Scribd
Upload
56 views2 pages

Perron Forbinius

This document summarizes key claims from the Perron-Frobenius theorem about the largest eigenvalue (λ1) and its associated eigenvector for non-negative matrices. It proves: 1) No entry in an eigenvector associated with λ1 can be zero. 2) λ1 has multiplicity 1 (is a simple root). 3) All entries in an eigenvector associated with λ1 must have the same sign. 4) If an eigenvector has all positive entries, its associated eigenvalue must be λ1.

Uploaded by

Michael Twito
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
56 views2 pages

Perron Forbinius

This document summarizes key claims from the Perron-Frobenius theorem about the largest eigenvalue (λ1) and its associated eigenvector for non-negative matrices. It proves: 1) No entry in an eigenvector associated with λ1 can be zero. 2) λ1 has multiplicity 1 (is a simple root). 3) All entries in an eigenvector associated with λ1 must have the same sign. 4) If an eigenvector has all positive entries, its associated eigenvalue must be λ1.

Uploaded by

Michael Twito
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/ 2

A closer look at the Perron-Frobenius theorem

Michael Twito
July 21, 2018

1 Introduction
This note as written due Graph Theory course, here we shall take a differ-
ent route and work with the alternative maximisation seen in the Rayleigh
principle

2 Claims:
Claim: If 0 6= x ∈ ε(λ1 ) then xi = 0 for no ∈ [n].
Proof. Let us define y = (|x1 |, ..., |xn |) as to have
X
||y||2 = x2i = ||x||2
i

Then
y T Ay ≥ xT Ax = λ1 ||x||2 = λ1 ||y||2
implying that y ∈ ε(λ1 ). Once we establish that y is an eigenvector associ-
ated with λ1 and by definition

∀i ∈ [n], yi ≥ 0

Suppose ∃i ∈ [n] s.t yi = 0. As y 6= 0 there is a j ∈ [n] s.t yj 6= 0.


Let P be an ij-path in G. As the ends of P are assigned one a zero and one
a non-zero value by y there must be an edge P say (k, l) s.t yk = 0 yet
yl 6= 0 and then yl ≥ 0. Then
X
0 = λ1 yk = yv ≥ yl > 0
v∼k

a contradiction.

1
Claim: λ1 has multiplicity 1 (i.e λ1 is simple).
Proof. Let x1 ∈ εA (λ1 ) and x2 ∈ εA (λ2 ) be independent. We know that no
entry in x1 is zero.Consequently we may choose a constant c s.t the first entry
of the vector x = x2 − cx1 = 0. Then

Ax = A(x2 − cx1 ) = λ2 x2 − cλ1 x1

Consequently we see that if λ1 = λ2 then x ∈ εA (λ1 ) but as x has a zero


entry this contradicts claim.

Claim: Let 0 6= x ∈ εA (λ1 ). ∀xi ∈ x xi have the same sign.


Proof. Let 0 6= x ∈ εA (λ1 ) and let y = (|x1 |, ..., |xn |); We know that y ∈
εA (λ1 ). We also know that no entry of x is zero and that dimεA (λ1 ) = 1(all
eigaenvalues of A are semi-simple) it follows that x = y or x = −y.

Claim: Suppose x ∈ εA (λ) for some eigenvalue λ of A and that xi > 0


∀i ∈ [n], Then λ = λ1
Proof. By the last Claim, ∃u ∈ εA (λ1 ) s.t ∀xi , xj |i, j ∈ [n] xi and xj have the
same sign. If λ 6= λ1 then u ⊥ x then at least one of those vectors have a
negaive or positive entry (according to our vector sign that we choose in the
assuption )

You might also like