OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger...
Transcript of OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger...
![Page 1: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/1.jpg)
Optimizing Counter-Terror Operations
Gustav Feichtinger
http://orcos.tuwien.ac.at/home/[email protected]
1 / 1
![Page 2: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/2.jpg)
1
Optimal Control of Terroristic Attacks
Gustav Feichtinger
Institute for Mathematical Methods in EconomicsVienna University of Technology
http://www.eos.tuwien.ac.at/OR/
Joint work with J.P. Caulkins, D. Grass, A. Novak
![Page 3: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/3.jpg)
2
• Introduction
• Two Sides of a Coin
• Terminology of Differential Games
• Obama vs. Osama
• Leader – Follower Games
• A Post September 11th Game on Terrorism
• Fire & Water Strategies
• Public Opinion as ‘Catalysator’
• Conclusions
Roadmap
![Page 4: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/4.jpg)
3
• Keohane and Zeckhauser, 2003: „The ecology of terror
defense”, state = ITO’s resources
• Castillo-Chavez & Song, 2003: „Models for the transmission
dynamics of fanatical behaviors“
Modeling a “Stock” of Terrorists is not Common
but Has Precedent
Introduction
![Page 5: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/5.jpg)
4
• Faria, 2004: „Terror cycles“
• Kaplan et al., 2005: „What happened to suicide bombings in
Israel? Insights from a terror stock model”
• Udwadia et al., 2006: „A dynamical model of terrorism“
• Caulkins et al., 2007: fire and water strategies for counter–
terrorism, “double-edged sword effect”; terrorism and global
reputation
• Grass et al., 2008
• Kress & Szechtman, 2009: effect of intelligence in counter-
insurgency operations
![Page 6: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/6.jpg)
• Think of there being a population or “stock” of
terrorists
• Focus on “flows” in and out of that stock
– Pay attention to the possibility that some interventions
that increase an outflow might simultaneously increase
inflow
– Levels of associated “reputation” or “perception” stocks
can affect those flows
5
![Page 7: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/7.jpg)
Data Limitations
• Terror stock data are poor
– So models are stylized and not validated
– Value stems from articulating ideas in a language
as precise as mathematics
• Not from hypothesis testing
• Providing specific, quantitative guidance
6
![Page 8: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/8.jpg)
7
Two Sides of a CoinA simple counter-terror model
x(t) … number of terrorists at time t
y(t) … power of the government
u(t) … counter-terror measures
… obsolence rate
… growth function
… impact of accumulated
terrorist fighting
![Page 9: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/9.jpg)
8
y x+ -
u
- -
government wants to minimize both the damage of terroristic
attacks and the cost of ist measures:
… damage, … costs
![Page 10: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/10.jpg)
9
4-D canonical system
Supercritical Hopf-bifurcation: stable limit cycles,
persistent oscillations make economic sense
Wirl & Feichtinger (1992)
![Page 11: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/11.jpg)
payoff rate , i = 1,…, N
infinite-horizon problem
10
![Page 12: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/12.jpg)
Markovian (Feedback or Closed-loop)
Information Pattern
stationary Markovian strategy
Open-loop Information Pattern
11
![Page 13: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/13.jpg)
Tractable Game Stuctures
• Linear quadratic games
Engwerda (2005)
• State-separable games (Linear state games)
Dockner et al. (1985), Dockner et al. (2000, Sect. 7.2)
Open-loop Nash equilibria qualify as Markovian solution
and for
12
![Page 14: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/14.jpg)
13
Obama vs. Osama
A dynamic game of the government vs. ITO
x(t) … resources of the terrorists at time t
v(t) … intensity of attacks; strategic
instrument of the terrorists
u(t) … counter-terror measures
Player 1: Obama (US): resource exploiter
Player 2: Osama (ITO): resource conservator
![Page 15: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/15.jpg)
14
g(x) … growth function (hump-shaped)
h(u,v) … ‚harvesting‘ function
![Page 16: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/16.jpg)
15
1st player: government : resource exploiter
2nd player: ITO: resource conserver
![Page 17: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/17.jpg)
16
shadow prices of the stock x to player 1 and 2
![Page 18: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/18.jpg)
17
Proposition 1.
