Upload
leo-kennedy
View
215
Download
1
Tags:
Embed Size (px)
Citation preview
Fat Tail Distributions and Efficiency of Fat Tail Distributions and Efficiency of Flow Processing on Complex NetworksFlow Processing on Complex Networks
Zoltán Toroczkai
Center for Nonlinear Studies, and Complex Systems Group, Theoretical Division, Los Alamos National Laboratory
LA-UR-03-5542 LANL LDRD-DR S.P.I.N. Project, 2003-06
What are Networks?
Interacting many “particle” systems where the interactions are propagated through a discrete structure, a graph (not a continuum).
Node (the “particle”) Link (edge)
Graph:
-- undirected
-- directed
The links [edges] represent interactions or associations between the nodes.
Where are Networks?
• Infrastructures:Infrastructures: transportation nw-s (airports, highways, roads, rail, water) energy transport nw-s (electric power, petroleum, natural gas)
• Communications:Communications: telephone, microwave backbone, internet, email, www, etc.
• Biology:Biology: protein-gene interactions, protein-protein interactions, metabolic nw-s, cell-signaling nw-s, the food web, etc.
• Social Systems:Social Systems: acquaintance (friendship) nw-s, terrorist nw-s, collaboration networks, epidemic networks, the sex-web
• Geology:Geology: river networks
Skitter data depicting a macroscopic snapshot of Internet connectivity, with selected backbone ISPs (Internet Service Provider) colored separately by K. C. Claffy email: [email protected] http://www.caida.org/Papers/Nae/
Communication NetworksCommunication Networks
Chemicals
Bio-Chemical reactions
Networks in BiologyNetworks in Biology
The metabolic pathway
Biochemical Pathways - Metabolic Pathways, Source: ExPASy
Networks in BiologyNetworks in Biology
Chemicals Bio-Chemical reactions
The metabolic pathway
The protein network
H. Jeong, S.P. Mason, A.-L. Barabasi, Z.N. Oltvai, Nature 411, 41-42 (2001)
P. Uetz, et al. Nature 403, 623-7 (2000).
proteins Binding
Structural properties: degree distributions and the scale-free character
Node degree: number of neighbors
Observation: networks found in Nature and human made, are in many cases “scale-free” (power-law) networks:
kkP )( kkP )(
i
Degree distribution, P(k): fraction of nodes whose degree is k (a histogram over the ki –s.)
ki=5
The Erdős-Rényi Random Graph (also called the binomial random graph)
),(, EVG pN
• Consider N nodes (dots).
• Take every pair (i,j) of nodes and connect them with an edge with probability p.
For the sake of definitions:
What is scale-free?
Poisson distribution
Non-Scale-free Network
Power-law distribution
Scale-free Network
=<k>
Capacity achieving degree distribution of Tornado code. The decay exponent -2.02.
M. Luby, M. Mitzenmacher, M.A. Shokrollahi, D. Spielman and V. Stemann, in Proc. 29th ACM Symp. Theor. Comp. pg. 150 (1997).
Erdős-Rényi Graph
Bacteria Eukaryotes
Archaea Bacteria Eukaryotes
Science citations www, out- and in- link distributions Internet, router level
Metabolic networkSex-web
Scale-free Networks: Coincidence or Universality?
• No obvious universal mechanism identified
•As a matter of fact we claim that there is none (universal that is).
• Instead, our statement is that at least for a large class of networks (to be specified) network structural evolution is governed by a selection principle which is closely tied to the global efficiency of transport and flow processing by these structures, and
• Whatever the specific mechanism, it is such as to obey this selection principle.
Need to define first a flow process on these networks.
Z. Toroczkai and K.E. Bassler, “Jamming is Limited in Scale-free Networks”, Nature, 428, 716 (2004)
Z. Toroczkai, B. Kozma, K.E. Bassler, N.W. Hengartner and G. Korniss “Gradient Networks”, http://www.arxiv.org/cond-mat/0408262
Gradient Flow NetworksGradient Flow Networks
Ex.:
Y. Rabani, A. Sinclair and R. Wanka, Proc. 39th Symp. On Foundations of Computer Science (FOCS), 1998: “Local Divergence of Markov Chains and the Analysis of Iterative Load-balancing Schemes”
Load balancing in parallel computation and packet routing on the internet
Gradients of a scalar (temperature, concentration, potential, etc.) induce flows (heat, particles, currents, etc.).
