Evolutionary Dynamics, Games and Graphs
-
Upload
barbara-ikica -
Category
Science
-
view
99 -
download
1
Transcript of Evolutionary Dynamics, Games and Graphs
![Page 1: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/1.jpg)
Evolutionary Dynamics, Games and Graphs
Barbara IkicaSupervisor: prof. dr. Milan Hladnik
FMF UL
5 January 2016
![Page 2: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/2.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
![Page 3: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/3.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
![Page 4: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/4.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
![Page 5: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/5.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:
◦ the replicator equation,◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:• amplifiers of random drift,• amplifiers of selection,• the replicator equation on graphs.
![Page 6: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/6.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,
◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:• amplifiers of random drift,• amplifiers of selection,• the replicator equation on graphs.
![Page 7: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/7.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,◦ Nash equilibria and evolutionary stability,
◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:• amplifiers of random drift,• amplifiers of selection,• the replicator equation on graphs.
![Page 8: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/8.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:• amplifiers of random drift,• amplifiers of selection,• the replicator equation on graphs.
![Page 9: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/9.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:
◦ evolutionary graph theory:• amplifiers of random drift,• amplifiers of selection,• the replicator equation on graphs.
![Page 10: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/10.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:
• amplifiers of random drift,• amplifiers of selection,• the replicator equation on graphs.
![Page 11: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/11.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:• amplifiers of random drift,
• amplifiers of selection,• the replicator equation on graphs.
![Page 12: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/12.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:• amplifiers of random drift,• amplifiers of selection,
• the replicator equation on graphs.
![Page 13: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/13.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
Overview
• Deterministic models:◦ the replicator equation,◦ Nash equilibria and evolutionary stability,◦ permanence and persistence.
• Stochastic models:◦ evolutionary graph theory:• amplifiers of random drift,• amplifiers of selection,• the replicator equation on graphs.
![Page 14: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/14.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Deterministic models
State of the population (with n species):
∆n :={
xxx = (x1, x2, . . . , xn) ∈ Rn : xi ≥ 0 inn∑
i=1xi = 1
}
Fitness (reproductive success) of the i-th species: fi(xxx)
![Page 15: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/15.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Replicator dynamics
The replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n
Average fitness: f (xxx) =∑n
i=1 fi(xxx)xi
![Page 16: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/16.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Replicator dynamics
The replicator-mutator equation
xi = xi
(fi(xxx) − fi(xxx)
∑nj=1,j 6=i
qij
)+∑n
j=1,j 6=i
xj fj(xxx)qji − xi f (xxx), , i = 1, 2, . . . , n
Average fitness: f (xxx) =∑n
i=1 fi(xxx)xi
![Page 17: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/17.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Replicator dynamics
The replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n
Average fitness: f (xxx) =∑n
i=1 fi(xxx)xi
![Page 18: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/18.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Game theory in replicator dynamics
Strategies:
∆N :={
ppp = (p1, p2, . . . , pN ) ∈ RN : pi ≥ 0 inN∑
i=1pi = 1
}
Payoff matrix: U = [uij ]Ni,j=1
Expected payoff of a ppp-strategist against a qqq-strategist:ppp ·Uqqq
![Page 19: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/19.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Game theory in replicator dynamics
Strategies:
∆N :={
ppp = (p1, p2, . . . , pN ) ∈ RN : pi ≥ 0 inN∑
i=1pi = 1
}
Payoff matrix: U = [uij ]Ni,j=1
Expected payoff of a ppp-strategist against a qqq-strategist:ppp ·Uqqq
![Page 20: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/20.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Game theory in replicator dynamics
How to incorporate a game?
1. i-th species (xi) pppi
2. A = [aij ]ni,j=1, aij = pppi ·Upppj
3. fi(xxx) = (Axxx)i =∑n
j=1 pppi ·Upppjxj
The replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n
Average fitness: f (xxx) =∑n
i=1 fi(xxx)xi
![Page 21: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/21.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Game theory in replicator dynamics
How to incorporate a game?
1. i-th species (xi) pppi
2. A = [aij ]ni,j=1, aij = pppi ·Upppj
3. fi(xxx) = (Axxx)i =∑n
j=1 pppi ·Upppjxj
The replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n
Average fitness: f (xxx) =∑n
i=1 fi(xxx)xi
![Page 22: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/22.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Game theory in replicator dynamics
How to incorporate a game?
