CODEWITHSUNDEEP
pythonmodelgridmesa
Clark
Clark

Reputation: 73

Create irregular shaped cellular automata space in Python Mesa

The examples within Mesa's readthedocs and in their Github all have fixed m x n grids. I'm trying to create an irregularly shaped agent space, for example:

3 x 3, connected to another 3 x 3 via a 1 x 1 "tunnel" (almost like a dumbbell shape). I'm happy to explicitly feed a series of tuples with grid coordinates but can't tell if that's possible. As a work around, I presently have agents in the system which don't interact, serving as a wall of sorts. In Mesa's Schelling example here's where the grid is formed:

class Schelling(Model):
    """
    Model class for the Schelling segregation model.
    """

    def __init__(self, width=20, height=20, density=0.8, minority_pc=0.2, homophily=3):
        """ """

        self.width = width
        self.height = height
        self.density = density
        self.minority_pc = minority_pc
        self.homophily = homophily

        self.schedule = RandomActivation(self)
        self.grid = SingleGrid(width, height, torus=True)

Any insight is appreciated. At a loss but I'd have to think this is possible.

Upvotes: 0

Views: 377

Answers (1)

Clark
Clark

Reputation: 73

Figured it out and answering my own question for the knowledge base.

It would appear that explicitly turning off cells is not built into Mesa's functionality. However, I produced a workaround for shaping, but it will only work with Mesa's SingleGrid which only allows one agent per grid location.

First, create an agent that doesn't interact or move:

class WallAgent(Agent):
    """
    Agent that acts as a wall. No interaction with other agents
    """

    def __init__(self, pos, model, agent_type):
        super().__init__(pos, model)
        self.pos = pos
        self.type = agent_type

Then, when placing agents within Mesa's model class you dynamically or explicitly state where to assign the wall agents:

       wall = [(0,0),(1,0),(2,0),(3,0),(4,0),(5,0),(6,0),(7,0),(8,0),(11,0),
            (12,0),(13,0),(14,0),(15,0),(16,0),(17,0),(18,0),(19,0),(0,1),(1,1),
            (2,1),(3,1),(4,1),(5,1),(6,1),(7,1),(8,1),(11,1),(12,1),(13,1),(14,1),
            (15,1),(16,1),(17,1),(18,1),(19,1)]
            
        for pos in wall:
            agent_type = 'wall'
            agent = WallAgent(pos, self, agent_type)
            self.grid.position_agent(agent, pos[0], pos[1])
            self.schedule.add(agent)

If using Mesa's visualization suite you'll get a result such as this where the black dots represent the wall/shaping. The blue and orange agents are ones that move and interact (for the problem I'm currently researching).enter image description here

Upvotes: 3

Related Questions