UNIT 9 DETERMINANTS

Structure
Introduction
Objectives
Defining Determinants
Properties of Determinants
Inverse of a Matrix
Product Formula
Adjoint of a Matrix
Systems of Linear Equations
The Determinant Rank
Summary
Solutions/Answers

9.1 INTRODUCTION
In Unit 8 we discussed thesuccessive elimination method for solving a system of linear
equations. In this unit we introduce you to another method, which depends on the
concept of a determinant function. Determinants were used by the German
mathematician Leibniz (1646-1716) and the Swiss mathematician Cramer (1704-1752)
to solve a system of linear equations. In 1771. the mathematician Vandermonde
(1735-1796) gave the first systematic presentation of the theory of determinants.
There are several ways of developing the theory of determinants. In Section 9.2 we
approach it inone way. In Section 9.3 you will study the propertiesof determinants and
certain other basic facts about them. We go on to give their applications in solving a
system of linear equations (Cramer's Rule) and obtaining the inverse of a matrix. We
also define the determinant of a linear transformation. We end with discussing a method
of obtaining the rank of a matrix.
Throughout this unit F will denote 9 field of characteristic zero (see Unit I), M,(F) will
denote the set.of n x n matrices over F and V, (F) will denote the space of all n x 1
matrices over F, that is,

The concept of a determinant must be understood properly because you will be using it
again and again. Do spend more time on Section 9.2, if necessary. We also advise you to
revise Block 2 before starting this unit.

Objectives
After completing this unit, you should be able to
evaluate the determinant of a square matrix, using various properties of
determinants;
obtain the adjoint of a square matrix;
compute the inverse of an invertible matrix,;using its adjoint;
apply Cramer's Rule to solve a system of linear equations;
evaluate the determinant of a linear transformation;
evaluate the rank of a matrix by using the concept of the determinant rank.

9.2 DEFINING DETERMINANTS

There are many ways of introducing and defining the determinant function from M,(F)
to F. In this section we give one of them, the classical approach. This was given by the
French mathematician Laplace (1749-1827), and is still very much in use.
Elgenvaluea and Elgenveetors We will define the determinant function det: M,(F) + F by induction on n. That is, we
will define it for n = 1,2,3,and then define it for any n, assuming the definition for
n - 1.
When n = 1, for any A € M,(F) we have A = [a], for some a € F. In this case we define
det (A) = det ([a]) = a.
For example, det ( [ 5 ] ) = 5 and det ([-51) = -5.

When n = 2, for any A = [:,:,] € M,(F), we define

def (A) = %,a22 - %,a,,.

When n = 3, for any A = 1: a12

a32 a33
€ M3(F), we define

det(A) using the definition for the case n = 2 as follows:

det (A) = det ([::: :it]) ([:

-al2det :::I) ([: :: 1.
+

That is, det (A) = (-1)"' a,, (det of the matrix left after deleting the row and column1
containing a,,) + ,,
a (det of the matrix left after deleting
the row and column containing a,,) + (-l)'+'a,, (det of the matrix
left after deleting the row and column containing a,, ).

Note that the power of (-1) that is attached to a,, is 1 + j Tor

j = 1,2,3.
We denote det (A)
by IAl also. For
example, the dctcnninant of

[: :]isdenotedby
SO,det (A) = a,, (a,, a,,- a,, a3,)-aI2 (a2, a,,) + (a21 a,,).

In fact, we could have calculated /A1 from the second row also as follows:

Similarly, expanding by the third row, we get

All 3 ways of obtaining IA 1 lead to the same value.

Consider the following example.

Example 1 :Let

Calculate (A1
1 Solution: We want to obtain

Let A,, denote the matrix obtained by deleting the ith row and jth column of A.
Let us expand by the first row. Observe that

Thus,
lA(=(-l)"'xl xIAII(+(- 1 ) " 2 ~ 2 +~( ~
- 1~) ' ~~ 'd~ 6 ~ ( ~ 1 ~ = 5 - 6 - 7 8
= - 79.
E El) Now obtain IAl of Example 1, by expanding by the second row, and the third row.
Does the value of IA( depend upon the row used for calculating it?