1. i-th species (xi) pppi
2. A = [aij ]ni,j=1, aij = pppi ·Upppj
3. fi(xxx) = (Axxx)i =∑n
j=1 pppi ·Upppjxj
The replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n
Average fitness: f (xxx) =∑n
i=1 fi(xxx)xi
![Page 23: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/23.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Game theory in replicator dynamics
How to incorporate a game?
1. i-th species (xi) pppi
2. A = [aij ]ni,j=1, aij = pppi ·Upppj
3. fi(xxx) = (Axxx)i =∑n
j=1 pppi ·Upppjxj
The replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n
Average fitness: f (xxx) =∑n
i=1 fi(xxx)xi
![Page 24: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/24.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Game theory in replicator dynamics
How to incorporate a game?
1. i-th species (xi) pppi
2. A = [aij ]ni,j=1, aij = pppi ·Upppj
3. fi(xxx) = (Axxx)i =∑n
j=1 pppi ·Upppjxj
The linear replicator equation
xi = xi((Axxx)i − xxx ·Axxx
), i = 1, 2, . . . ,n
Average fitness: f (xxx) = xxx ·Axxx
![Page 25: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/25.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Nash equilibria and evolutionary stability
(Symmetric) Nash equilibriumA strategy ppp ∈ ∆N such that for all ppp ∈ ∆N ,
ppp ·Uppp ≥ ppp ·Uppp .
Evolutionary stable strategyA strategy ppp ∈ ∆N such that for all ppp ∈ ∆N\{ppp},
ppp ·U(εppp + (1− ε)ppp
)> ppp ·U
(εppp + (1− ε)ppp
)holds for all sufficiently small ε > 0.
![Page 26: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/26.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Nash equilibria and evolutionary stability
(Symmetric) Nash equilibriumA strategy ppp ∈ ∆N such that for all ppp ∈ ∆N ,
ppp ·Uppp ≥ ppp ·Uppp .
Evolutionary stable strategyA strategy ppp ∈ ∆N such that for all ppp ∈ ∆N\{ppp},
ppp ·U(εppp + (1− ε)ppp
)> ppp ·U
(εppp + (1− ε)ppp
)holds for all sufficiently small ε > 0.
![Page 27: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/27.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Nash equilibria and evolutionary stability
TheoremA strategy ppp is an ESS iff (for 0 < ε < ε) the following twoconditions are satisfied:• equilibrium condition: ppp ·Uppp ≥ ppp ·Uppp for all ppp ∈ ∆N ,• stability condition: if ppp 6= ppp and ppp ·Uppp = ppp ·Uppp, then
ppp ·Uppp > ppp ·Uppp.
![Page 28: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/28.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Nash equilibria and evolutionary stability
(Symmetric) Nash equilibriumA strategy ppp ∈ ∆N such that for all ppp ∈ ∆N ,
ppp ·Uppp ≥ ppp ·Uppp .
Evolutionary stable strategyA strategy ppp ∈ ∆N such that for all ppp ∈ ∆N\{ppp},
ppp ·U(εppp + (1− ε)ppp
)> ppp ·U
(εppp + (1− ε)ppp
)holds for all sufficiently small ε > 0.
![Page 29: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/29.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Nash equilibria and evolutionary stability
(Symmetric) Nash equilibriumA state of the population xxx ∈ ∆n such that for all xxx ∈ ∆n ,
xxx ·Axxx ≥ xxx ·Axxx .
Evolutionary stable stateA state of the population xxx ∈ ∆n such that for all xxx 6= xxx in aneighbourhood of xxx in ∆n ,
xxx ·Axxx > xxx ·Axxx .
![Page 30: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/30.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Nash equilibria and evolutionary stability
(Symmetric) Nash equilibriumA state of the population xxx ∈ ∆n such that for all xxx ∈ ∆n ,
xxx ·Axxx ≥ xxx ·Axxx .
Evolutionary stable stateA state of the population xxx ∈ ∆n such that for all xxx 6= xxx in aneighbourhood of xxx in ∆n ,
xxx ·Axxx > xxx ·Axxx .
