Scribd
Upload
125 views

Percolation: Theory and Applications: Daniel Genin, NIST

This document discusses percolation theory and its applications. It begins with an introduction to percolation theory and its origins modeling water flow through porous rocks. It then covers the basic setup and results of percolation models, including defining key quantities like the percolation probability and critical probability. Finally, it provides an example application to modeling network robustness using site percolation on random graphs.

Uploaded by

Anonymous j6r5KRtrH2
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
125 views

Percolation: Theory and Applications: Daniel Genin, NIST

This document discusses percolation theory and its applications. It begins with an introduction to percolation theory and its origins modeling water flow through porous rocks. It then covers the basic setup and results of percolation models, including defining key quantities like the percolation probability and critical probability. Finally, it provides an example application to modeling network robustness using site percolation on random graphs.

Uploaded by

Anonymous j6r5KRtrH2
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 24

Percolation: Theory and

Applications

Daniel Genin, NIST

October 17, 2007


OUTLINE

• Introduction/Setup

• Basic Results

• Example of Application

1
Introduction

Original problem: Broadbent and Hammers-


ley(1957)

Suppose a large porous rock is submerged un-


der water for a long time, will the water reach
the center of the stone?

Related problems:

How far from each other should trees in an or-


chard (forest) be planted in order to minimize
the spread of blight (fire)?

How infectious does a strain of flu have to be


to create a pandemic? What is the expected
size of an outbreak?

2
Setup: 2D Bond Percolation

• Stone: a large two dimensional grid of chan-


nels (edges). Edges in the grid are open or
present with probability p (0 ≤ p ≤ 1) and
closed or absent with probability 1 − p.

• Pores: open edges and p determines the


porosity of the stone.

A contiguous component of the graph of open


edges is called an open cluster. The water will
reach the center of the stone if there is an open
cluster joining its center with the periphery.

Similarly, in the orchard example, p is the prob-


ability that blight will spread to an adjacent
tree and minimizing the spread corresponds to
minimizing the size of the largest open cluster.
3
Setup: 2D Bond Percolation

p=0.25 p=0.48

p=0.52 p=0.75
4
Setup: Bond Percolation

General Bond Percolation Model

• The space of the model is Zn or any infinite


graph.

• The edges are open or closed with probabil-


ity p, which may depend on the properties
of the edge (e.g. degree).

• Open cluster is a connected component of


the open edge graph.

• The network is said to percolate if there


is an infinite open cluster containing the
origin.

If the graph is translation invariant there is no


difference between the origin and any other
vertex.
5
Setup: Site Percolation

Site Percolation Model

• The space of the model is Zn or any infinite


graph.

• The vertices are open or closed with prob-


ability p, which may depend on the proper-
ties of the vertex (e.g. degree).

• Open cluster is a connected component of


the open vertex graph.

• The network is said to percolate if there


is an infinite open cluster containing the
origin.

Every bond percolation problem can be real-


ized as a site percolation problem (on a differ-
ent graph). The converse is not true.
6
Setup: Why Percolation?

• Percolation provides a very simple model


of random media that nevertheless retains
enough realism to make its predictions rel-
evant in applications.

• It is a test ground for studying more com-


plicated critical phenomena and a great source
of intuition.

7
Basic Results: Quantities of Interest

• |C| — the size of the open cluster at 0,


where C stands for the open cluster itself;

• θ(p) — percolation probability, defined as

θ(p) = Pp(|C| = ∞);

8
Basic Results: Percolation Probability

Exact shape of θ(p) is not known but it is be-


lieved to be a continuous function of p

θ( p)

p c (d) 1 p

Percolation thus has three distinct phases


1) subcritical if p < pc
2) critical if p = pc
3) supercritical if p > pc

9
Basic Results: Quantities of Interest

• |C| — the size of the open cluster at 0,


where C stands for the open cluster itself;

• θ(p) — percolation probability, defined as

θ(p) = Pp(|C| = ∞);

• pc(d) — critical probability, defined as

pc(d) = sup{p : θ(p) = 0};

10
Basic Results: Critical Probability

Theorem. If d ≥ 2 then 0 < pc(d) < 1.

The exact value of pc(d) is known only for a


few special cases:

• pbond
c (1) = psite
c (1) = 1

• pbond
c (2) = 1/2, psite
c (2) ≈ .59

• pbond
c (triangular lattice) = 2 sin(π/18)

• pbond
c (hexagonal lattice) = 1 − 2 sin(π/18)

Theorem. Probability that an infinite open clus-


ter exists is 0 if p < pc(d) and 1 if p > pc(d).

