Evolutionary Stability in the Asymmetric Volunteer`s Dilemma

advertisement
1
Evolutionary Stability in the Asymmetric Volunteer’s Dilemma
2
Supporting Information
3
Jun-Zhou He, Rui-Wu Wang and Yao-Tang Li
4
Appendix S1: The equilibrium points and Local stability analysis
5
From the definition of equilibrium point of nonlinear systems, we know that the
6
equilibrium points of the equation (2.3) in text are the solutions of the following
7
equations
 S ( 1  S )K( S U S N 1W ) 0
.

N 2

(
1


)
K
(

U



)
0
W
W
W
W
S
W

8
9
(1)
Now, we solve the equation (1). Solving  S (1   S )( K S  U S W N 1 )  0 , we get
 S  0,  S  1, W   K S U S 
1 ( N 1)
10
.
11
Solving W (1  W )( KW  UW  S W N  2 )  0 , we get W  0, W  1 and
12
KW  UW  S W N  2  0 .
(2)
13
Obviously, points (  S , W )  (0,1), (0, 0), (1, 0), (1,1) are the solutions of the
14
equations (1). We denote them as A(0,1) , B (0, 0) , C (1, 0) , D (1,1) , respectively,
15
where  S and W are respectively the frequencies of defectors of “strong” and
16
“weak” players.
17
Substituting  S  1 in Equation (2) yields:
W   KW UW 
1 ( N  2)
18
  ,    1,  K
UW 
1 ( N 2)

19
that is,
20
denote it as E (1, W* ) ,here W*   KW UW 
21
S
W
W
,
is also a solution of the equations (1), and we
1 ( N  2)
Substituting W   K S U S 
1 ( N 1)
.
in Equation (2), we obtain
1
 S   KW UW  K S U S 
  KW UW  K S U S 
22
( N  2) ( N 1)
( N 11) ( N 1)
  KW UW  K S U S  K S U S 
1 ( N 1)
 U S KW K SUW  K S U S 
1 ( N 1)


, W   U S KW K SUW  K S U S 
.
,
 KS
denote
it
1 ( N 1)
US 
1 ( N 1)

23
So
24
the
25
 S**  U S KW K SUW  K S U S 
26
From above discussion, we obtain six equilibrium points of the nonlinear system (2.3):
27
A(0,1) , B (0, 0) , C (1, 0) , D (1,1) , E (1, W* ) , F (  S** , W** ) ,
28
S
equations
(1),
we
and
1 ( N 1)
where W*   KW UW 
1 ( N  2)
F (  S** , W** )
as
and W**   K S U S 
1 ( N 1)
,  S**  U S KW K SUW  K S U S 
1 ( N 1)
is also a solution of
,
here
.
, W**   K S U S 
1 ( N 1)
.
29
30
The following is the local stability analysis of the system of volunteer dilemma
31
game. In text, the linearization of the replicator dynamics (2.3) at an equilibrium point
32
(  S , W ) is
 f (  S , W )
 g (  S , W )
  
33
34
S
f ( S ,
g ( S ,
)  W
  (S
W)
S
.
W
,
W
)  ( S , W
(3)
)
For convenience, let
 f (  S , W )  S
J (  S , W )  
 g (  S , W )  S
35
36
S
f (  S , W ) W 
,

g (  S , W )  S  (  ,  ) (  ,  )
S W
S W
(4)
where
37
f (  S , W )  S  (1  2 S )( K S  U S W N 1 ) ,
38
f (  S , W ) W  ( N  1)U S W N  2  S (1   S ) ,
39
g (  S , W )  S  UW W N 1 (1  W ) ,
40
g (  S , W )  S  (1  2W ) KW  [ N W  ( N  1)]UW  S W N  2 ,
2
(  S , W ) represents the equilibrium points of the nonlinear system (2.3),  S and
41
and
42
W are respectively the frequencies of defection of “strong” and “weak” players.
43
44
From (4), we obtain the Jacobi matrices of the linear system (3) of the nonlinear
system (2.3) at these six equilibrium points:
0 
,
KW 
 K US
J ( 0 , 1) S
 0
0 
,
 KW 
 K
J ( 1 , 0) S
 0
45
K
J (0, 0)   S
 0
46
 [ K S  U S ( W* ) N 1 ]

0
0
U S  K S

*
J (1,1)  
 , J (1, W )  
* 
UW  KW 
0
( N  2) KW (1  W ) 
 0

47

0
( N  1)U S  S** (1   S** )( W** ) N 2 
J (  S** , W** )  
.
**
** N 1
( N  2)UW  S** (1  W** )( W** ) N 2 
 UW (1  W )( W )
0 
,
KW 
48
The local stability of the nonlinear system (2.3) at equilibrium points is
49
determined by the eigenvalues of the Jacobi matrices. Now, we analyze character of
50
the eigenvalues of above six matrices.
51
Since 0  K S  KW  UW  U S and
J (0, 0) , J (1,1) , J (0,1) and
J (1, 0) are
52
diagonal matrices, we can easily get that the eigenvalues of J (0, 0) and J (1,1) are
53
all positive, and the eigenvalues of J (0,1) are negative. On thus condition, one of
54
the eigenvalues of J (1, 0) is positive and another is negative.
55
For J (1, W* ) , we obtain easily that its eigenvalues are ( N  2) KW (1  W* ) and
56
[ K S  U S ( W* ) N 1 ] . Obviously,
57
0  K S U S  KW UW  1 , we know that yield
58
W*   KW UW 
59
negative if and only if K S U S   KW UW 
60
J (1, W* ) are all negative under the special condition
1 N 2
( N  2) KW (1  W* )  0 .
) demand K S U S   KW UW 
[ K S  U S ( W* ) N 1 ]  0
N 1 N  2
N 1 N  2
In addition, from
(here
. So another eigenvalue is
, that is, the eigenvalues of
 KS
US 
N 2
  KW UW 
N 1
.
3
61
As to J (  S** , W** ) , its characteristic equation is
 2  a22   a12 a21  0 .
62
63
Solving Equation (8), we get

64
65
2
a22  a22
 4a12 a21
,
2
where
66
a12  ( N  1)U S  S** (1   S** )( W** ) N 2 ,
67
a21  UW (1  W** )( W** ) N 1 ,
68
a22  ( N  2)UW  S** (1  W** )( W** ) N  2 .
69
70
71
(8)
Obviously, a12  0, a21  0
and
a22  0 ,
then
it
is
easy
to
obtain
2
a22  a22
 4a12 a21 . Hence, one of the eigenvalues of J (  S** , W** ) is a positive real
number and another is a negative.
72
From above analysis of the eigenvalues and the stability theory of differential
73
equations (Hofbauer and Sigmund 1998; Hirsch et al. 2004), we obtain that A(0,1)
74
and E (1, W* ) are sinks (stable); and that B (0, 0) and D (1,1) are sources (unstable);
75
C (1, 0) and
F (  S** , W** ) are saddles (unstable).
76
77
References
78
Hirsch, M. W., Smale, S., Devaney, R. L., 2004. Differential Equations, Dynamical
79
80
81
Systems, and an Introduction to Chaos. Calif. Acad. Press.
Hofbauer, J., Sigmund, K., 1998. Evolutionary Games and Population Dynamics.
Cambridge Uni. Press.
4
Download