Upload
prentice
View
42
Download
0
Embed Size (px)
DESCRIPTION
Punctuated Equilibrium and Criticality on Network Structures. I. Introduction II. Random Neighbor Model III. BS Evolution Model on Network Structures IV. Results V. Summary. Sungmin Lee, Yup Kim Kyung Hee Univ. I. Introduction. The "punctuated equilibrium" theory. - PowerPoint PPT Presentation
Citation preview
12004 년 봄 물리학회성균관대학교 ( 수원 )
Punctuated Equilibrium and
Criticality on Network Structures
I. Introduction
II. Random Neighbor Model
III. BS Evolution Model
on Network Structures
IV. Results
V. Summary
Sungmin Lee, Yup KimKyung Hee Univ.
22004 년 봄 물리학회성균관대학교 ( 수원 )
I. Introduction
Self-organized critical steady state
S.J.Gould (1972)
Instead of a slow, continuous movement, evolution
tends to be characterized by long periods of virtual
standstill ("equilibrium"), "punctuated" by episodes
Of very fast development of new forms
The "punctuated equilibrium" theory
32004 년 봄 물리학회성균관대학교 ( 수원 )
The Bak-Sneppen evolution model
0.2 0.30.15
0.40.45
0.7 0.90.35
0.10.55
0.75
0.5 0.80.65
0.60.25
Fitness - An entire species is represented by a single fitness - The ability of species to survive - The fitness of each species is affected by other species to which it is coupled in the ecosystem.
At each time step, the ecology is updated by (i) locating the site with thelowest fitness and mutating it by assigning a new
randomnumber to that site, and
10 if ),,1( Ni PBC
P.Bak and K.sneppenPRL 71,4083 (1993)
0.2 0.30.15
0.40.45
0.7 0.90.95
0.47
0.22
0.75
0.5 0.80.65
0.60.25
Lowest fitness
(ii) changing the landscapes of the two neighbors by assigning new random numbers
to those sites
New lowest fitness
Snapshot of the stationary
state66702.0cf
M.Paczuski, S.Maslov, P.BakPRE 53,414 (1996)
42004 년 봄 물리학회성균관대학교 ( 수원 )
Gap and Critical fitness
)(min sf : The lowest fitness at step s
)](,),0(max[)( minmin sffsG
cfsG )(
52004 년 봄 물리학회성균관대학교 ( 수원 )
Avalanche - subsequent sequences of mutations through fitness below a certain threshold
Distribution of avalanche
sizes in the critical state
Punctuated equilibria - long periods of passivity interrupted by sudden bursts of activity
The activity versus time in a local segment
of ten consecutive sites.
SSP ~)(
1d 2d
1.07(1) 1.245(10)
62004 년 봄 물리학회성균관대학교 ( 수원 )
(1) biospecies 의 연관구조가 Scale-free Network 나 Random Network 일 때 evolution 의 특성 연구
Motivation of this study
(2) Evolution and Punctuated
Equilibrium 의 Network
structures 에서의 Self-Organized
Criticality (3) What is the best structure
for the adaptation of species-correlation? (Is there evolution-free
network?)
72004 년 봄 물리학회성균관대학교 ( 수원 )
Exactly solvable model
0.2 0.30.15
0.40.45
0.7 0.90.35
0.10.55
0.75
0.5 0.80.65
0.60.25
- At each time step, the ecology is updated by (i) locating the node with the lowest fitness and mutating it by assigning a new random number to that site (ii) changing the landscapes of randomly
selected K-1 sites by assigning new random
numbers to those sites.
II. Random Neighbor Model
0.2 0.30.15
0.77
0.45
0.34
0.90.35
0.52
0.55
0.75
0.50.22
0.65
0.60.89
Lowest fitness
New lowest fitness
ix: the i-th smallest fitness value
)(xpi : the distribution for ix)(xp : the distribution of all fitness values in the ecology
)()()()( 1)!()!1(
! xQxpxPxp iNiiNi
Ni
x
xpxdxP0
)()( 1
)()(x
xpxdxQwhere
82004 년 봄 물리학회성균관대학교 ( 수원 )
),(),()1,( 11 txptxptxp N
NK
NNK txptxp
),(),( 11
11
The evolution equation for ),( txp
NKxp )(
1)( KKxp
)( 11NK x
)( 11NKx
for
for
SSP ~)( 5.1
(1) identifying each burst with a node(2) and each of K new fitness values resulting from a burst - with either a branch rooted in that node (if the fitness value is less than the threshold value) - with a leaf rooted in the same node (if the fitness value is above threshold)
Avalanches
Kcf10 1
The limit is necessary to obtain the tree structure.N
t
92004 년 봄 물리학회성균관대학교 ( 수원 )
- generate network structures with N nodes- A random fitness equally distributed between 0 and
1, is assigned to each node.- At each time step, the ecology is updated by (i) locating the node with the lowest fitness and
mutating it by assigning a new random number to that site (ii) changing the landscapes of the linked neighbors
by assigning new random numbers to those nodes.
