parallant

Compilation

make all

Running

There is a script that runs algorithm_seq and algorithm_tiled versions with both available models: model_langton and model_forams versions:

./test.sh Width Height Depth PopulationSize Iterations

or simply:

make test - to run with default parameters (20, 10, 1, 10, 100)

You can run also another configuration using erlang shell:

1> parallant:start(Width, Height, Depth, PopulationSize, Steps, Options)

where:

• Width :: pos_integer() - width of the world
• Height :: pos_integer() - height of the world
• Depth :: pos_integer() - depth of the world
• PopulationSize :: pos_integer() < Width*Height*Depth - how many agents
• Steps :: pos_integer() - how many simulation steps
• Options :: proplists:proplist() - e.g. [{model, model_langton}, {algorithm, algorithm_seq}, {log, false}] - Run the simulation of a sequential algorithm with Langton's Ant model and don't log the state of every simulation step. The full list of possible options is described below.
  • {algorithm, Algorithm} :: algorithm_seq | algorithm_tiled
  • {model, Model} :: model_langton | model_forams - model_langton represents Langton's Ant simulation whereas model_forams is a simple algae dispersion in a grid simulation
  • {agents, AgentsImpl} :: agents_lists | agents_gbtree | agents_ets - how agents (cells) are stored (using lists, a tree from gbtree module or an ETS table).
  • {log, Log} :: true | false - the simulation state of every step is displayed if true (only the first and the last if false)
  • {animate, Animate} :: true | false - if true every logged step overwrites the previous one (might be not supported by the shell)

Writing your own model

In order to run your own model, you should provide a module implementing a behaviour model. Therefore, you need to implement three callback functions:

• initial_population/3 - returns initial population in a form of a list of tuples of position and desired state
• move/3 - takes a position of a current agent and current environment and returns updated environment
• get_agent_char/2 - takes an agent state and returns a character that represents this agent on the displayed grid.

Current agents' states can be fetched and updated by means of functions from the agents module:

• agents:get_agent/3 - returns current agent state on given position or empty if there is no agent on this position
• agents:update_agent/4 - updates given position with given agent state; pass empty as the agent_state to delete the agent; returns updated environment

Examples of this behaviour: model_langton, model_forams

In order to run it, you should provide the name of your module in the Options proplist. e.g. [{model, my_custom_model}].