Naturally, gradients will induce flows on networks as well.
Setup:
Let G=G(V,E) be an undirected graph, which we call the substrate network.
}1,...,2,1,0{},...,,{ 110 NxxxV N The vertex set:
loops)-self (no ),,( , , ExxjixxeEeVVE ji The edge set:
A simple representation of E is via the Nx N adjacency (or incidence) matrix AA
Eji
EjiaxxA ijji ),( if 0
),( if 1),(
Let us consider a scalar field Vh :}{
Set of nearest neighbor nodes on G of i :)1(
iS
(1)
Definition 1 The gradient h(i) of the field {h} in node i is a directed edge:
))(,()( iiih
Which points from i to that nearest neighbor }{)1( iSi for G for which the increase in the
scalar is the largest, i.e.,:
)(maxarg)(}{)1(
jiSj
hii
The weight associated with edge (i,) is given by:
ihhih )(
)(),()( then )( If iiiihii 0 The self-loop )(i0.. is a loop through i
with zero weight.
Definition 2 The set F of directed gradient edges on G together with the vertex set V forms the gradient network:
),( FVGG
(3)
(2)
If (3) admits more than one solution, than the gradient in i is degenerate.
In the following we will only consider scalar fields with non-degenerate gradients. This means:
0}),( if {Prob. Ejihh ji
Theorem 1 Non-degenerate gradient networks form forests.
Proof:
Theorem 2 The number of trees in this forest = number of local maxima of {h} on G.
In-degree distribution of the Gradient Network when In-degree distribution of the Gradient Network when G=GG=GN,pN,p . . A A
combinatorial derivationcombinatorial derivationIn-degree distribution of the Gradient Network when In-degree distribution of the Gradient Network when G=GG=GN,pN,p . . A A
combinatorial derivationcombinatorial derivation
Assume that the scalar values at the nodes are i.i.d, according to some distribution (h).
First, distribute the scalars on the node set V, then find those link configurations which contribute to R(l) when building the GN,p graph.
Without restricting the generality, calculate R(l) for node 0.
Consider the set of nodes with the property 0hh j
Let the number of elements in this set be n, and the set be denoted by [n].
The complementary set of [n] in V\{0} is :][nC
Version: Balazs Kozma (RPI)
lnpp
)1(
nlNnpp
1
)1(1
In order to have exactly l nodes pointing their gradient edges into 0:
• they have to be connected to node 0 on the substrate
• they must NOT be connected to the set [n]
For l nodes:
Also need to require that no other nodes will be pointing their gradient directions into node 0 :
(Obviously none of the [n] will.)
So, for a fixed h0 and a specific set [n] :
nlNnln ppppl
nN
1 )1(1)1(
1
The probability Qn for such an event for a given n while letting h-s vary according to their distribution:
0
)( )( 0
h
hdhh
nNn hh 1 0
0 )(1)(
N
hhhdhn
NQ nNn
n
1)(1)( )(
1 1 0
000
For one node to have its scalar larger than h0:
For exactly n nodes:
Thus:
Combining:
nlNnlnN
nnN pppp
l
nNQlR
1
1
0
)1(1)1(1
)(
Finally:
lnlnNnN
nN pppp
l
nN
NlR
1 1
0
)1()1(111
)(
, 1 , 1
)(
: 1 , . , ,0limit In the
Npzlzl
lR
zconstNpzNp
N
lnlnNnN
nN pppp
l
nN
NlR
11
0
)1()1(111
)(
What happens when the substrate is a scale-free network?
Gradient Networks and Transport Efficiency
- every node has exactly one out-link (one gradient direction) but it can have more than one in-link (the followers)
- the gradient network has N-nodes and N out-links. So the number of “out-streams” is Nsend = N
- the number of RECEIVERS is
1
)(
l
inlreceive NN
)0(11)(
01
)(
N
Gh
in
Gh
l
inl
Ghsend
receive RN
N
N
N
N
NJ
- J is a congestion (pressure) characteristic.
