Linear Algebra: Assignment #2

Professor N. Padmanabhan
Due on September 16, 2019 at 12:00 PM

Prayag Savsani
AU1841035

1
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 1
If the columns of a matrix are dependent so are the rows. T/F, with reason or counter example

Solution: False
 
Counterexample: A = 1 1 1 . Here, the column vectors are dependent but the only single row vector is
independent.

Problem 2
The columns of a matrix form a basis for the column space of the matrix. T/F, with reason or counter
example

Solution: False
 
1 2
Counterexample: Any matrix with dependent columns eg. A = .
2 4
 
1
Here the second column is dependent on the first column and hence, the column space is C(A) = .
2
The basis set of the column space does not contain all the columns of the matrix.

Problem 3
The column space C(A)) of a matrix A is the same as the column space C(rref (A)) of the row reduced
echelon form of A. T/F with reason or counter example.

Solution: False
Counterexample:

−1
   
1 2 3 1 0
A= 3
 4 5 , rref (A) = 0 1
  2
4 6 8 0 0 0

Here, as rref (A) has two pivot columns which are column 1 and column 2, those respective columns form the
    
 1 0 
basis for its column space. So C(rref (A)) = 0 , 1 which spans only R2 . And the same columns
 
0 0
   
 1 2 
1 and 2 form the basis for column space of A. So C(A) = 3 , 4 which in fact spans R3 . Hence, as
 
4 6
the basis vectors change, the column spaces also change.

2
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 4
dim(C(A)) = dim(N (A)). T/F with reason or counter example.

Solution: False
For a matrix Amxn , we know that
rank + nullity = n (1)
Given that rank = nullity, Eq. (1) becomes,

rank + rank = n
∴ n = 2 ∗ rank

Hence, no. of columns in the matrix would be a multiple of 2. So, a matrix with odd number of columns
serves as our counterexample. One such matrix is,
 
1 2 3
A = 5 6 7
9 10 13

Problem 5
Two vectors in R3 are always linearly independent. T/F with reason or counterexample.

Solution: False
Counterexample: Consider the vectors,
   
1 2
a = 0 and b = 0
  
0 0

Clearly, a, b ∈ R3 but a and b are not linearly independent. The reason being that b is a linear combination
of a.

Problem 6
A is a square matrix. If the rows are linearly independent, then the columns are also linearly independent.
T/F with reason or counter example.

Solution: True
As for every square matrix A, dim(R(A)) = dim(C(A)).
Here, dim(R(A) =⇒ No. of linearly independent rows in the matrix and dim(C(A) =⇒ No. of linearly
independent columns in the matrix. Them being equal proves the given implication.

3
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 7
Ax = b has infinitely many solutions. Is it possible for Ax = b0 (different RHS) to have a unique solu-
tion? Can Ax = b00 have no solution for some b00 ? Give good reasons for your answer. No example or
counterexample.

Solution
As Ax = b has infinitely many solutions, the system has at least one free variable and the matrix A doesn’t
have a full rank i.e. rank 6= n, where n is the number of columns.

(i) No such b0 is possible as for an unique solution to exist, there should be no free variables. And no
change in RHS can convert the existing free variables in the matrix to non-free variables.

(ii) ∀ b00 ∈
/ Rm , where m is the number of rows in A, the system Ax = b00 will have no solution.

Problem 8
Am×n is a matrix with rank r and Bm×n is a matrix with rank r0 . C is a matrix obtained from A by
appending the columns of B. What can you say about the rank of C? Give good reasons.