(i) The Hamiltonian of player 1 is concave in the control u as
.
(ii) The shadow price is positive and is thus concave in v.
(iii) The shadow price is positive, iff the marginal utility of
the governments‘ activities exceeds the marginal costs .
![Page 19: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/19.jpg)
18
Proposition 2. Given the specifications of functions as
above, a necessary condition for the existence of purely
imaginary eigenvalues is that the terrorists are more
impatient than the governments, i.e.
has to hold
![Page 20: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/20.jpg)
19
computer code BIFDD
Hopf bifurcation at
Stable cycles at
![Page 21: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/21.jpg)
20
![Page 22: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/22.jpg)
21
![Page 23: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/23.jpg)
A Post September 11th Game on
Terrorism
22
![Page 24: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/24.jpg)
23
![Page 25: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/25.jpg)
Nash equilibrium
24
![Page 26: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/26.jpg)
implies
implies
Stationary Nash equilibrium
unstable steady state25
![Page 27: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/27.jpg)
1.1.
26
![Page 28: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/28.jpg)
The Stackelberg Equilibrium with the ITO as Leader
Step 1: ITO announces v
Step 2:
Step 3:
s.t.
(**)
given by (**)
27
![Page 29: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/29.jpg)
28
,
Due to that state separability the adjoint variable of the leader
w.r.t. the adj. var. of the followers plays no role in the opt.
problem
![Page 30: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/30.jpg)
Comparison of the Nash and Stackelberg
Equilibrium Solutions
and
29
![Page 31: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/31.jpg)
30
„Fire strategies“: territorial bombing, aggressively searching all
people, activities involving significant collateral damage
→ inconvenience to third parties, resentment by population,
stimulation of recruitment rates, elimination of current
terrorists
„Water strategies“: intelligence driven arrests or „surgical“ operations
against almost certainly guilty individuals
→ no harm to innocent parties, higher acceptance by
population, expensive, difficult to apply
Fire & Water Strategies
![Page 32: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/32.jpg)
31
Occurrence of terroristic attack at t=0
Stage 1: modest counter measures, „water strategy“
Stage 2: additional, more aggressive measures, „fire strategy“
(side effect: increased inflow of recruits to terror organisation)
x(t) … number of terrorists at time t
u(t) … „water strategy“ at time t
v(t) … „fire strategy“ at time t
![Page 33: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/33.jpg)
32
Stage 1:
s.t.
Stage 2:
s.t.
where
![Page 34: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/34.jpg)
33
![Page 35: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/35.jpg)
34
![Page 36: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/36.jpg)
35
Public Opinion as Catalysator
•Is it important to manage public opinion while fighting terrorism?
Caulkins et al. (2007):
x … size of the terrorist organizationy … reputation, level of public sympathy for
counter-terror forcesu … counter-terror operations
y decreases with u but increases with xy
x
+ -u
-
-
-
-
![Page 37: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/37.jpg)
36
s.t.
![Page 38: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/38.jpg)
37
![Page 39: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/39.jpg)
38
![Page 40: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/40.jpg)
39
![Page 41: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/41.jpg)
40
Conclusion
• Charlatanerie in violence / peace „research“
• Stock-flow models of counter-terrorism
• Qualitative insights: advantage of the maximum principle
•Persistent oscillations (stable limit cycles) as
(open-loop) Nash solutions
• Complex behavior of „optimal“ strategies in
differential games
• Statement of J. Lesourne
![Page 42: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/42.jpg)
41
http://www.springer.com/economics/game+theory/book/978-3-540-77646-8
http://www.eos.tuwien.ac.at/OR/OCMat/