Now, let us see how this definition is extended to define det(A) for any n X n matrix A,
nfl. [a" ~ I Z . . . a h 1
a21 822 . .. 82n

When A = E M,(F), we define det(A) by expanding from

- ail an2 ... ann

the ith row as follows:
+
det (A) = (- l)i"a,r det (Air ) (- 1)jt2ai 2 det (A i 2 ) + +
... (- l)'*"aindet(Ain') ,where Aii
n the (n-1) x (n-1) matnx obtained from A by deleting the ith row and the jth column,
and i is a fixed integer with 1 d i =S n.

We, thus, see that det (A) = (- l)'"ai, det (Ai,),

i=l
Eigenvalues and Eigenvectors defines the determinant of an n x n matrix A in terms of 'the determinants of the
, .
(n-1) x (n-1) matrices A,,, j = 1 , 2 ,...... , n.

Note: While calculating ( A ( ,we prefer to expand along g row that has the maxidurn
number of zeros. This cut\ d w n the number of terms to be calculated.

-:I.
The following example will help you to get used to calculating determinants.
Example 2: Let

A = [;
-3 -2 0 2
~a~culatel~,.

2 1 -3
Solution

The first three rows have one zero each. Let us expand along the third row. Observe that
a,, = 0.So we don't need to calculate (A32(. Now,

We will obtain lA,il, lA331, and JA341by expanding along the second, third and second
row, respectively.

(expansion along the second row)

= (-1) .6 + 0 + (-1) (-1).6
=-6+6=0.

+ (-1)" 1 1 I
-3 -2
(expansion along the third row)
 + (-I)'+' .o. 1 -3 -2
I (expansion along the second row)

Thus, the required determinant is given by

E E2) Calculate 1 A t ) , where A is the matrix in

'

a) Example 1,
b) Example 2.

At this point we mention that there are two other methods of obtaining
determinants -via permutations and via multilinear forms. We will not be doing these
methods here. For purposes of actual calculation of determinants the method that we
have given is normally used. The other methods are used to prove various properties of
determinants.
So far we have looked at determinants algebraically only. But there is a geometrical
interpretation of determinants also, which we now give.
Determinant as area and volume: Let u = (a,, a,) and v = (b,, b& be two vectors in R2.
Then, the magnitude of the area of the parallelogram spanned by u and v (see Fig. 1) is
the absolute value of det (u, v) = Fig. 1: 'Ihe shaded area is
det (u, v)
In fact, what we have just said is true for any n > 0.Thus, if u,, u2,....u, are n vectors in 9
 n g e ~ ~ u l n g e ~Rn,
~ then
~ the absolute value of det (u,, 9....,u,) is the magnitude of the volume of the

-
n-dimensional box spanned by u,,u, ,.... u,.
Try this exercise now.
IM (C,, C ,,....., C,) dcwta..-,
act(A)* A = (C~*
C,) is the matrix whose
E E3) What is the magnitude of the volume of the box in R3spanned by i, j and k?
columns are C,, q.. ..,C,.

Let us, now study some properties of the determinant function.

9.3 PROPERTIES OF DETERMINANTS

In this section we will state some properties of determinants, mostly without proof. We
will take examples and check that these properties hold for them.
..
Now, for any A E M,,(F) we shall denote its columns by~C,.C,, ...,C, Then we have the
following 7 properties, PI-W. ..
PI: If tiis an n x 1vector over F, then
det (C, ,....,C,,, Ci + t i , Ci+,,...., c,)
= det (C,. ...,C, ,Ci, Ci+l,.....,C,) + det(C, ,..,Ci-, Ci, Ci+,,..., C,).
Pt:If Ci = C, for any i # j, then det (C,,C, ,...,C,) = 0.
P3: If Ci and C, are interchanged (i # j) to form a new matrix B, then
det B = - det (C,, q,.. .,C,).
P4: For a €F.
det (C, ..., C,, a C i , Ci+,,...,C,,) = a det (C,, G ,...,C,).
Thus, det (=C,,aG ,..., aC,) = andet (C, ,...,C,).
Now, using PI, P2 and P4. we find that for i # j and a E F,
det (C, ,...,Ci + a C j,...,C, ,...,C,) = det (C1,...,Ci,---,Cj,"',Cn) + (Ci,...,
ci,...cj,... C,,).
= det (C,,G,. ..,C,).
Thus, we have
PS: For anya E F and i # j, det (C, ,...,Ci + aCj,Ci+,,...,C,) = det (C,,G ,...,CJ.
Another property that we give is
P6: det(A) = det(At)VA E M,,(F). (In E2 you saw that this property was true for
Examples 1and 2. Its proof uses the permutation approach to determinants.)
Using P6, and the fact that det (A) can be obtained by expandingalong any row, we get
W:For A E we can obtain det(A) by expanding along any column. That is, for
a fixed k,
I A 1 = (- 1)'"alr +(- 1)~"a2k IA2d + +
... (- 1)""a.~ 1A.d.
An important remark now.
Rcmuk: Using P6, we can immediately say that PI-P5 are valid when columns are
repla,h by rows.
Using the n0t'Ition of Unit 8, P3 says that
-
det (%(A)) = det(A) = det (C,,(A)) .
,P4says that
det (R,(a ) (A))'= a' det (A) = det (C, ( a) (A)), Y a E F, and P5 says that
det (R,,( a ) (A)) = det (A) = det (C,,)( a) (A), Y a E F.
I
1 We will now illustrate how useful the properties P I - P7 are. Determinants