Solution
We try to define bounds for the rank of C. Talking about the minimum bound first, consider matrix A with
rank r and its first r rows to be linearly independent. The remaining m−r rows will be dependent. Similarly,
matrix B of rank r0 with first r0 rows linearly independent and remaining dependent. In such a case, when
matrix C is formed, its rank will always be greater than or equal to both r and r0 . There will be no case
where after appending the columns, no. of linearly independent rows becomes less than the maximum of r
and r0 as if after appending some row becomes dependent, it would have to be dependent in both matrix A
and B which will lead us to a contradiction regarding the values of r and r0 .

rank of C ≥ max(r, r0 ) (1)

Now, coming to the maximum bound, matrix C is of dimensions m × 2n. Hence, the maximum value of its
rank can be 2n. Consider both matrix A and B to have full ranks, i.e. r = r0 = n.

rank of C ≤ 2n (2)
∴ rank of C ≤ n + n (3)
0
∴ rank of C ≤ r + r (Since r = r’ = n) (4)

∴ max(r, r0 ) ≤ rank of C ≤ r + r0 (From (1) and (4)) (5)

4
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 9
Show that N (A) = C(A) ⇒ A2 = 0

Solution
Given that N (A) = C(A), N (A) ⊂ Rn and C(A) ⊂ Rm , m must be equal to n. Also, by the rank-nullity
theorem, as dimensions of null space and column space are equal, the no. of columns in the matrix should
be of even order. All this implies that matrix A is a square matrix of even order.
Moreover, through the property N (A) ⊥ R(A), we can say that C(A) ⊥ R(A). This implies that for any
two given vectors in C(A) and R(A), their dot product is zero.
Hence,

A2 = A ∗ A (1)
   
a11 · · · a1n a11 ··· a1n
2  .. .. ..  ·  .. .. ..  (n is even)
A = . . .   . . .  (2)
an1 · · · ann an1 ··· ann
A2 = 0 (Since, column vectors are orthogonal to the row vectors) (3)

Hence, proved.

5
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 10
 
A
Amxn is a matrix and B = is a 2m × n matrix obtained from A. Compare the null and column spaces
A
of the two matrices.

Solution
Let the rank of matrix A be r ≤ n. Hence,
 
a11 ··· a1n  
 .. .. ..  =⇒ rref (A) =   −F
A= . . .  Ir×r F =⇒ N (A) = (1)
I
am1 ··· amn

Moreover, the basis of column space of A will contain the r linearly independent columns which in this case
will span a subset of Rm . Now, matrix B on the other hand,
 
a11 ··· a1n
 .. .. .. 

 . . . 

 am1 ··· amn 
 
B= (2)
a(m+1)1 · · · a(m+1)n 

.. .. ..
 
.
 
 . . 
a(2m)1 · · · a(2m)n
 
a11 · · · a1n
 . .. .. 
 .. . . 
 
am1 · · · amn 
 
After row operations, B can be reduced to =  (3)
 0 ··· 0 

 .. .. .. 
 
 . . . 
0 ··· 0
   
Ir×r F −F
∴ rref (B) = =⇒ N (A) = (4)
0 0 I

From Eq. (1) and (4), we can see that null spaces of both A and B are same.
As far as column space of B is concerned, as the rank of A and B stays the same, the no. of basis vectors
stays the same. But, as basis vectors of B are of dimension 2m, the column space of B will be a subset of
R2m unlike column space of A which is a subset of Rm . Hence, their column spaces are in fact different.

6
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 11
Let U and W be the following subspaces of R4 :
U = {(a, b, c, d) : b + c + d = 0}, W = {(a, b, c, d) : a + b = 0, c = 2d}.
Find the dimension and basis of (i)U, (ii)W, (iii)U ∩ W .

Solution
(i) U : As U is of dimension atmost 4, consider a linear combination of four vectors u1 , u2 , u3 , u4 ∈ U
such that xu1 + yu2 + zu3 + wu4 = 0
This can be written as AX = 0, where

a1 a2 a3 a4
 
 b1 b2 b3 b4 
A=  c1 c2 c3 c4 


d1 d2 d3 d4
Given that bi + ci + di = 0,

a1 a2 a3 a4
 
 b1 b2 b3 b4 
A= 
 c1 c2 c3 c4 
−(b1 + c1 ) −(b2 + c2 ) −(b3 + c3 ) −(b4 + c4 )
a1 a2 a3 a4
 
 b1 b2 b3 b4 
A=  c1 c2 c3 c4  (R4 → R4 + R2 + R3 )


0 0 0 0

Hence, dim(U ) = 3.
  