- 0 J 1. J=0: minimum congestion, J=1: maximum congestion
1
1
1)1(1
1),(,
N
n
nNnG ppN
pNJ pN
11
1
11
ln
ln1),(,
N
O
pN
NpNJ pNG
In the scaling limit , , const. Np
- for large networks we get maximal congestion!
In the scaling limit , , ,0 zpNNp
1
0
)()(1
),(, zzeG zeEizEiz
edxpNJzx
pN
1 ...ln
1),( 1,
zG
z
CzpNJ pN
- becomes congested for large average degree.
- For scale-free structures, the congestion factor becomes independent on the system (network) size!!
For LARGE and growing networks, where the conductance of edges is the same, and the flow is generated by gradients, scale-free networks are more likely to be scale-free networks are more likely to be selected during network evolution than scaled structuresselected during network evolution than scaled structures.
For LARGE and growing networks, where the conductance of edges is the same, and the flow is generated by gradients, scale-free networks are more likely to be scale-free networks are more likely to be selected during network evolution than scaled structuresselected during network evolution than scaled structures.
The Configuration model
A. Clauset, C. Moore, Z.T., E. Lopez, to be published.
K-th Power of a Ring
Generating functions: i
ki zkzg )(
1
0 )1(
)()1(1 )(
g
xgxzgdxzR
Degree distribution of the gradient network for the K-th power of a ring Degree distribution of the gradient network for the K-th power of a ring
ijijij ab
hklN inlR )(0,
)(
)(1)( 0 jijiji hhbbjH
:let then , and ,, If )1(0SiVji
So:
1
10
1
0
1
1
1
10
)(0 )(1)(
N
jjijij
N
i
N
ji
N
ii
in hhbbjHak
)1(][1,
11
0 0 1 )(
11)(
NPbl
N
n nni
n
j jin
N
NlR
0 1
)(1i
n
jjib
n
jjnn ST
1
)1()(
where )(),...,1( nn is an n-subset of the set {1,2,…,N-1}.
)1( NPn denotes the set of all possible n-subsets of {1,2…,N-1}.
n
NNPn
1)1(
is always zero, if there is a node from the n-subset connected to i, or i belongs to the n-subset.
Let which is the union of the disks of all nodes from the n-subset.
Thus, one needs to find the number of coverings of the ring with n disks, each of radius K, that misses exactly l nearest neighbors of the origin.
KlK
KlKlKlKlK
K
KlKKKK
KK
KllKlKlKlK
KlKK
lR K
2 ,14
1
121 ,)32)(22)(12(
124
,)33)(23)(13(3
7726
11 ,)32)(22)(12)(2(
24934
)(
2
2
)2(
2K+l
Power law with exponent =- 3
Competition Games on Networks
Collaboration with:• Marian Anghel (CCS-3)
• Kevin E. Bassler (U. Houston)
• György Korniss (Rensselaer)
References:
M. Anghel, Z. Toroczkai, K.E. Bassler and G. Korniss, Competition-driven Network Dynamics: Emergence of a Scale-free Leadership Structure and Collective Efficiency, Phys.Rev.Lett. 92, 058701 (2004)
Z. Toroczkai, M. Anghel, G. Korniss and K.W. Bassler, Effects of Inter-agent Communications on the Collective, in Collectives and the Design of Complex Systems, eds. K. Tumer and D.H. Wolpert, Springer, 2004.
Resource limitations lead in human, and most biological populations to competitive dynamics.
The more severe the limitations, the more fierce the competition.
Amid competitive conditions certain agents may have better venues or strategies to reach the resources, which puts them into a distinguished class of the “few”, or elites.
Elites form a minority group.
In spite of the minority character, the elites can considerably shape the structure of the whole society:
since they are the most successful (in the given situation), the rest of the agents will tend to follow (imitate, interact with) the elites creating a social structure of leadership in the agent society.