t Example 3: Obtain det (A), where A is

Solution: a) Since the first and third rows of A (R, and R,) coincide, I A I = 0,by P2 and
P6.

1 2 -1 -3
= 2 4 5 0 , by adding R, toR,.

= 0, since R, = R, .
Try the following exercise now
E E4) Calculate 1 3 0 2 3 5
2 1 2 and1 0 1.
1 3 0 4 6 10

Now we give some examples of determinants that you may come across often.
Exampk 4: Let

, wherea,bER.
b b a b
b ' b b a

Calculate 1 A (

a b b b
I A J= b
b
a
b
b
a
b
b
b b b a

a+3b a+3b a+3b a+3b (by adding the second, third and fourth rows
b' a b b to the first row, and applying P5)
- b b a b
b b b a
 a+3b 0 0 0 (by subtracting the first column from
- b a-b 0 Id every other column, and using PS)
b 0 a-b 0
b 0 0 g- b

= (a+3b)a-b 0 0 (expanding along the first row)

0 a-b 0
0 0 a-b

= (a+3b) (a-b)?

In Example 4 we have used an important, and easily proved fact, namely,

det (dhg (a,, +..,
aJ) = a, ~z a,,, a, E F V i....
This is true because,

Example 5: Show that

(This is known as the Vandermonde's determinant of order 4.)

Solotlon:The given determinant

1 0 0 0 (by subtracting
- XI X2 - XI X3 - XI X4 - XI the first column from
XI
2
x22- xI2 x: - XI 2 xd2 - XI
2 every other column)
x13 -
xZ3 x13 ~3~ - XI 3
x1
3
- x13

x2 - XI X3 --.XI X4 - XI
(x2 - XI) + XI)
(x2 - XI) + XI) (x3 (x3 (x4- XI) (x4 + XI)
X2 - XI) (x? + + h ~ l ) - XI) (h2
x12 + xI2 +
(x3 ~ 3 ~ 1(XI) - XI)(%^ + x12 + ~ 4 ~ 1 )

(by expanding along the first row and factorising the entries)

=(%-XI) (5-x1) (x4-x1) 1 1 1

%+XI 5+ x1 x4 + X l
x;+ x: + x; + x: + x3x1 x42 + x12 + x4x1
(by @king out (%-x1), (5-x,), and (x4-x,) fmm Columns 1.2 and 3 respectively). '

(by subtracting the k t column from thejsecond and third columns)

 "
, (
(expanding by the first row and factorising the entries)

' Try the following exercise now.

E E5)What are a 0 0 a d e
a b O a n d O b f ?
P T C o o c

The answer of E4 is part of a general phenomenon, namely, the determinantof an upper

or lower triangular matrix is the product of its diajgonal elements.
The proof of this is immediate because,
811 * . *
0 a2? ... * a22 * . . . *
0 0 0 . . *
. . . =a11 . 833.
. . (expanding along C,)
. . . . . .
0 0 ... an. 0 0 . . .ann

_ . . ,-_ a11a 2 2 . . a,,

- each time expanding along the first column.

In the Calculus course you must have come across dfldt =fP(t), where f is a function oft.
The next exercise involves this.
E E6) Let us define the function 0(t) by

Show that /j'(t) =

,
-.
'. And now, let us study a method for obtaining the inverse of an invertible matrix.