1 1 1 
  

 
1 1  0 
 
One basis set can be  0 ,
 ,
−1
 
 1 

 
 
−1 0 −1
(ii) W : As W is of dimension atmost 4, consider a linear combination of four vectors w1 , w2 , w3 , w4 ∈ W
such that xw1 + yw2 + zw3 + ww4 = 0
This can be written as AX = 0, where

a1 a2 a3 a4
 
 b1 b2 b3 b4 
A=  c1 c2 c3 c4 


d1 d2 d3 d4
Given that ai + bi = 0, ci = 2di ,

a1 a2 a3 a4
 
−a1 −a2 −a3 −a4 
A=  2d1 2d2

2d3 2d4 
d1 d2 d3 d4
a1 a2 a3 a4
 
0 0 0 0
A=  (R2 → R1 + R2 , R3 → R3 − 2R4 )
0 0 0 0
d1 d2 d3 d4

Problem 11 continued on next page. . . 7

Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Hence, dim(W ) = 2.
  
1 1 


 
−1 −1
 
 0 ,
One basis set can be    
 2 

 
 
0 1

(iii) U ∩ W : First, we try to find dimension of U + W using the same combination of vectors as used above.
Hence, in AX = 0, A is,

a1 + p1 a2 + p2 a3 + p3 a4 + p4
 
 b1 + q1 b2 + q2 b3 + q3 b4 + q4 
A=
 c1 + r1

c2 + r2 c3 + r3 c4 + r4 
d1 + s1 d2 + s2 d3 + s3 d4 + s4

Given that bi + ci + di = 0, pi + qi = 0, ri = 2si ,

a1 + p1 a2 + p2 a3 + p3 a4 + p4
 
 a1 + b1 a 2 + b2 a 3 + b3 a 4 + b4 
A=
c1 − 2q1 − 2r1

c2 − 2q2 − 2r2 c3 − 2q3 − 2r3 c4 − 2q4 − 2r4 
q1 + r1 + s1 q2 + r2 + s2 q3 + r3 + s3 q4 + r4 + s4

(R4 → R4 + R2 + R3 , R2 → R1 + R2 , R3 → R3 − 2R4 )
Hence, dim(U + W ) = 4.
Now, we know dim(A + B) = dim(A) + dim(B) − dim(A ∩ B). Plugging in U and W , we get

dim(U + W ) = dim(U ) + dim(W ) − dim(U ∩ W ) (1)

∴ dim(U ∩ W ) = dim(U ) + dim(W ) − dim(U + W ) (2)
∴ dim(U ∩ W ) = 3 + 2 − 4 (3)
∴ dim(U ∩ W ) = 1 (4)

Now, basis set of U ∩ W will contain one vector u = {a, b, c, d} such that b + c + d = 0, a + b = 0, c = 2d.
 
 3 

 
−3

One basis set can be 
 2 
 
 
1

8
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 12
1 −1 1 −1 1
   
0 1 −1 1 1
Consider A =  . Find the inverse of this matrix and solve Av =  .
0 0 1 −1  1
0 0 0 1 1
Guess the result of inverse of a 5 × 5 ”alternating matrix” and verify by multiplication.

