LA - W12 Diag, Split, A (λ) &g (λ)

WEEK 12
Diagonalization (II)
Kwok-Wing Tsoi
In the previous week, we proved that a matrix A ∈ Mn (F ) is diagonalizable if and only if
there exists a basis of F n which consists of eigenvectors of A (or the existence of an ‘eigen-
basis’). However, establishing the existence of an eigen-basis requires explicit computation
of eigenspaces, which could be fairly clumsy. Therefore, we wish to develop alternative,
and hopefully simpler criterion to determine the diagonalizability of a matrix. It turns out
that one such criterion can be formulated in the case when F = C.
12.1 Splitting of polynomials

What is so special about the field C ? To see this, we introduce a concept regarding
factorization of polynomials.
Definition 12.1.1. Let F be a field. A non-zero polynomial f (x) ∈ F [x] is said to

split over F if it can be written as a product of linear factors. i.e.
f (x) = a · (x − α1 )(x − α2 )...(x − αn )
where a, α1 , ..., αn ∈ F .
In other words, a polynomial f (x) splits over F if we can factorize f (x) completely
into a product linear factors (these linear factors can be ‘repeated’, as shown below).
Example. Let f (x) = x4 + 2x2 + 1.
• f (x) = (x2 + 1)2 does not split over R because we cannot factorize x2 + 1 any
further over R.
√ √
• f (x) = (x + −1)2 (x − −1)2 splits over C.
What distinguishes the field of complex numbers C from other fields is the following
famous, yet difficult, fact from algebra
Theorem 12.1.1 (Fundamental Theorem of Algebra).

Every non-zero polynomial in C[x] splits over C.
Proof. Omitted. A standard proof of FTA is usually discussed in a first course of

complex analysis.
In literature, we say that C is
an algebraically closed field.
116
12.2 Algebraic and geometric multiplicities The theory works equally well
when C is replaced by any alge-
Now we focus on matrices over C or linear operators of a complex vector space V .
braically closed field. (Do you
Definition 12.2.1. Let A ∈ Mn (C). Write know any other than C?)
pA (x) = (−1)n · (x − λ1 )r1 (x − λ2 )r2 ...(x − λk )rk
where λ1 , ..., λk are distinct. For each eigenvalue λ = λi ,
1. the algebraic multiplicity of λi , denoted by a(λi ), is defined to be the number of

copies of (x − λi ) in the factorization of pA (x). i.e.
def
a(λi ) = ri .
2. the geometric multiplicity of λi , denoted by g(λi ), is defined by
def
g(λi ) = nullity(A − λi In )
Example. Find the algebraic and geometric multiplies of the eigenvalues of

 
1 1 0
A = 1 2 1 .
 
1 −1 2
– 117 –
Now we can extend the list of invariants for similar matrices.
Theorem 12.2.1. If A, B ∈ Mn (C) are similar, then they have the same
1. characteristic polynomials;
2. set of eigenvalues (counted with multiplicities);
3. algebraic multiplicities for each eigenvalue;
4. geometric multiplicities for each eigenvalue.
Proof. We proved (1) last time. (2) and (3) follow immediately. Proving (4) is left
as an exercise (Hint. use Rank inequality).
It turns out that the geometric multiplicity of an eigenvalue is always bounded from
above by its algebraic multiplicity.
Theorem 12.2.2. Let A ∈ Mn (C). For each eigenvalue λ of A, we have
1 ≤ g(λ) ≤ a(λ).
– 118 –
12.3 Diagonalizability criterion II : a(λ) = g(λ)
In the case when we deal with complex matrices (or matrices over an algebraically closed
field), diagonalizability can be characterized by the algebraic and geometric multiplies.
Theorem 12.3.1. Let A ∈ Mn (C). The following are equivalent.
1. A is diagonalizable (over C).
2. There exists a basis {v1 , v2 , ..., vn } of Cn where each vi is an eigenvector of A.
3. For each eigenvalue λ of A, we have a(λ) = g(λ).
We have proved (1) ⇔ (2). It suffices to prove (1) implies (3) and (3) implies (2). At
the outset, we suppose A ∈ Mn (C) and
pA (x) = (−1)n (x − λ1 )r1 (x − λ2 )r2 · · · (x − λk )rk
where λ1 , ..., λk are distinct.

(1) implies (3).
– 119 –
(3) implies (2).
– 120 –
12.4 Applications : Finding An (Method 1)
As an application of diagonalization, we can compute a high power of a matrix efficiently.
Theorem 12.4.1. Let A, Q ∈ Mn (F ) for which Q is invertible.
(Q−1 AQ)n = Q−1 · An · Q.
Example. In Homework 10, we consider the Fibonacci sequence where F0 = 0, F1 = 1

and Fn = Fn−1 + Fn−2 for n ≥ 2 and proved that
!n !
1 1 Fn+1 Fn
A =
n
= .
1 0 Fn Fn−1
(a) Find an invertible Q such that Q−1 AQ is diagonal.
(b) Write down An .
(c) Hence derive a formula (in n) for the n-th Fiboncci number Fn .
– 121 –

LA - W12 Diag, Split, A (λ) &g (λ)

Uploaded by

Copyright:

Available Formats

LA - W12 Diag, Split, A (λ) &g (λ)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

LA - W12 Diag, Split, A (λ) &g (λ)

Uploaded by

Copyright:

Available Formats

WEEK 12

12.1 Splitting of polynomials

Definition 12.1.1. Let F be a field. A non-zero polynomial f (x) ∈ F [x] is said to

f (x) = a · (x − α1 )(x − α2 )...(x − αn )

Example. Let f (x) = x4 + 2x2 + 1.

Theorem 12.1.1 (Fundamental Theorem of Algebra).

Proof. Omitted. A standard proof of FTA is usually discussed in a first course of

pA (x) = (−1)n · (x − λ1 )r1 (x − λ2 )r2 ...(x − λk )rk

where λ1 , ..., λk are distinct. For each eigenvalue λ = λi ,

1. the algebraic multiplicity of λi , denoted by a(λi ), is defined to be the number of

2. the geometric multiplicity of λi , denoted by g(λi ), is defined by

Example. Find the algebraic and geometric multiplies of the eigenvalues of

2. set of eigenvalues (counted with multiplicities);

3. algebraic multiplicities for each eigenvalue;

4. geometric multiplicities for each eigenvalue.

Theorem 12.2.2. Let A ∈ Mn (C). For each eigenvalue λ of A, we have

Theorem 12.3.1. Let A ∈ Mn (C). The following are equivalent.

1. A is diagonalizable (over C).

2. There exists a basis {v1 , v2 , ..., vn } of Cn where each vi is an eigenvector of A.

3. For each eigenvalue λ of A, we have a(λ) = g(λ).

pA (x) = (−1)n (x − λ1 )r1 (x − λ2 )r2 · · · (x − λk )rk

where λ1 , ..., λk are distinct.

Theorem 12.4.1. Let A, Q ∈ Mn (F ) for which Q is invertible.

(Q−1 AQ)n = Q−1 · An · Q.

Example. In Homework 10, we consider the Fibonacci sequence where F0 = 0, F1 = 1

(a) Find an invertible Q such that Q−1 AQ is diagonal.

(b) Write down An .

You might also like