![Page 31: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/31.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Equilibria of the linear replicator equation
Nash equilibria
Stable equilibria ω-limit sets of orbits in int∆n
Asymptotically stable equilibri
aEvolutionary stable stat
es
Strict Nashequilibria
1
![Page 32: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/32.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
The Hawk–Dove Game
H D T
H G−C2 G G(C−G)
2C
D 0 G2
G(C−G)2C
T G(G−C)2C
G(G+C)2C
G(C−G)2C
e1
e2
e3
x�
=
1
3,
2
3, 0
![Page 33: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/33.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
The Hawk–Dove Game
H D T
H G−C2 G G(C−G)
2C
D 0 G2
G(C−G)2C
T G(G−C)2C
G(G+C)2C
G(C−G)2C
e1
e2
e3
x�
=
1
3,
2
3, 0
![Page 34: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/34.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
The Rock–Scissors–Paper Game
A =
R S P
R 0 1 + ε −1
S −1 0 1 + ε
P 1 + ε −1 0
![Page 35: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/35.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Permanence and persistence
PermanenceA dynamical system on ∆n ispermanent if there exists aδ > 0 such thatxi = xi(0) > 0 fori = 1, 2, . . . ,n implies
lim inf t→+∞ xi(t) > δ
for i = 1, 2, . . . ,n.
¶ = 1
e1
e2
e3
![Page 36: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/36.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Permanence and persistence
PersistenceA dynamical system on ∆n ispersistent if xi = xi(0) > 0 fori = 1, 2, . . . ,n implies
lim supt→+∞ xi(t) > 0
for i = 1, 2, . . . ,n.
¶ = -0,5
e1
e2
e3
![Page 37: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/37.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Permanence and persistence
Strong persistenceA dynamical system on ∆n isstrongly persistent ifxi = xi(0) > 0 fori = 1, 2, . . . ,n implies
lim inf t→+∞ xi(t) > 0
for i = 1, 2, . . . ,n.
¶ = 0
e1
e2
e3
![Page 38: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/38.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Index theory
SaturationAn equilibrium ppp of the replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n,
is saturated if fi(ppp) ≤ f (ppp) holds for all i with pi = 0.
General index theorem for the replicator equationThere exists at least one saturated equilibrium for thereplicator equation. If all saturated equilibria ppp are regular,i.e. det Jfff (ppp) 6= 0, the sum of their Poincaré indices
∑ppp i(ppp)
is (−1)n−1, and hence their number is odd.
![Page 39: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/39.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Index theory
SaturationAn equilibrium ppp of the replicator equation
xi = xi(fi(xxx)− f (xxx)
), i = 1, 2, . . . ,n,
is saturated if fi(ppp) ≤ f (ppp) holds for all i with pi = 0.
General index theorem for the replicator equationThere exists at least one saturated equilibrium for thereplicator equation. If all saturated equilibria ppp are regular,i.e. det Jfff (ppp) 6= 0, the sum of their Poincaré indices
∑ppp i(ppp)
is (−1)n−1, and hence their number is odd.
![Page 40: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/40.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Index theory
SaturationAn equilibrium ppp of the linear replicator equation
xi = xi((Axxx)i − xxx ·Axxx
), i = 1, 2, . . . ,n,
is saturated if (Appp)i ≤ ppp ·Appp holds for all i with pi = 0.
General index theorem for the replicator equationThere exists at least one saturated equilibrium for thereplicator equation. If all saturated equilibria ppp are regular,i.e. det Jfff (ppp) 6= 0, the sum of their Poincaré indices
∑ppp i(ppp)
is (−1)n−1, and hence their number is odd.
![Page 41: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/41.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Index theory
SaturationAn equilibrium ppp of the linear replicator equation
xi = xi((Axxx)i − xxx ·Axxx
), i = 1, 2, . . . ,n,
is saturated if (Appp)i ≤ ppp ·Appp holds for all i with pi = 0.
(Symmetric) Nash equilibriumA state of the population ppp ∈ ∆n such that for all xxx ∈ ∆n ,
xxx ·Appp ≤ ppp ·Appp .
![Page 42: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/42.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary graph theory
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
![Page 43: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/43.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary graph theory
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
ri∑Sk=1 nkrk
![Page 44: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/44.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11
![Page 45: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/45.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
0,80,2
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11
![Page 46: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/46.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
1110
![Page 47: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/47.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11100,7
0,3
![Page 48: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/48.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
1110
12
![Page 49: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/49.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
1110
12
0,3
0,2
0,5
![Page 50: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/50.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11
![Page 51: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/51.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11
![Page 52: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/52.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11
0,6
![Page 53: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/53.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
1
3
6
2 4
9 5
7 8
10
12
11
0,7
0,3
0,8
1
0,4
0,6
1
0,2 0,8
1
0,9
0,1 1
0,7
0,3
0,3
0,2
0,50,5
0,5
1
3
6
2 4
9 5
7 8
10
12
11
1
3
6
2 4
9 5
7 8
10
12
11
4
![Page 54: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/54.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
The Moran process in a homogeneous populationConsider a complete graph with N vertices and identicaledge weights. The corresponding fixation probability of asingle mutant with relative fitness r 6= 1 (in a population ofresidents with fitness 1) is given by
ρM := 1−1/r1−1/rN .
