[image of digits]
Embarcadero C++ Builder - Segregation & Assimilation Patterns

SEGREGATION & ASSIMILATION MODELS

Agents (cells) move to a new location (segregation) or change their identity (assimilation) based upon their local preferences for neighbors.
There are four types of cells (agents): cyan, magenta, yellow and black (empty space).No agent may displace another agent.
The global patterns that you see below are the result.

Unless otherwise noted:
All simulations begin with a randomly distributed population of equal quantities of cyan, magenta, yellow and black cells (agents).
All simulations run with one single unchanging set of rules. Changing the rules during the run (adding historic events) will complexify the patterns considerably.

SIGNIFICANCE

We study segregation and assimilation models because they mirror social processes with which we are familiar.
They are among the simplest models of complex systems while still providing compelling visualizations of emergence.
Emergence is a term used to describe the process by which local rules produce global patterns of behavior.

Each global pattern that you see below were produced by one specific set of local rules.

Give the global patterns, can you infer the local rules that produced them? Can you explain how they came into being?
There is no mathematics, no algebra, no calculus, no statistics that can tell us how this happened (retrodiction).

Given the local rules, can you deduce the global patterns that they will produce?
Again, there is no mathematics, no algebra, no calculus, no statistics that can tell us what will happen (prediction).
However, we can build a multiagent (multicausal) model and run it to completion to find the answer.
There is no shortcut to answering the question "What if?".

Understanding how the world works requires looking beneath the surface of things to understand how they came to be the way they are.
It requires teasing out the local rules that produced the global behavior we observe.
It requires searching for the individual processes that produce the aggregate pattern.
We speak of levels of emergence, along a bottom-up hierarchy from the individual, to the family, the group, to the Nation and the globe.
Some influences may be top-down and categories of emergence may cross-cut one another in a heterarchy, but let's keep in simple.
In these investigations one often hears the admonition "KISS."
(keep it simple stupid)

Assimilation Models

All assimilation simulations were run to 500 iterations.

All assimilation simulations were run to 500 iterations.
Segregation - Preference or Prejudice?

These simulations were run until they stabilized.

These simulations were run until they stabilized.
Segregation - Preferences of increasing complexity: Attending to your own, to one other, and to combinations of others.

These simulations were run until they stabilized.
Segregation - Attending to empty space.

This simulation was run until it stabilized with each agent desiring one neighboring empty cell.
Segregation - Conflicting preferences.

This simulation was run until it stabilized.
Segregation - Temporal delay constraints on different types.

These simulations were run until they stabilized.
Segregation - Spatial mobility constraints on different types.

This simulation was run until it stabilized.
Segregation - CROWDS, COPS & AGITATORS
Inspired by Rob Axtell
Crowds (yellow) influenced by aggitators (magenta) and police (cyan).
Aggitators and police move randomly. Members of the crowd run away.
Left panel or diagonal: Mappings of who is where.
Right panel or diagonal: Mappings of crowd levels of calm (green) and excitement (orange).

These simulations were run until they stabilized.

Initialized with one large crowd with a few cops and few aggitators inserted in two different locations.
Members of the crowd avoid both the cops and aggitators.

This simulation was run until it stabilized.

Initialized with one large crowd with a few cops and few aggitators circling them.
Members of the crowd avoid both the cops and aggitators.
This simulation was run until it stabilized.

Initialized with one large crowd with a few cops and few aggitators inserted in two different locations.
Members of the crowd avoid everyone.
This simulation was run until it stabilized.

Initialized with one large crowd with a few cops and few aggitators circling them.
Members of the crowd avoid everyone.
This simulation was run until it stabilized.