A .

' 1' 9.4 INVERSE OF A MATRIX

$1 In this section we first obtain the determinant of the product of two matrices and then
define an adjoint of a matrix. Finally,we see the conditions under which a matrix is
invertible, and, when it is invertible, we give its inverse in terms of its adioint.
~igenvduesand Ei#envcetors 9.4.1 Product Formula
In Unit 7 you studied matrix multiplication. Let us see what happens to the determinant
of a product of matrices.
Themern 1:Let A and B be n x n matrices over F. Then det (AB) = det(A) det (B).
We will not do the proof here since it is slightly complicated. But let us verify Theorem
1for some cases.

Example 6: Calculate ( AI,( B 1 and 1AB ( when

Solution: We want to verify Theorem 1for our pair of matrices. Now, on expanding by
the third row, we get ( A( = 1.
Also, IB( = 30, which can be immediately seen since B is a triangular matrix.

You can verify Theorem 1for the following situation.

E E7) Show that (ABI = IAl ( B ( , where

Theorem 1 can be extended to a product of m n x n matrices,

A,,A ,,..., A,. That is,
,....
det (A, A A,,,) = det (A,) det (A,) .....
det (A,,,)
Now let us look at an example in which Theorem 1simplifies calculations.
Example 7: For a, b,c E R, calculate
 Sdution: The solution is very simple. The given matrix is equal to

we get the required determinant to be

a b c 2
(by Theorem 1)
b c a

because a b c
c a b
b c a
= a I t 1 - 1 b
a I + c ( L PC)
= a(a2-bc) - b(ac-b2)+ c (c2-ab)
= a3+ b3+ c3-3abc.

Now, you know that AB f BA, in general. But, det (AB) = det(BA), since both are
equal to the scalar det(A) det(B).

,4zl
On the other hand, det (A+B) f det(A) + det(B), in general. The following exercise is
i.;: an example.

E E8) Let A = [:(:] ,B =

0 -1
01 . Show that det(A+B) f det(A) + det (B).

What we have just said is that det. is not a linear function.

::
$,! We now give an immediate corollary to Theorem 1.
Corollary 1: If AEMJF) is invertible,then d e t ( ~ - ' ) ~lldet
= (A).
Roof: Let BE MJF) such that AB = I. Then det (4B)= det(A) det(B) = det(1) = 1
Thus, det(A) f 0 and det (B) = lldet(A). In particular,
det (A-') = l/det(A).

Another corollary to Theorem 1 is

Corollary 2: Similar matrices have the same determinant.
Proof: If B is similar to A, then B = P-' AP for some invertible matrix P. Thus, by
Theorem 1, det(B) = det(~-'AP) A matrix B is similar to a matrix
= det (P-') det(A) det (P) = l/det(P). det(P). det(A), by Cor.1. A if there exists a non-singular
=det(A). matrix P such that P-'AP = B.

We use this corollary to introduce you to the determinant of a linear transformation. At

each stage you have seen the very close relationship between linear transformations and
matrices. Here too, you will see this closeness.
Defdtion: Let T:V+ V be a linear transformation on a finite-dimensionalnon-zero
vector space V. Let A = [TI, be 'he matrix.of T with respect to a given basis I3 of V.
Then we define the determinant of T by det(T) = det(A).
I
I ~$cavdu;bl a d ~gckvcetors This definition is independent of the basis of V that is chosen because, if we choose
another basis B' of V we obtain the matrix A: = me',,which is similar to A (see Unit 7,
I Cor. to Theorem 10). Thus, det (A') = det (A).
We have the following example and exercises.
Example 8: Find det(T) where we define T:R3 + R3 by
T(xl,$,x3) = (3~,+~~,-2x,+x,,-x,+2 x2+4x3)
Solution: Let B = {(1,0,0), (0,1,0), (0,0,1)) be the standard ordered basis of R3. NOW,

So, by definition,
3 0 1
det(T)= det(A) = -2 1 0
-1 2 4

E E9) Find the determinant of the zero operator and the identitfoperator from R3+ R3.

E E10) Consider the differential operator

D: P,+ P, : D (%+alx+%x2) = a, +2%~.
What is det(D)?

