What is ABMing

1
What is ABMing?
ABMing is a computational methodology that allows
the analyst to create and analyze global emergent
complexity and self-organizing patterns as a
result of agents local behavioral interactions.
ABMs also known as individual or entity based
models
Origins in AI primary to develop Intelligent
Software
2
Prisoner's Dilemma
Wiki page link. Non-zero sum game. In the
prisoner's dilemma, you and your friend are
picked up by the police and interrogated in
separate cells without a chance to communicate
with each other. You are both told the same
thing
If you both confess, you will both get 4 years
in prison. If neither of you confesses, the
police will be able to pin part of the crime
on you, and you'll both get 2 years. If one of
you confesses but the other doesn't, the
confessor will make a deal with the police
and will go free while the other one goes to jail
for 5 years.
The objective is to minimize your jail time!!!
3
What Are Cellular Automata?

• Cellular automata simple models, which are used
  for studying complex systems behavior in
  different fields of science
• Automata are discrete dynamic systems, which work
  can be completely described in the terms of local
  interactions
• The Game of Life

4
The Game of Life

• The Rules
• For a space that is 'populated'
• Each cell with one or no neighbors dies, as if
  by loneliness. Each cell with four or more
  neighbors dies, as if by overpopulation.Each cell
  with two or three neighbors survives.
• For a space that is 'empty' or 'unpopulated
• Each cell with three neighbors becomes
  populated.
• The Game of Life - http//www.bitstorm.org/gameofl
  ife/
• Wiki page link.

5
The Schelling Tipping Game

• An interesting and important puzzle
• after 1964 housing discrimination was illegal
• since 1950 racial prejudice has declined
• yet neighborhoods remain highly segregated
• T. C. Schelling (1978) hypothesized that
  segregation
• does not need to be imposed (top-down)
• self-organizes through dynamic interaction

6
Schelling Tipping Game
Interior agent A up to 3 neighbors
Interior agent A up to 8 neighbors
Interior agent A up to 5 neighbors

• Each agent is happy (no need to move) if more
  than 1/3 of its neighbors are
• of same type.
• Boxes below give number of neighbors that must be
  same type for happiness
• given the total neighbors an agent has, from 0
  to 8.

Happiness Rule -
7
Starting Pattern for theSchelling Tipping Game
8
the Game!!!

• Given the pattern on previous slide, everyone is
• happy and no one moves. Now remove 10 randomly
  selected agents from the board.
• Starting from the top row, moving from left to
  right, row by row, check for unhappy agents.
• Every time you encounter an unhappy agent, if
  possible move him to a tolerable vacant square
  where he is happy otherwise remove him.
• Keep going until there are no unhappy agents left
  on the board. What degree of segregation does the
  resulting pattern display?

• Get your Nobel Prize!!!

9
Principles in devising an ABM

• In Principle, modeler
• constructs an environment
• or virtual world on a
• computer populated by
• various agent types.
• sets initial world conditions.
• observe how the world develops over time.

10
Steps in devising an ABM
Model type
Conceptual models investigating typical patterns
of interaction (e.g what determines
cooperation?). Realistic models portraying a
real, specific system are in general more
complex, must have parameters that can be
validated on real data (e.g. biological systems)
11
Steps in devising an ABM

• Not spatially explicit - agent location is
  irrelevant (computer network)
• Spatially explicit - agents are associated with a
  location in geometrical space.
• Situated vs. Mobile where agents can move.
• Continuous (real valued) vs. Discrete (integer
  valued, grid-like, grid-based) space.
• cellular automata - spatially-explicit/grid-bas
  ed/situated ABM.

12
Physical-spatial aspects

• Capturing the physical-spatial aspects of
  entity and system behavior in modeling dynamic
  biological systems is a Challenge!!!
• Equation-based approaches are generally unable to
  cope with spatio-temporal dynamics when the
  system is highly complex.
• Simplification may result in loss of accuracy.
• In biological systems, modeling in
  physical-spatial terms is more tractable.
• Majority of spatially explicit ABM in biology is
  in discrete space.