III. BS Evolution Model on Network Structures
0.2
0.75
0.6
0.7
0.8
0.9
0.4
0.3
0.5
0.1
0.25
0.45
0.2
0.21
0.6
0.7
0.62
0.9
0.4
0.3
0.5
0.31
0.25
0.98
102004 년 봄 물리학회성균관대학교 ( 수원 )
dkkPNk
max
)(1
dkkNk
max
1
11
~max Nk
kkP ~)(Scale-free network
3
dkkPkkk
max
0
22 )(
13
~~ 3max
2
Nkk
13
21
1~
1~
Nkfc
Prediction for our model
)(kP: degree distribution
Mean degree 4 kin this work
- We predict the critical value is proportional to the fluctuation of degree
13
21
1~
1~
Nkfc
3
- We predict the critical behavior of random network is similar to random neighbor model.
Random network
Scale-free network
( logarithmic)3
similar to random network
kkckk f 11
11
kcf 1~ )1( kk
112004 년 봄 물리학회성균관대학교 ( 수원 )
IV. ResultsRandom Network
10-6 10-5 10-4 10-30.3
0.4
0.5
0.6 N f
c(N)
-------------------------1000 0.425410000 0.3603100000 0.33221000000 0.3325
f c(N
)
1/N
3/1cf
1 10 100 1000
10-6
10-5
10-4
10-3
10-2
10-1
100
-1.650(8)
S
P(S
)
)8(650.1
4 kMean degree
510N
122004 년 봄 물리학회성균관대학교 ( 수원 )
Scale-free Network
3 4 kMean degree
10-6 10-5 10-4 10-3
0.01
0.1
=2.34 =2.75
0.26(1)
0.379(4)
f c(N
)
1/N
10-6 10-5 10-4 10-3
0.05
0.10
0.15
0.20
0.25
0.30 =3.01
f c(N
)
1/N
13
21
1~
Nfc
2462.0
1~N
0714.0
1~N
)34.2(
)75.2(
xc Nf
1~
Logarithmic(?)
132004 년 봄 물리학회성균관대학교 ( 수원 )
0 500 1000 1500 2000
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
1 10 100 1000
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
101
P(S
)
S
-1.88(1)
P(S
)
S
(?)0cf
34.2
510N
10-6 10-5 10-4 10-3
0.0
0.1
0.2
0.3
f c(N
)
1/N
------------------------------- 3.01 0.035(6) 2.75 0.021(5) 2.34 0.004(2)
)( Nfc
142004 년 봄 물리학회성균관대학교 ( 수원 )
0 2000 4000 6000 8000 1000010-7
10-6
10-5
10-4
10-3
10-2
10-1
100
1 10 100 1000 1000010-7
10-6
10-5
10-4
10-3
10-2
10-1
100
S
P(S
)-1.74(1)
S
P(S
)
75.2
01.3
510N
0 10000 2000010-6
10-5
10-4
10-3
10-2
10-1
100
1 10 100 1000 1000010-6
10-5
10-4
10-3
10-2
10-1
100
S
P(S
)
-1.64(1)
P(S
)
S
510N
152004 년 봄 물리학회성균관대학교 ( 수원 )
Scale-free Network
10-6 10-5 10-4 10-30.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
f c(N
)
1/N
3
1 10 100 1000 1000010-6
10-5
10-4
10-3
10-2
10-1
100
-1.65(1)
S
P(S
)
))1(15.0(finitefc
30.44 kMean degree
510N
162004 년 봄 물리학회성균관대학교 ( 수원 )
10-6 10-5 10-4 10-30.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
f c(
N)
1/N
))1(28.0(finitefc
68.5
1 10 100 100010-5
10-4
10-3
10-2
10-1
100
-1.64(2)
S
P(S
)
510N
172004 년 봄 물리학회성균관대학교 ( 수원 )
IV. Summary
cf
2.341.88(1)+exponenti
al
2.751.74(1)+exponenti
al
3.01 1.64(1)+(?)
4.30 1.65(1)
5.68 1.64(2)
(?)0cf
finitefc
SSP ~)(
cf
1.650(8)
SSP ~)(3/1cf
◆ Random Network
◆ Scale-free Network
◆ On SFN the system is not critical because all species are adapted.
3
◆ SFN is critical 3
◆ On SFN the system is critical and it’s behavior is similar to RN.
3
◆ is similar to Earthquake model on RN.
3, RNS.Lise and M.Paczuski PRL 88, 228301 (2002)