Let us now see what the adjoint of a square matrix is, and how it will help us in obtaining
I
the inverse of an invertible matrix. f
,rj

1 I

9.4.2 Adjoint of a Matri* I

i.
In Section 9.2 we used the notation Ailfor the matrix obtained from a square matrix A by
deleting its ith row and jth column! elated to this we define the (i,j)th cofactor of A (or I

the cofactor of au) to be (-ly+l IA,]I.It is denoted by Cij. That is C , = (- 1)"' IA,]I.
1'
l
Consider the following example. , <
 Exampk 9: Obtain the cofactors C,, and Cu of the matrix A =

, Solution: C12= (- 1)'" IAI2) = - 1 :1=-16

In the following result we give a relationship between the elements of a matrix and their
cofactors.

Theorem 2: Let A = [aij],,,. Then,

a) a,, Cil + ai2Ci2+.... + ainCin= det(A) = aliCli+%iC2i+ .. +a,,C,.

b) ail Cjl + ai2Ci2+.... + ainCjn= 0 = aliClj+qiC2j+ .. +aniCn, if i # j.
We will not be proving this theorem here. We only mention that (a) follows immediately
I+
from the definition of det (A), since det (A) = (-1)"' a,, (A,, ... + (-l)'+" a, IA~, 1.
E Ell) Verify (b) of Theorem 2 for the matrix in Example 9 and i = l , j=2 or 3.

Now, we can define the adjoint of a matrix.

Defdtion: Let A = [a,] be any n x n matrix. Then the adjoint of A is the n x n matrix,
;. denoted by Adj(A), and defined by
r i i

"v',
where C, denotes the (ij)th cofactor of A .
i' '
Thus, Adj(A) is the n X n matrix which is the transpose of the matrix of corresponding
cofactors of A.
Let us look at an example.
cos0 0 -sin 0
Example 10: Obtain the adjoint of the matrix k =

1b I
1
Solution: Cll = (-1)'"
COY0 =

c12 = (-1)1+2 0
sine COS~
0 1 =0
 +
CZ1= 0, q2= COS% sin% = 1, q3= 0.
C31 = sin 0, C,, = 0, C,, = cos 0.

cos0
.,Adj(A)=[O
sin 0
0 - sin 8 '
; o 6] =
cos
cos 0
[ o- sin 8 ;
0
.I
sin0
cos 0

Now you can try the following exercise.

E E12) Find Adj(A), where A =

[I :-:]
0 0 6 .

In Unit 7 you came across one method of finding out if a matrix is invertible. The
following theorem uses the adjoint to give another way of finding out if a matrix A is
invertible. It also gives us A-l, if A is invertible.
Theorem 3: Let A be an n x n matrix over F, Then
A. (Adj(A)) = (Adj(A)). A = det(A) I.

Proof: ~ i c a lmatrix
l multiplication from Unit 7. Now
- ...
dl I 812 ... C l l C21
82 I 822 ... CI? C22 . . C,,:
A (Adj (A)) =

an I an2 ... arm Cln ...CI, C,,

Ry Theorem 2 we know that ailCi,+ a&,, + ... + a, Cin= det (A), and
ailCjl + + ... + a,Cjn = 0 if i f j. Therefore,
1;
det (A) 0 ... 0
det (A) . ..

...
...
... det'(A)

= det (A)
 Similarly, (Adj(A)) .A = det (A)I.
An immediate corollary shows us how to calculate the inverse of a matrix, if it exists.
Corollary: Let A be an n x n matrix over F. Then A is invertible if and only if
det (A) # 0. If det(A)# 0, then
A-' = (l/det(A)) Adj(A).
Proof: If A is invertible, then A-' exists and A-' A = I. So, by Theorem 1,
d e t ( ~ - ' )det(A) = det(1) = 1. .: ,det(A) # 0.
7 $1I Conversely, if det(A) # 0, then Theorem 3 says that
,jbt

iy 1
' I .: A-' = -Adj (A).
IAl

It We will use the result in the following example.

Example 11: Let

I
cos 8 0 - sin 8
A=[ 0 1 Find A-' .
sin 8 0 cos
o8

Solution:

det(A) = (- .l. cos 0 -sin 0 (by expansion along the

sin 0 cos 0 second row)
= cos20+ sin20= 1
Also, from Example 10 we know that

Therefore, A-'= (lldet (A)) Adj(A) = Adj(A).

