Академический Документы
Профессиональный Документы
Культура Документы
Review
IS 795/895
Fall 2009
Outline
Introduction to chaos and complexity
Why use agent-based models?
Advantages
Pitfalls
IS 795/895
Fall 2009
IS 795/895
Fall 2009
Chaos
What do you think of?
IS 795/895
Fall 2009
Chaos Theory
Theory of mathematics
Nonlinear systems
Simultaneous equations
IS 795/895
Fall 2009
Nonlinearity
Linear
a bx e
dx
dt
y x
dy
dt
x z
dz
dt
xy
Nonlinear
IS 795/895
z
Fall 2009
IS 795/895
Fall 2009
Mathematical chaos
Deterministic equations
Erratic behavior over time
Focus on dynamics
IS 795/895
Fall 2009
IS 795/895
Fall 2009
10
Attractors
Graphical technique to illustrate coherence of
chaotic systems
Types
Point
Limit cycle
Strange
IS 795/895
Fall 2009
11
Implications
Simple systems can produce chaos
Saperstein: the prediction of unpredictability
Rumsfeld:
As we know, there are known knowns; there are things we
know we know. We also know there are known unknowns;
that is to say we know there are some things we do not know.
But there are also unknown unknownsthe ones we dont
know we dont know.
IS 795/895
Fall 2009
12
Implications
Sensitivity to initial conditions
slight differences may lead to vastly different outcomes
IS 795/895
Fall 2009
13
Chaos Theory
Questions?
IS 795/895
Fall 2009
14
15
IS 795/895
Fall 2009
Notice . . .
Each example is:
Decentralized, leaderless
Massively parallel
Interdependent decision-making
Spatial
Predictably unpredictable
16
IS 795/895
Fall 2009
Organization
What do we mean?
An organization (entity)
Versus organization (state of being)
17
IS 795/895
Fall 2009
18
IS 795/895
Fall 2009
Self Organization
Waldrop 1992: matters incessant attempts to
organize itself into ever more complex structures,
even in the face of the incessant forces of
dissolution p. 102
Autopoiesis
19
IS 795/895
Fall 2009
Self Organization
Organization for free
No external guidance
No leadership
Contra Second Law
Fall 2009
20
Self Organization
Examples
IS 795/895
Fall 2009
21
Emergence
Definitions
Gilbert and Troitzsch
Pepinsky, pp. 373-4: those phenomena that appear at
an aggregate level, based not on specific micro-level
interactions of agents but rather on the complex and
often unpredicted effects of many such interactions
Fall 2009
22
Second-Order Emergence
Probably unique to social systems
Agents recognize emergent structures
Emergence thus affects agent behavior
Double hermeneutic (Giddens)
IS 795/895
Fall 2009
23
Implications
The phenomenon of emergence requires us to
model interactions across levels of analysis
Agents and structures
Mathematics is no help
IS 795/895
Fall 2009
24
IS 795/895
Fall 2009
25
IS 795/895
Fall 2009
26
Uses of ABM
Gilbert & Troitzschs seven uses
IS 795/895
Fall 2009
27
Space
Physical (GIS)
Network
IS 795/895
Fall 2009
28
Quasi-experimentation
Counter factuals (Fearon)
micro foundations
IS 795/895
Fall 2009
29
Rational
Choice
ABM
Assumptions
Interests
Behavior
Inferences
Behavior
Interests
Hoffmann
Lustick
IS 795/895
Fall 2009
30
ABM for IS
Core concerns of IR
Change through actor learning
Gilpin (1981): interaction, systemic and system change
Counterfactual reasoning
Fearon (1991)
IS 795/895
Fall 2009
31
IS examples
Pepinsky, p. 371
Political behavior
Arms races
Trade and economics
Ethnic conflict
Escalation
War and conquest
IS 795/895
Fall 2009
32
IS 795/895
Fall 2009
33
Logic of Simulation
Combination of induction and deduction
Deduction: formal specification of
Actors
Variables
Rules of behavior
Fall 2009
34
Epistemology
Underspecified and incomplete
Two issues
Explanation vs. Prediction
Point vs. Trend Prediction
IS 795/895
Fall 2009
35
IS 795/895
Fall 2009
36
IS 795/895
Fall 2009
37
Model
All agent-based models:
IS 795/895
Fall 2009
38
IS 795/895
Fall 2009
39
Outputs
Similar to dependent variables
Parameters
Not variables because we are not sampling
Data
Empirical: from the target
Meta: from the model
IS 795/895
Fall 2009
40
Epistemology
Questions?
IS 795/895
Fall 2009
41
Steps of Research
1. Puzzle
2. Definition of target
3. Observations of
target
4. Assumptions and
design the model
a.
b.
c.
d.
5.
6.
7.
8.
9.
Simulate
Record data
Verification
Validation
Sensitivity analysis
Environment
Agents
Parameters and rules
Time
IS 795/895
Fall 2009
42
Pitfalls (1)
Incorrect model specification
Agents: thick ontological presuppositions (Pepinsky)
In that there is little consensus about the nature of the
international system, simulation may not be an
acceptable research tool for this enterprise. p. 379
Importance of validation
IS 795/895
Fall 2009
43
Pitfalls (2)
Sensitivity of parameter specification: Chaos
Importance of verification and sensitivity analysis
IS 795/895
Fall 2009
44
III. INTRODUCTION TO
NETLOGO
IS 795/895
Fall 2009
45
Interface Elements
46
Elements of NetLogo
Interface tab
Procedures tab
Information tab
47
Interface Tab
48
Procedures Tab
49
Information Tab
50
Buttons
Sliders
Switches
Choosers
Monitors and
plots
51
0, 0
+ 16
33 x 33 square patches
Cartesian coordinates
World is customizable
Size, Scale
Shape of patch (square versus
hexagon)
Wrapping (none, top-bottom,
left-right, all)
Origin
+ 16
16
16
52
Introduction
Questions?
IS 795/895
Fall 2009
53