If r = 1, ρM = 1/N .
![Page 55: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/55.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
Classification of graphs according to ρG
1. If ρG = ρM , then the graph G is ρ-equivalent to theMoran process; it has he same balance of selection andrandom drift.
2. A graph G is an amplifier of selection ifρG > ρM for r > 1 and ρG < ρM for r < 1 .
3. A graph G is an amplifier of random drift ifρG < ρM for r > 1 and ρG > ρM for r < 1 .
![Page 56: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/56.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
Classification of graphs according to ρG
1. If ρG = ρM , then the graph G is ρ-equivalent to theMoran process; it has he same balance of selection andrandom drift.
2. A graph G is an amplifier of selection ifρG > ρM for r > 1 and ρG < ρM for r < 1 .
3. A graph G is an amplifier of random drift ifρG < ρM for r > 1 and ρG > ρM for r < 1 .
![Page 57: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/57.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Fixation probability ρG
Classification of graphs according to ρG
1. If ρG = ρM , then the graph G is ρ-equivalent to theMoran process; it has he same balance of selection andrandom drift.
2. A graph G is an amplifier of selection ifρG > ρM for r > 1 and ρG < ρM for r < 1 .
3. A graph G is an amplifier of random drift ifρG < ρM for r > 1 and ρG > ρM for r < 1 .
![Page 58: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/58.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
ρ-equivalence to the Moran process
The isothermal theoremA graph G is ρ-equivalent to the Moran process if and onlyif it is isothermal.
0,3
0,6
0,10,20,3
0,40,1
![Page 59: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/59.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Amplifiers of random drift
Construction of amplifiers of random driftSuppose 1/N ≈ 0. Choose a fitness r > 1 and a constantρ ∈ (1/N , ρM ) or, alternatively, a fitness r < 1 and aconstant ρ ∈ (ρM , 1/N ). There exists a graph G on Nvertices such that ρG = ρ.
ρG(N1) :=N1
N1− 1/r
1− 1/rN1.
![Page 60: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/60.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Amplifiers of random drift
Construction of amplifiers of random driftSuppose 1/N ≈ 0. Choose a fitness r > 1 and a constantρ ∈ (1/N , ρM ) or, alternatively, a fitness r < 1 and aconstant ρ ∈ (ρM , 1/N ). There exists a graph G on Nvertices such that ρG = ρ.
ρG(N1) :=N1
N1− 1/r
1− 1/rN1.
![Page 61: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/61.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Amplifiers of selection
TheoremLet G(L,C ,D) be a superstar with D > 2. In the limit as Land C tend to infinity, for r > 1,
1− 1r4(D−1)(1−1/r)2 ≤ ρ ≤ 1− 1
1+r4D ,
and for 0 < r < 1,
ρ ≤((1/r)4T
)−δ+1 .
Here, T and δ > 1 are appropriately chosen natural numberswith T satisfying (D − 1)(1− r)2 ≤ T ≤ D.
![Page 62: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/62.jpg)
Slika 5.4: Superzvezda G(6, 4, 2). Barve vozlišč sorazmerno nakazujejo njihove temperature– svetlomodra vozlišča s temperaturo 1/24 sodijo med najbolj ohlajena, rdeče obarvan korens temperaturo 6 je najbolj vroče vozlišče, na vsakem deblu pa zasledimo eno rahlo in enomočno osvetljeno rdeče vozlišče s pripadajočima temperaturama 4 in 1 zaporedoma.
Za N � 1 se navedene verjetnosti odražajo v različnih časovnih okvirih, znotrajkaterih poteka dinamika na različnih vrstah vozlišč superzvezde. Spremembe senajpočasneje odvijajo na zbiralnih vozliščih, na katerih je posameznik v povprečjunadomeščen enkrat na vsakih N2 časovnih korakov. Srednje hitro potekajo navozliščih z debel (od drugega naprej), na katerih prihaja do sprememb približnoenkrat na vsakih N korakov. Najhitreje pa se zgodijo na korenu in na prvih vozliščihdebel, kjer se vozlišča v povprečju posodabljajo enkrat na vsakih
√N korakov.