It is known that no infinite open cluster exists


for p = pc(d) if d = 2 or d ≥ 19.
11
Basic Results: Critical Probability

Some bounds on the critical probability are


known
Theorem. If G is an infinite connected graph
and maximum vertex degree ∆ < ∞. The crit-
ical probabilities of G satisfy
1
≤ pbond
c ≤ psite
c ≤ 1 − (1 − pbond
c )∆ .
∆−1
In particular, pbond
c ≤ psite
c and strict inequality
holds for a broad family of graphs.

12
Basic Results: Quantities of Interest

• |C| — the size of the open cluster at 0,


where C stands for the open cluster itself;

• θ(p) — percolation probability, defined as


θ(p) = Pp(|C| = ∞);

• pc(d) — critical probability, defined as


pc(d) = sup{p : θ(p) = 0};

• χ(p) — the mean size of the open cluster


at the origin, defined as
χ(p) = Ep[|C|];

• χf (p) — the mean size of the finite open


cluster at the origin, defined as
χf (p) = Ep[|C| : |C| < ∞];

13
Basic Results: Subcritical Phase

If p < pc all open clusters are finite with prob-


ability 1.
Theorem. Probability of a cluster of size n at 0
decreases exponentially with n. More precisely,
there exists α(p) > 0, α(p) → ∞ as p → 0 and
α(pc) = 0 such that

Pp(|C| = n) ≈ e−nα(p) as n → ∞

This also implies that χ(p) is finite for all p in


the subcritical region.
Theorem. Probability distribution for cluster
radii decays exponentially with the radius, i.e.

Pp(0 ↔ ∂B(r)) ≈ e−r/ξ(p)


where ξ(p) — the characteristic length of ex-
ponential decay — is the mean cluster radius.

14
Basic Results: Supercritical Phase

If p > pc, with probability 1 at least one infinite


open cluster exists.
Theorem. The infinite open cluster is unique
with probability 1.
Theorem. Probability of a finite open cluster
of size n at 0 decreases exponentially with n.
More precisely, there exist functions β1(p) and
β2(p), satisfying 0 < β2(p) ≤ β1(p) < ∞, such
that

exp(−β1(p)n(d−1)/d) ≤ Pp(|C| = n)
≤ exp(−β2(p)n(d−1)/d)

Because χ(p) is infinite for p > pc the truncated


mean — χf (p) — over finite clusters only is
considered.

15
Basic Results: χ(p)

The general shape of χ(p) is believed to be as


follows

f
χ( p) χ ( p)

p c (d) 1 p

16
Application: Network Robustness and
Fragility

Problem: How many random nodes can be re-


moved before a network looses connectivity?
How many of the highly connected nodes can
be removed before the network looses connec-
tivity?

Use site percolation model on a random graph


with a given degree distribution pk and ver-
tex occupation probability qk depending on the
vertex degree.

Allowing qk to vary with k allows to study vari-


ous types of attacks: random if qk = q is inde-
pendent of k, targeted deletion of high degree
nodes if qk = H(kmax − k).

17
Application: Network Robustness and
Fragility

Using formalism of generating functions it can


be shown that the generating function H0 of
cluster size |C| at a random vertex satisfies

H0(x) = 1 − F0(1) + xF0(H1(x))


1
H1(x) = 1 − (F0′ (1) + xF0′ (H1(x)))
z
∞
pk q k x k
X
F0(x) =
k=0
and z is the mean graph degree, and

χ(q) = H0′ (1)

18
Application: Network Robustness and
Fragility

Although closed form solutions to the above


equations do not exist in general, it is possible
to compute H0 to any degree of accuracy by it-
erating equations for H1 and then substituting
into the equation for H0.

In the case qk = q (uniform distribution) it can


be shown that χ(q) diverges at
1
qc =
G′′(1)
where G = 1z k pk xk . This is the percolation
P

threshold probability.

19
Application: Network Robustness and
Fragility

(
0 if k = 0
pk =
Ck−τ e−k/κ if k ≥ 1

20
Application: Network Robustness and
Fragility

If the highest degree vertices are removed first,


qk = H(kmax − k), the probability that a ran-
dom vertex does not belong to the giant open
cluster is

S = 1 − H0(1) = F0(1) − F0(u)


where u solves
1 ′
u = 1 − (F (1) + F ′(u))
z
These equations can be solved numerically.

21
Application: Network Robustness and
Fragility

22
Bibliography

Reka Albert and Albert Laszlo Barabasi, Sta-


tistical mechanics of complex networks, Re-
views of modern Physics, 74, Jan. 2002.

Duncan Callaway, M. E. J. Newman, Steven H.


Strogatz, and Duncan J. Watts, Network ro-
bustness and fragility: Percolation on random
graphs, arXiv:cond-mat/0007300, Oct. 2000.

Geoffrey Grimmet, Percolation, Grundlehren der


mathematischen Wissenschaft, vol 321, Springer,
1999.

M. E. J. Newman, The structure and function


of complex networks, arXiv:cond-mat/0303516v,
Mar. 2003.

23

You might also like