You should also verify that Adj(A) is A-' by calculating A. Adj(A) and Adj(A). A.
You can use Theorem 3 for solving the following exercises.
E E13) Can you find A-' for the matrix in E 12?

E E14) Find the adjoint and inverse of the matrix A in E7.

 Elgenvnluea and Elsenvectors

E E15) If A-' exists, does [Adj(A)]-' exist? If so, what is [Adj(A)]-I?

Now we go to the next section, in which we apply our k.nowledge of determinants to

obtain solutions of systems of linear equations.

9.5 SYSTEMS OF LINEAR EQUATIONS

Consider the system of n linear equations in n unknowns, given by

anlX1 + an2X2 + .. . + BnnXn = bn.

which is the same as

In Section 8.4 we discussed the Gaussian elimination method for obtaining a solution of
this system. In this section we give a rule due to the mathematician Cramer, for solving
a system of linear equations when the number of equations equals the number of
variables.

Theorem 4: Let the matrix equation of a system of linear equations be

Theorem 4 is called Cramer's

Rule.

Let the columns of A be C,, C,,. ..,Cn. If det(A) f 0, the given system has a unique
solution, namely,
x1 = DI/D,....,xn = Dn/D,where
D, = det (C,,... C,-,,B,C,+,,..,Cn)
= determinant of the matrix obtained from A by replacing the ith column by B, and
D = det (A).
 h f i Since IAl f 0, the corollary to Theorem 3 says that A-' exists.
NOWAX = B A-'AX = A-'B
-1X = ( I D ) Adj(A) B

C!I c21
CI? C2? ...
...
* X = (1/D)
,.I' . .
"
bn
din d2n .. . cnn

Thus,

Now, Di = det (C, ,..., C,,, B, Ci+,,..., C,). Expanding along the ith column, we get
+
Di = Clibl CZib,+ ... + Cnibn.

Thus, - -
-
x1 Dl
Xz "2

=1/D .

-Xn - Dn-,

which gives us Cramer's Rule, namely,

X1 = D I D , X2 = D2/D,....,X, = D,/D.

The following example and exercise may help you to practise using Cramer's Rule.
Example 12: Solve the following system using Cramer's Rule:

'
i
A= 2
2 1 -6
, = 11, [i]
Solution: The given system is equivalent to AX = B, where
2 3 -1
= ,Therefore, applying the rule, we get

After calculating, we get

x = -23, y = 14, z = - 6.
 Eigenvduu ~d Elgenvectom Substitute these values in the given equations to check that we haven't made a mistake
in our calculations.
E E16) Solve, by Cramer's Rule, the following system of equations.

Now let us see what happens if B = 0. Remember, in Unit 8 you saw that AX = 0 has
n - r linearly independent splutions, where r = rank A. The following theorem tells us
this condition in terms of det(A).
Theorem 5: The homogeneous system AX = 0 has a non-trivial solution if and only if
det(A) = 0.
Proof: First assume that AX = 0 has a non-trivial solution. Suppose, if possible, that
det(A) f 0. Then Cramer's Rule says that AX = 0 has only the trivial solution X = 0
X is non-trivial if x # 0. (because each Di=O in Theorem 4). This is a contradiction to our assumption.
Therefore, det (A) = 0.
Conversely, if det (A) = 0,then A is not invertible. :., the linear mapping
A :Vn(F) -+ Vn(F): A(X) = AX is not invertible. .'., this mapping is not one-one.
Therefore, ~ eA r# 0, that is AX = 0 for some non-zero X E Vn(F).Thus, AX = 0 has
a non-trivial solution.
You can use Theorem 5 to solve the following exercise.

E E17) Doesthesystem 2x + 3y + z =0
x-y - 2 =o
4x + 6y + 22 =0
have a non-zero solution?
 And now we introduce you to the determinant rank of a matrix, which leads us to
, another method of obtaining the rank of a matrix.
I

9.6 THE DETERMINANT RANK

In Units 5 and 8 you were introduced to the rank of a linear transformation and the rank
of a matrix, respectively. Then we related the two ranks. In this section we will discuss
the determinant rank and show that it is the rank of the concerned matrix. First we give
a necessary and sufficient condition for n vectors in V,(F) to be linearly dependent.
Theorem6: Let XI, &,.....,X,E V,(F). Then XI, %,. ..,Xn are linearly dependent over
the field F if and onlyif det (XI, X,,...., X,)= 0.