![Page 43: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/43.jpg)
Optimal Control and Differential Games of
Violence
How to Fight Terror and Insurgencies
Gustav Feichtinger
http://orcos.tuwien.ac.at/home/[email protected]
Joint work with A.J. Novak, A.Seidl, G. Tragler, and S. Wrzaczek
1 / 18
![Page 44: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/44.jpg)
Outline
1 Research Goals and Methodological Objectives
2 Optimizing Counter-terror Operations in Asymmetric Lanchester Models
3 Multi-Stage Differential Games
4 Popular Support to Terror and Counter-Terror
2 / 18
![Page 45: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/45.jpg)
Research Goals and Methodological Objectives
Introduction
OR & ManSci: Modeling of dynamic optimization problems both withand without strategic interactions
Seierstad & Sydsaeter (1987); Leonard & Long (1992); Sethi &Thompson (2000); Feichtinger & Hartl (1986); Grass et al. (2008)
Basar & Olsder (1995); Dockner et al. (2000); Long (2010)
Future work: combining multi-stage optimal control and optimalimpulse control methods with dynamic games
3 / 18
![Page 46: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/46.jpg)
Research Goals and Methodological Objectives
War on Terror: terror attacks & counter-terror measures
enormous economic problems, but only very few ’hard-core’ dynamicoptimization models
MAthematics of PUblic SEcurity (MAPUSE)
Conflicts between governments and terrorists: both continous policychanges and abrupt ’game-changing’ events
4 / 18
![Page 47: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/47.jpg)
Research Goals and Methodological Objectives
Research Goals
1 Mathematical advances - both analytical and numerical
2 Modeling war on terror
Methodological issues
(i) Bifurcation theory of nonlinear dynamical systems: multipleequilibria and limit cycles; mechanisms generating complexbehavior
(ii) Multi-stage modeling: switch between consecutives regimes;timing of switch: exogenous, stochastic or endogenous.
(iii) Differential games: interactions between different decisionmakers in a dynamic framework
(iv) Impulse control: control instruments executed only atoptimally determined points in time (not continuously!)
5 / 18
![Page 48: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/48.jpg)
Lanchester-type Models
Optimizing Counter-terror Operations in Asymmetric
Lanchester Models
Lanchester (1916): Battle between two parties
Guerilla warfare: Deitchman (1962), Schaffer (1968)
Kress & Szechtman (2009), Kaplan et al. (2010): levels ofintelligence, government vs. small guerrilla group
Asymmetric nature of counter-terror
Government forces: limited awareness of enemiesGuerillas: can observe location of goverment forces
Lanchester square law, Lanchester linear law
6 / 18
![Page 49: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/49.jpg)
Lanchester-type Models
Kress & Szechtman (KKS): Lanchester paradigm with intelligence
2 state variables :
G ... size of government forcesI ... size of terrorist group (0 ≤ I ≤ 1)
intelligence rate µ ∈ [0, 1]
atttrition rates α and γ
β & ǫ recruitment rates of government and terrorists
G = −αI + β
I = −γG(µ+ (1− µ)I) + ǫ+ θ(C)
7 / 18
![Page 50: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/50.jpg)
Lanchester-type Models
Double edge sword effect θ(C)
Collateral casualities
C = γG(1 − µ)(1− I)
µ = 1: classical Lanchester’s square law
µ = 0: guerrilla model
Control variables:
intelligence rate µ
recruitment rate α
8 / 18
![Page 51: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/51.jpg)
Lanchester-type Models
Resulting Optimization Problem
minµ,β
∫
∞
0
e−rt(
D(I) + C(µ)−A(G) +K(β))
dt,
s.t. G = −αI + β,
I = −γG(µ + (1− µ)I) + ǫ+ θ(C),
0 ≤ µ ≤ 1, G ≥ 0, I ≥ 0.
D(I) ... damage created by insurgents
C(µ) ... cost of intelligence
A(G) ... costs of army
K(β) ... costs of hiring and firing
9 / 18
![Page 52: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/52.jpg)
Lanchester-type Models
Analysis of a Dynamic Version of the Lanchester Model
Lanchester (1916): forerunner of Operations Research
Since then: descriptive models
Optimal control models: Multiple equilibria and path-dependence:
- occur in several models of counter-terror (e.g., Caulkinset al., 2008, 2009)
- “tipping” between two or more long-run solutions- “eradicate” vs. “accomodate” policy- Existence of three-fold indifference-threshold point inthe (G, I)−state space?
