Lec01 network flows
-
Upload
danny-luk -
Category
Technology
-
view
722 -
download
1
description
Transcript of Lec01 network flows
Lecture 1: Network Flows
Wai-Shing Luk (陆伟成)
Fudan University
2012年 8月 11日
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 1 / 13
Basic Elements of Network
Definition
[Network] A network is a collection of finite-dimensional vector spaces of
nodes and edges
V = {v1, v2, · · · , vN}, where |V | = N.
E = {e1, e2, e3, · · · , eM} where |E | = M
which satisfies 2 requirements:
1 The boundary of each edge is comprised of the union of nodes
2 The intersection of any edges is either empty or a boundary node of
both edges.
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 2 / 13
Orientation
Definition (Orientation)
An orientation of an edge is an ordering of its boundary node {s, t}, where
s is called a source/initial node
t is called a target/terminal node
Definition (Coherent)
Two orientations to be the same is called coherent
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 3 / 13
Incidence Matrix AT
Definition (Incidence matrix)
A(i , j) =
+1 if vj is the initial node of ei
−1 if vj is the terminal node of ei
0 if vj is not the boundary node of ei
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 4 / 13
Chain (τ)
An edge (node) chain τ is an M (N)-tuple of scalar which assigns a
coefficient to each edge (node), where M (N) is the number of
distinct edges (nodes) in the network.
A chain may be viewed as an (oriented) indictor vector representing a
set of edges (nodes).
Example
[0, 0, 1, 1, 1], [0, 0, 1,−1, 1]
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 5 / 13
Discrete Boundary Operator ∂
The incidence matrix natually maps edges into their corresponding
boundary elements:
β = AT τ
A chain is called a cycle if it is in the nullspace of the boundary
operator, i.e.
AT τ = 0
A chain β is called a boundary of τ if it is in the range of the
boundary operator
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 6 / 13
Co-boundary Operator d
Definition (Co-boundary)
The co-boundary (or differential) operator
d = ∂∗ = (AT )∗ = A
Note
Nullspace of A is #components of a graph
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 7 / 13
Generalized Stokes’ Theorem
Conventional (integration):∫S
d ω̃ =
∮∂Sω̃
Discrete (pairing)
[τ,Aω] = [AT τ, ω]
where
τi =
1 if ei ∈ S ,
0 otherwise .
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 8 / 13
Fundamental Theorem of Calculus
Conventional (integration):∫ b
af (t)dt = F (b)− F (a)
Discrete (pairing)
[τ1,Ac0] = [AT τ1, c0]
2 4 3 5
7
c0
-1 -1 1
1
AT τ1
2 -1 2
4Ac0
1 1 1
1τ1
7
7
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 9 / 13
Divergence and Flow
Definition (Divergence)
div x = AT x
Definition (Flow)
x is called a flow if∑
div x = 0, where all positive entries of (div x) are
called sources and negative entries are called sinks.
Definition (Circulation)
A network is called a circulation if there is no source or sink. In other
words, div x = 0
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 10 / 13
Tension and Potential
Definition (Tension)
A tension (in co-domain) y is a differential of a potential u, i.e. y = Au.
Theorem (Tellgen’s)
Flow and tension are bi-orthogonal (isomorphic).
Proof.
0 = [AT x , u] = (AT x)Tu = xT (Au) = xT y
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 11 / 13
Path (P)
Definition (Path indicator vector)
A path indicator vector τ of P that
τi =
1 if ei ∈ P ,
0 otherwise .
Theorem
[total tension y on P] = [total potential on the boundary of P].
Proof.
yT τ = (Au)T τ = uT (AT τ) = uT (∂P) .
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 12 / 13
Cut (Q)
Definition (Cut)
Two node sets S and S ′ (the complement of S , i.e. V − S). A cut Q is an
edge set, denoted by [S , S ′]−. A cut indicator vector q (oriented) of Q is
defined as Ac where
ci =
1 if vi ∈ S ,
0 otherwise .
Theorem (Stokes’ theorem!)
[total divergence of x on S] = [total x across Q].
Proof.
(div x)T c = (AT x)T c = xT (Ac) = xTq .
W.-S. Luk (Fudan Univ.) Lecture 1: Network Flows 2012年 8月 11日 13 / 13