Podrobneje bomo proučili le srednje hitro dinamiko, torej tisto, ki se odvija nadeblih. Zgornji razmislek nam namreč omogoča, da z njenega vidika stanja zbiralnihvozlišč zaradi počasnejšega posodabljanja obravnavamo kot konstantna, stanja hitrospreminjajočih se vozlišč pa aproksimiramo s slučajnimi spremenljivkami. Sedajsprostimo zahtevo r ≈ 1 in se lotimo dokazovanja.
119
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Amplifiers of selection
0 < r < 1 :
ρ ≤((1/r)4T
)−δ+1
(D − 1)(1− r)2 ≤ T ≤ D
r > 1 :
1− 1r4(D − 1) (1− 1/r)2 ≤ ρ ≤ 1− 1
1 + r4D
![Page 63: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/63.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary game theory on graphs
Strategies: R1,R2, . . . ,Rn ; payoff matrix: A = [aij ]ni,j=1
Graphs: N vertices, undirected and unweighted edges,k-regular
Payoff of a Ri-strategist withkj neighbouring Rj-strategists:
Fi =n∑
j=1kjaij
Fitness of a Ri-strategist: fi = 1− w + wFi , w ∈ [0, 1]intensity of selection
![Page 64: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/64.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary game theory on graphs
Strategies: R1,R2, . . . ,Rn ; payoff matrix: A = [aij ]ni,j=1Graphs: N vertices, undirected and unweighted edges,k-regular
Payoff of a Ri-strategist withkj neighbouring Rj-strategists:
Fi =n∑
j=1kjaij
Fitness of a Ri-strategist: fi = 1− w + wFi , w ∈ [0, 1]intensity of selection
![Page 65: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/65.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary game theory on graphs
Strategies: R1,R2, . . . ,Rn ; payoff matrix: A = [aij ]ni,j=1Graphs: N vertices, undirected and unweighted edges,k-regular
Ri
Payoff of a Ri-strategist withkj neighbouring Rj-strategists:
Fi =n∑
j=1kjaij
Fitness of a Ri-strategist: fi = 1− w + wFi , w ∈ [0, 1]intensity of selection
![Page 66: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/66.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary game theory on graphs
Strategies: R1,R2, . . . ,Rn ; payoff matrix: A = [aij ]ni,j=1Graphs: N vertices, undirected and unweighted edges,k-regular
Ri
RjRk
Rl
Payoff of a Ri-strategist withkj neighbouring Rj-strategists:
Fi =n∑
j=1kjaij
Fitness of a Ri-strategist: fi = 1− w + wFi , w ∈ [0, 1]intensity of selection
![Page 67: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/67.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary game theory on graphs
Strategies: R1,R2, . . . ,Rn ; payoff matrix: A = [aij ]ni,j=1Graphs: N vertices, undirected and unweighted edges,k-regular
Ri
RjRk
Rl
Payoff of a Ri-strategist withkj neighbouring Rj-strategists:
Fi =n∑
j=1kjaij
Fitness of a Ri-strategist: fi = 1− w + wFi , w ∈ [0, 1]intensity of selection
![Page 68: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/68.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary game theory on graphs
Strategies: R1,R2, . . . ,Rn ; payoff matrix: A = [aij ]ni,j=1Graphs: N vertices, undirected and unweighted edges,k-regular
Ri
RjRk
Rl
Payoff of a Ri-strategist withkj neighbouring Rj-strategists:
Fi =n∑
j=1kjaij
Fitness of a Ri-strategist: fi = 1− w + wFi , w ∈ [0, 1]intensity of selection
![Page 69: Evolutionary Dynamics, Games and Graphs](https://reader031.fdocuments.in/reader031/viewer/2022020119/58a60b681a28ab773d8b47d1/html5/thumbnails/69.jpg)
EvolutionaryDynamics, Games
and Graphs
Barbara Ikica
DeterministicmodelsThe replicator equation
Nash equilibria andevolutionary stability
Permanence and persistence
Stochastic modelsEvolutionary graph theory
Amplifiers of random drift
Amplifiers of selection
The replicator equation ongraphs
Evolutionary game theory on graphs
Let xi(t) denote the expected frequency of Ri-strategists attime t ≥ 0.The replicator equation on graphsSuppose k > 2 and N � 1. In the limit of weak selection,w → 0, the following equation can be derived to describeevolutionary game dynamics on graphs.
xi = xi((
(A + B)xxx)
i − xxx · (A + B)xxx), i = 1, 2, . . . ,n .
Here, the elements of the matrix B = [bij ]ni,j=1 are given by
bij = aii + aij − aji − ajjk − 2 .