I
Proofi Let U = (XI, Xi,..,X,,) be then x n matrix whose column vectors are XI, &....,
X,,. Then XI, &,.....,X,, are linearly dependent over F if and only if there exist scalars
al, a2,....,an E F, not all zero, such that a, X, + a,% + .... + a,X, = 0.

'i Thus, X,, &,...,X, are linearly dependent over F if and only if UX = 0 for some non-

But this happens if and only if det (U) - 0, by Theorem 5. Thus, Theorem 6 is proved.
..
Theorem 6 is equivalent t o the statement X, ,X2,. ,Xn E Vn(F) are linearly independent if
and only if det (Xl,X2,...&)# 0.

r]
You can use Theorern, 6 for solving the following exercise.

E E18) Check if the vectors

[ ] [- [ i ]
, 7. are linearly independent

over R.

61
Now, consider the matrix A = 0 4 5

Sincetwo rows of A are equal we know that IAI = 0. But consider its 2 x 2 submatrix
A submatrix of A IS a matrlx
that can be obtalned from A by
deleting some rows and
columns. 2.3
~ . k c s u d ~ v a t a s
A13= [ ] Its determinant is - 4 + 0. In this case we say that the determinant
rank of A is 2.
In general, we have the following definition.
Definition: Let A be an m x n matrix. If A # 0, then the determinant rank of A is the
largest positive integer r such that
i) there exists an r x r submatrix of A whose determinant is non-zero, and
ii) for s > r, the determinant of any s x s submatrix of A is 0.
Note: The determinant rank r of any m x n matrix is defined, not only of a square
matrix. Also r lmin (m, n).
Consider the following example.

Example 13: Obtain the determinant rank of A = 2 5 .

[1
Solution: Since A is a 3 x 2 matrix, the largest possible value of its determinant rank can
be 2. Also, the submatrix 1 4 of A has determinant (-3) # 0.
[2 51
.: ,the determinant rank of A is 2.
Try the following exercise now.

E E19) Calculate the determinant rank of A, where A =

And now we come to thetreason for introducing the determinant rank-it gives us
another method for obtaining the rank of a matrix.
Theorem 7: The determinant rank of an m x n matrix A is equal to the rank of A.
Proof: Let the determinant rank of A be r. Then there exists an r x r submatrix of A
whose determinant is non-zero. By Theorem 6, its column vectors are linearly
independent. It follows by the definition of linear independence, that these column
vectors, when extended to the column vectors of A, remain linearly independent. Thus,
A has at least r linearly independent column vectors. Therefore, by definition of the
rank of a matrix,
r s rank (A) = p (A) ...... . (1)
Also, by definition of p (A), we know that the number of linearly independent rows that
A has is p (A). These rows form a p (A) x n matrix B of rank p (A). Thus, B will have
p(A) linearly independent columns. Retaining these linearly independent columns of B
we get a p (A) x F (A) submatrix C of B. So, C is a submatrix of A whose determinant
will be non-zero, by Theorem 6, since its columns are linearly independent. Thus, by the
definition of the determinant rank of A,we get
F

............... (2)
/
L
P (A)< r
(1) and (2) give us us p(A) = r.
We will use Theorem 7 in the following example.
! Example 14: Find the rank of
r

A= [-1
2
3
3
1
2
41 .

,
I
Solution: det (A) = 0.But det
([:. I]: =-7.0.

Thus, by Theorem 7, p (A)=2.

Remark: This example shows how Theorem 7 can simplify the calculation of the rank
1 of a matrix in some cases. We don't have to reduce a matrix to echelon form each time.
But, in (a) of the following exercise, we see a situation where using this method seems
to be as tedious as the row-reduction method.
C
E E20) Use Theorem 7 to find the rank of A, where A =
E20 (a) shows how much time can be taken by using this method. On the other hand,
E20 (b) shows how little time it takes to obtain p (A), using the determinant rank. Thus,
the method to be used for obtaining p (A) varies from case to case.
We end this unit by briefly mentioning what we have cover in it.

