scenario.
• Simulators’ steps can be triggered based on available output of other simulators.
• Output data can be specified as persistent or non-persistent (i.e., transient).
• Simulators can request steps asynchronously to react to external events.
• When there’re no interactions, simulators can compute the state transformation of several fundamental time steps at once (max_advance).

Powerful ecosystem:

In our readthedocs we list all our own and external components we know with a short description. All components of the mosaik ecosystem provided by the mosaik team can be found at Gitlab.