Definition: a leader is an agent that has at least one follower at that moment. The influence of a leader is measured by the number of followers it has. Leaders can be following other leaders or themselves.
The non-leaders are coined “followers”.
The El Farol bar problemThe El Farol bar problem
A B
…
[W. B Arthur(1994)]
A binary (computer friendly) version of the El Farol bar problem:
[Challet and Zhang (1997)]
The Minority Game (MG)The Minority Game (MG)
A = “0” (bar ok, go to the bar)
B = “1” (bar crowded, stay home)
World utility(history): (011..101)
latest bit
m bits
l {0,1,..,2m-1}
(Strategies)(i) =
S(i)1(l)
S(i)2(l)
S(i)
S(l)
(Scores)(i) = C (i)(k), k = 1,2,..,S.
(Prediction) (i) =)}({max )(* kCk i
k }1,0{)()( )(
* lSiP i
k
3-bit history 000 001 010 011 100 101 110 111
associated integ.
0 1 2 3 4 5 6 7
Strategy # 1 0 0 0 1 1 0 0 1
Strategy #2 1 1 0 0 1 0 0 0
Strategy #3 1 1 1 0 0 0 1 0
t
A(t)
Attendance time-series for the MG:
World Utility Function:
2)2/( NA
Agents cooperate if they manage to produce fluctuations below (N1/2)/2 (RCG).Scaling variable:
NN
P m2
The El Farol bar game on a social networkThe El Farol bar game on a social network
…
A B
The Minority Game on Networks (MGoN)The Minority Game on Networks (MGoN)
Agents communicate among themselves.
Social network:Social network: 2 components:
1) Aquintance (substrate) network: G (non-directed, less dynamic)
2) Action network: A (directed and dynamic)
G
AA G
Communication types (more bounded rationality):
Majority ruleMajority rule Minority ruleMinority rule
Critic’s ruleCritic’s rule: an agent listens to the OPINION/PREDICTION of all neighboring agents on G, scores them (self included) based on their past predictions, and ACTS on the best score.
(not rational)
(not rational)
(more rational, uses reinforcement learning)
(Links)(i) = (Scores)(i) = F (i)(j), j= 1,2,..,K.
(Prediction) (i) =)}({max )(* jFj i
j }1,0{)()( )(
*
lSiP j
k
)(1iL)(
2iL
)(iKi
L
i
Emergence of scale-free leadership structure:
Emergence of scale-free leadership structure:
Robust leadership hierarchy
RCG on the ER network produces the scale-free backbone of the leadership structure
1for ,1);,(
);,()();,(
)();,(
);,();,(
0
1
1
mpmNf
pmNfkpapmNN
papmNN
pmNNkpmNN
kk
k
kk
k
iouti
The influence is evenly distributed among all levels of the leadership hierarchy.
m=6
Structural un-evenness appears in the leadership structure for low trait diversity.
The followers (“sheep”) make up most of the population (over 90%) and their number scales linearly with the total number of agents.
Network Effects: Improved Market EfficiencyNetwork Effects: Improved Market Efficiency
A networked, low trait diversity system is more effective as a collective than a sophisticated group!
Can we find/evolve networks/strategies that achieve almost perfect volatility given a group and their strategies (or the social network on the group)?
, 1 , 1
)(
: 1 , . , ,0limit In the
Npzlzl
lR
zconstNpzNp
N
Conclusions Conclusions :
• We defined Gradient Networks as directed sub-graphs formed by local gradients of a scalar distributed on a substrate graph G.
• When the gradient direction is unique these Gradient Networks form forests.
• Gradient Networks typically arise when there is a local extremizing dynamics at the node level (Agent-based Systems such as markets, routers, parallel computers, etc..)).
• Gradient Networks can be scale-free graphs even on substrate networks that are NOT scale-free networks (such as E-R graphs)!!
• Gradient Networks can be highly dynamic, their evolution driven by the dynamics of the scalar field on G and they are not solely defined through the topological properties of G!! (such as in the case of preferential attachment).
• G. N.-s give a natural explanation for why scale-free large networks might emerge if the edges have the same conductance and the flows are generated by gradients.