13
What is Agent?

• An agent is a physical or virtual entity
  which is situated in some environment and capable
  of
• acting in an environment
• communicate directly with other agents
• driven by a set of individual objectives
• perceiving its environment (but to a limited
  extent)
• possesses skills and can offer services
• reproduce itself

14
An agent can be

• Passive agents have no behavior of their own
  rather, any changes in their state are brought
  about by other components of the model.
• Active agents are those that have both state and
  a defined behavior.
• Reactive agents are those whose behavior is fully
  determined by the state of its environment.
• Pro-active agents are those whose behavior can be
  determined by its own state as well as the state
  of its environment.
• Non-adaptive agents have a determinate set of
  rules.
• Adaptive agents have rules that can change
  dynamically with changes in state (both the
  agent's and the environment's).

15
Swarm intelligence

• Collective system capable of accomplishing
  difficult tasks in dynamic and varied
  environments without any external guidance or
  control and with no central coordination
• Achieving a collective performance which could
  not normally be achieved by an individual acting
  alone

16
Adapted from http//www.scs.carleton.ca/arpwhite/
courses/95590Y/notes/SI20Lecture203.pdf
17
Adapted from http//www.scs.carleton.ca/arpwhite/
courses/95590Y/notes/SI20Lecture203.pdf
18
Adapted from http//www.scs.carleton.ca/arpwhite/
courses/95590Y/notes/SI20Lecture203.pdf
19
Immune System as Swarm intelligence - AIS
Modelling strategy
For each agent in turn.
step
Adjust position of all cells to ensure no overlap
The characteristics of each and of the world are
tracked through time vs. averaging!!!
20
Cell cycle control cell as an agent
G1 GROWTH PHASE
M
G2
G0
G1
S
21
Cell cycle control cell as an agent
G1 GROWTH PHASE
M
G2
G0
G1
S
22
Cell cycle control cell as an agent
G1-G0 checkpoint
M
G2
G0
G1
S
23
Cell cycle control cell as an agent
G0 QUIESCENT PHASE
M
G2
G0
G1
S
24
Cell cycle control cell as an agent
G1 GROWTH PHASE
M
G2
G0
G1
S
25
Cell cycle control cell as an agent
S PHASE (CHROMOSOME REPLICATION)
M
G2
G0
G1
S
26
Cell cycle control cell as an agent
G2 PHASE (HOUSEKEEPING)
M
G2
G0
G1
S
27
Cell cycle control cell as an agent
M
M PHASE - DIVISION
G2
G0
G1
S
28
ABM model of the DC infection
M PHASE - DIVISION
29
Net Logo
M PHASE - DIVISION
30
Ant Colony Optimization
ABMing - additional applications
31
ACO Concept

• Ants (blind) navigate from nest to food source
• Shortest path is discovered via pheromone trails
• each ant moves at random
• pheromone is deposited on path
• ants detect lead ants path, inclined to follow
• more pheromone on path increases probability of
  path being followed

32
Applications

• Efficiently Solves NP hard Problems
• Routing
• TSP (Traveling Salesman Problem)
• Vehicle Routing
• Sequential Ordering
• Assignment

33
Why ABM?

• a technique for theorizing
• that is designed to address complex real-world
  issues
• a practical approach to real-world issues
• that permits modeling tools to be adapted to the
  problem instead of having to adapt the problem
  to the tools
• and a fun way to explore real-world issues
• that permits creative experimentation with new
  ideas
• that encourages out of control programming that
  can surprise and inform

34
ABM limitations

• Requires detailed knowledge of behavior.
• May require considerable coding expertise to
  develop as well as considerable computer time to
  run.
• Typically requires many simulations to evaluate
  any particular situation as it is based upon an
  underlying stochastic model.

35
ABMs are Great!!!

• Tell everybody

Any Questions?