Scribd
Upload
15 views

M235 Lect 10

This document summarizes key concepts in linear algebra, including: 1) The range and null space of a matrix A define its rank and nullity, which are related by the rank-nullity theorem. 2) The trace of a matrix is the sum of its diagonal elements, while the determinant can be defined in several equivalent ways and is used to determine invertibility. 3) The eigenvalues and eigenvectors of a matrix A are used to diagonalize A, with similar matrices having the same eigenvalues, trace, and determinant.

Uploaded by

Clinton Batchanki
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
15 views

M235 Lect 10

This document summarizes key concepts in linear algebra, including: 1) The range and null space of a matrix A define its rank and nullity, which are related by the rank-nullity theorem. 2) The trace of a matrix is the sum of its diagonal elements, while the determinant can be defined in several equivalent ways and is used to determine invertibility. 3) The eigenvalues and eigenvectors of a matrix A are used to diagonalize A, with similar matrices having the same eigenvalues, trace, and determinant.

Uploaded by

Clinton Batchanki
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Linear Algebra

1 Range and null space


For A ∈ Mm×n (C), define its range

ran A := {Ax, x ∈ Cm },

and its null space


ker A := {x ∈ Cn : Ax = 0}.
These are linear subspaces of Cm and Cn , respectively. The rank and the nullity of A are
defined by
rk A := dim(ran A), nul A := dim(ker A).
They are deduced form one another by the rank-nullity theorem

rk A + nul A = n.

Recall that A ∈ Mm×n (C) is injective if ker A = {0}, and surjective if ran A = Cm . Note that
a square matrix A is injective (or surjective) iff it is both injective and surjective, i.e., iff it is
bijective. Bijective matrices are also called invertible matrices, because they are characterized
by the existence of a unique square matrix B (the inverse of A, denoted by A−1 ) such that
AB = BA = I.

2 Trace and determinant


The trace and determinants are functions taking square matrices and returning scalars. The
trace of A ∈ Mn (C) is the sum of its diagonal elements, i.e.,
n
X
trA := ai,i where A = [ai,j ]ni,j=1 .
i=1

Notice that the trace is linear (i.e., tr(λA + µB) = λtr(A) + µtr(B)) and that

tr(AB) = tr(BA) whenever A ∈ Mm×n (C) and B ∈ Mn×m (C).

As for the determinant, it can be defined in several equivalent ways:

1. As a function of the columns of a marix, it is the only function f : Cn ×· · ·×Cn → C that is


linear with respect to each columns (f (. . . , λx + µy, . . .) = λf (. . . , x, . . .) + µf (. . . , y, . . .)),
alternating (f (. . . , x, . . . , y, . . .) = −f (. . . , y, . . . , x, . . .)), and unit-normalized (f (I) = 1).
This can be used to derive the identity

det(AB) = det(A) det(B) for all A, B ∈ Mn (C).

1
X
2. det A = sgn(σ)a1,σ(1) · · · an,σ(n) ,
σ∈Sn
where Sn is the set of n! permutations of [1 : n] and sgn(σ) = (−1)s , s = number of
pairwise interchanges composing σ (hence the computation rules for 2 × 2 and 3 × 3
determinants). This can be used to prove that

det A> = det A for all A ∈ Mn (C).

3. Laplace expansion with respect to a row or a column, e.g. with respect to the ith row
n
X
det A = (−1)i+j ai,j det Ai,j ,
j=1

where Ai,j is the submatrix of A obtained by deleting the ith row and the jth column. The
matrix B ∈ Mn (C) with entries bi,j := (−1)i+j det Ai,j is called the comatrix of A — note
that B > is also called the adjoint of A (classical adjoint, not to be confused with hermitian
adjoint). Laplace expansion can be used to prove that AB > = (det A)I. In turn, it is
deduced that A ∈ Mn (C) is invertible iff det A 6= 0, in which case A−1 = (1/ det(A))B > .

3 Eigenvalues and eigenvectors


Given a square matrix A ∈ Mn (C), if there exist λ ∈ C and x ∈ Cn , x 6= 0, such that

Ax = λx,

then λ is called an eigenvalue of A and x is called an eigenvector corresponding to the eigen-


value λ. The set of all eigenvectors corresponding to an eigenvalue λ is called the eigenspace
corresponding to the eigenvalue λ — it is indeed a linear space. Note that λ is an eigenvalue
of A iff det(A − λI) = 0, i.e., iff λ is a zero of the characteristic polynomial of A defined by

pA (x) := det(A − xI).

Observe that pA is a polynomial of the form

pA (x) = (−1)n xn + (−1)n−1 tr(A) xn−1 + · · · + det(A).

Since this polynomial can also be written in factorized form as (λ1 − x) · · · (λn − x), where
{λ1 , . . . , λn } is the set of eigenvalues of A (complex and possibly repeated), we have

tr(A) = λ1 + · · · + λn , det(A) = λ1 · · · λn .

The existence of n linearly independent eigenvectors v1 , . . . , vn ∈ Cn corresponding to eigen-


values λ1 , . . . , λn of A ∈ Mn (which occurs in particular if A has n distinct eigenvalues) is

2
equivalent to the existence of a invertible matrix V ∈ Mn and of a diagonal matrix D ∈ Mn
such that
A = V DV −1 .
The columns of V are the vi0 s and the diagonal entries of D are the λi ’s. In this case, we say
that the matrix A is diagonalizable. More generally, two matrices A and B are called similar
if there exists an invertible matrix V such that A = V BV −1 . Note that two similar matrices
have the same characteristic polynomial, hence the same eigenvalues (counting multiplici-
ties), and in particular the same trace and determinant.

4 Exercises

Ex.1: We recall that rk A∗ = rk A, where A∗ ∈ Mn×m (C) denotes the conjugate transpose
of a matrix A ∈ Mm×n . In general, is it true that nul A∗ = nul A? Establish that
ker A = ker A∗ A, deduce that nul A = nul A∗ A and that rk A = rk A∗ A = rk A∗ = rk AA∗ ,
and finally conclude that ran A = ran AA∗ .

Ex.2: Calculate tr(A∗ A) and observe that A = 0 iff tr(A∗ A) = 0.

Ex.3: For A, B ∈ Mn (C), prove that AB = I implies BA = I. Is this true if A and B are not
square?

Ex.4: Determine the eigenvalues and eigenvectors of the matrix


 
1 t ··· t
 .. 
t 1 t .
A= ..
,
.
 
t · · · t
t ··· t 1

and diagonalize it.


Pn
Ex.5: For A ∈ Mn (Z), suppose that there exists a prime number p dividing j=1 ai,j for all
i ∈ [1 : n]. Prove that p divides det(A).

Ex.6: Determine if the following statement is true or false: there exists A ∈ Mn (R) such that
A2 + 2A + 5I = 0 if and only if n is even.

You might also like