9.7. SUMMARY
In this unit we have covered the following points.
1) The definition of the deierminant of a square matrix.
2) The properties PI-P7, of determinants.
3) The statement and use of the fact that det(AB) = det(A) det (B).
4) The definition of the determinant of a linear transformation from U to V, where
dim U = dim V.
5) The definition of the adjoint of a square matrix.
(6) The use of adjoints to obtain the inverse of an invertible matrix.
7) The proof and use of Cramer's Rule for solving a system of linear equations.
8) The proof of the fact that the homogeneous system of linear equations AX = O.has
, a non-zero solution if and only if det(A) = 0.
9) The definition of the determinant rank, and the proof of the fact that rank of A =
determinant rank of A.

9.8 SOLUTIONSIANSWERS
E l ) On expanding by the 2nd row we get

JAl = -5 (A211+ 41A221- IA23l.

Expanding by the 3rd row, we get

= 7(-22) -3(-29) + 2(-6) = -79.

Thus, JAl = -79, irrespective of the row that we use to obtain it.

2 4 3 . .'.,on expanding by the first row, we get

I 6 1 21

b)At= -2
l o
1 0
0 1 - 3 I Since the 3rd row has the maximum
number of zeros, we expand along it. Then

E3) The magnitude of therequired volume is the modulus of

1 0 1
0 1 01 =l.
0 0 1
We draw the box in Fig. 2.

E4) The first determinant is zero, using the row .equivalent of P2. The second
determinant is zero, using the row equivalent of P5, since R, = 2R,.

Fig. 2

d df
since -(fg) = -g
dt dt
+f-
dg
dt

E7) Note that B is obtained from A by interchanging,C, and C,.

Also JAB1
1: I:
= -14 10 -6
-2:

-6 3 -2
= -14 10 -6 adding 2% to R,.
0 -1 6

21 + I$ 1; 1 ,expanding along R,.

=8-108=-100= IAl (BI.

E9) Let B be the standard basis of R3.The zero operator is

0 : ~ ' - R 3 : 0 ( x ) = 0 V x€R3.Now,[0],=0.
:. det (0) = 0.
I : R' -- R3: I (x) = xV x E R3,is the identity operator or R3. NOW,[IIB= I,.
... det(1) = det (I,) = 1.
4 e b c n - t - E10) The,standard basis for P, is {l,x,x2).
Now D(1) = 0, D(x)= 1, D(x2) = 2x.

+ aI3CD= 2(-1)'+
E l l ) a,~C, + al2CZ2

Similarly, check that a,,C,, + %,C,, + a,,C, = 0,

E12) C,, = 0, CI2= 0, CI3= 0, C2, = -15, C2, = 10, Cu = 0, C,, = 18, C3, = -12, ,
c,, = 0.
. Ad,(.) = [;
0 -15 18
1; I:].

E13) Since IAl = 0, A-I does not exist.

E14) From E7 we know that (A1= 10.
NOW,C,, = 4, C,, = 6, C,, = -6,
C,, = 3, C2, = 8, Cu ='3,
C3,=2,c3,=2,C,,=2.

.: A-' = -1
IAl
Adj (A) ==
1
--.

.lo [-% i i].

Verify that the matrix we have obtained is right, by multiplying it by A.
E15) Since A. Adj (A) = IA( I = Adj (A). A, and (A1+ 0, we find that
1
[ ~ d j ( ~ ) ]exists,
-' and is - A.
I
E16) This is of the form AX = B, where

A = [21 3
1 2 4
-3
0 -11, X = [I]
,B=[i]

D , = 2
I: : :I
3 3 =1
 -

Determinants
E17) The given system is equivalent to AX = 0, where

A = 1 -1

Now, the third row of A is twice the first row of A.

.'.,by P2 and P4 of Section 9.3, 1 A1 = 0.
.'.,by Theorem 5, the given system has a non-zero solution.
E18) 1 0 2 --
0 -1 3 = -3 + 2 = -1 f 0. .'.the given vectors are linearly independent.
1 1 Cl

E19) a) Since IA( # 0, the determinant rank of A is 3.

b) As in Example 13, the determinant rank of A is 2.
E20) a) The determinant rank of A S 3.
Now the determinant of the 3 X 3 submatrix

Also, the determinant of the 3 x 3 submatrix Il

3 2 5
is zero.

In fact, you can check that the determinant of any of the 3 x 3 submatrices is
zero. Now let us look at the 2 x 2 submatrices of A. Since 3 1 = 5 # 0,
I1 21
we find that p (A) = 2.
b) The determinant rank of A G 2.