- Conjecture by F. Wagener: OC models with n statevariables, multiplicity of indifference points at mostn+ 1
Persistent oscillations
10 / 18
![Page 53: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/53.jpg)
Lanchester-type Models
Lanchester Model in a Differential Game Setting
Lucas critique applies for terror OC models
Feedback mechanisms within Lanchester model in a DG
11 / 18
![Page 54: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/54.jpg)
Multi-Stage Differential Games
Multi-Stage Differential Games Between a Government
and a Terrorist Organization
DG beteween a government (player 1) and ITO (player 2)
multi-stage approach in OC: see e.g. Tomiyama & Rossana (1989),Makris (2001), Grass et al. (2008, Sect. 8.1.3)
state x: number of terrorists
control instruments of government:
u1: water measures: surgical operationsv1: fire measures: also kill civilians; stimulate recruitment rates ofterrorists
control instruments of terrorists:
u2: attack rate
P1: disutility from x, u1, v1 (if used) and u2
P2: utility and disutility from same quantities, costs from u2
ITO’s political objecitves: positive assessment of both strategies of P1
12 / 18
![Page 55: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/55.jpg)
Multi-Stage Differential Games
Applying Multi-stage Methods and Impulse Controls to Differential
Games
Stage 1: Substantial terror attack at t = 0; P1 applies u1, P2 uses u2
Stage 2: At time ts P1 decides to switch from water to fire; P1 usesv1 and u1, P2 attacks with u2
multi-stage approach in games: Reddy et al. (2012) (cooperativegame), Wrzaczek & Kort (2012) (threat of entry of competitor)
Two or more decision makers influence switching time betweenconsecutive regimes
Anticipation of Terror Attacks and Counter Measures
Government expects terror attackGoverment cannot anticipate terror attackGovernment knows when attack will happen (but cannot prevent it)Government tries to prevent (impact) of terror attack
13 / 18
![Page 56: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/56.jpg)
Popular Support to Terror and Counter-Terror
Popular Support to Terror and Counter-Terror
Fighting terrorism means influencing public opinion
Civilian population between the terrorists and government
Both players need popular support
Population source for recruitment of terrorists
14 / 18
![Page 57: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/57.jpg)
Popular Support to Terror and Counter-Terror Managing Public Opinion While Fighting Terrorism
Managing Public Opinion While Fighting Terrorism
Caulkins et al. (2009) introduced reputation (level of public sympathyfor counter-terror forces) as second state variable
Fighting terrorists: higher security for population, but collateraldamage to innocent parties (→ erodes sympathy for government)
OC model, government decision maker, who minimizes strength ofterorrists and costs for counter-terror operations
See also Grass et al. (2008, Sect. 6.3)
Existence of mulitple equilibria, indifference curve
15 / 18
![Page 58: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/58.jpg)
Popular Support to Terror and Counter-Terror Managing Public Opinion While Fighting Terrorism
Extension: terror organisation as additional decision maker
two-person non-zero sum DG
Catalysator-type property of reputation: if reputation is high,counter-terror measures more effective
Catalysator effect: general character of variety of problems: e.g.,contagious diseases (quarantines and shutting down of publictransport)
16 / 18
![Page 59: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/59.jpg)
Popular Support to Terror and Counter-Terror Fighting Terror in a Host Country
Fighting Terror in a Host Country
Population’s interaction with terrorist as three player DG:
Population as additional player
Brito & Intriligator (1992): guerrillas, government and drug lords
Langlois & Langlois (2011): repeated game between US, host countryand terrorists
host will push back against terrorists in anticipation offuture involvement of Western power
More than three players: local insurgents with local objectives
Multi-player Games: Two parties might cooperate
17 / 18
![Page 60: OptimizingCounter-TerrorOperations...1 Optimal Control of Terroristic Attacks Gustav Feichtinger Institute for Mathematical Methods in Economics Vienna University of Technology gustav@eos.tuwien.ac.at](https://reader033.fdocuments.us/reader033/viewer/2022053007/5f0b57ed7e708231d4300c64/html5/thumbnails/60.jpg)
Popular Support to Terror and Counter-Terror Terror Escalations Along the Progress Curve
Terror Escalations Along the Progress Curve
Johnson et al. (2011):
number of attacks as proxy for accumulated terrorist experiencetiming of terror attacks power progress lawescalations of terror shrinks intervals of attacksescalation rate (speed of terrorist learning) between 0 and 1.5
Hartl et al. (2012): impulse optimal control model
18 / 18