Solution
To find A−1 ,

1 −1 1 −1 1 0 0 0
 
0 1 −1 1 0 1 0 0
∼ 
0 0 1 −1 0 0 1 0
0 0 0 1 0 0 0 1
(R1 → R1 + R4 , R3 → R3 + R4 , R2 → R2 − R4 )
1 −1 1 0 1 0 0 1
 
0 1 −1 0 0 1 0 −1
∼
0

0 1 0 0 0 1 1
0 0 0 1 0 0 0 1
(R2 → R2 + R3 , R1 → R1 − R3 )
1 −1 0 0 1 0 −1 0
 
0 1 0 0 0 1 1 0
∼
0

0 1 0 0 0 1 1
0 0 0 1 0 0 0 1
(R1 → R1 + R2 )
1 0 0 0 1 1 0 0
 
0 1 0 0 0 1 1 0
∼
0 0 1 0

0 0 1 1
0 0 0 1 0 0 0 1
1 1 0 0
 
0 1 1 0
∴ A−1 =
0

0 1 1
0 0 0 1

Now,

1
 
1
Av = 
1


1
1
 
1
∴ v = A−1 


1
1

Problem 12 continued on next page. . . 9

Prayag Savsani Linear Algebra : Assignment #2 AU1841035

1 1 0 0 1
  
0 1 1 0 1
∴v=
0
 
0 1 1 1
0 0 0 1 1
2
 
2
∴v=
2


By the inverse of 4 × 4 alternating matrix, we can guess the inverse of 5 × 5 alternating matrix to be,
 
1 1 0 0 0
0 1 1 0 0
 
−1
A = 0 0 1 1 0
 
 
0 0 0 1 1
0 0 0 0 1

Solving for v,
 
1
1
 
v = A−1 1
 
 
1
1
  
1 1 0 0 0 1
0 1 1 0 0 1
  
∴ v = 0 0 1 1 0 1
  
  
0 0 0 1 1 1
0 0 0 0 1 1
 
2
2
 
∴ v = 2
 
 
2
1

Now, we know that a 5 × 5 alternating matrix looks like,

 
1 −1 1 −1 1
0 1 −1 1 1
 
A = 0 0 1 −1 1
 
−1
 
0 0 0 1
0 0 0 0 1

Verifying now,
    
1 −1 1 −1 1 2 1
0
 1 −1 1 1 2 1
   
Av = 0 0 1 −1 1  2 = 1
    
−1 2 1
    
0 0 0 1
0 0 0 0 1 1 1

Hence, our guess was correct.

10
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 13
How many matrices can you form of order 2 × 2 containing only 1’s and 0’s as entries? How many of these
are invertible?. List them.

Solution
Number of such 2 × 2 matrices = 2 ∗ 2 ∗ 2 ∗ 2 = 24 =
 16.           
0 1 1 0 0 1 1 0 1 1 1 1
Number of such invertible matrices = 6. They are: , , , , ,
1 0 0 1 1 1 1 1 1 0 0 1

Problem 14
Construct a matrix whose null space consists of all multiples of (4, 3, 2, 1).

Solution
4
 
 
  4
. Comparing with −F we get, −F = 3 , I = [1]
3
N (A) = α 
2 I
2
1
Hence, our matrix A must be of the form [I F]. Substituting,
1 0 0 −4
 

A = 0 1 0 −3
0 0 1 −2
4
 
3
Here, the basis vector for N (A) is 
2 whose span will contain all its multiples.


Problem 15
Construct a matrix whose column space contains (1, 1, 0) and (0, 0, 1) and whose null space contains
(1, 0, 1) and (0, 0, 1).

Solution
Let the desired matrix be Am×n . As both N (A) and C(A) are a subset of R3 , m = n = 3.
A is a square matrix.
Now, as the null space is supposed to contain two vectors, nullity of the matrix is 2. By the rank-nullity
theorem, the rank of the matrix must be 1 but the given column space contains 2 basis vectors.
Hence, no such matrix is possible.

11
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 16
Construct a matrix whose column space contains (1, 1, 5) and (0, 3, 1), while the nullspace contains
(1, 1, 2).

Solution
As C(A) contains (1, 1, 5) and (0, 3, 1) and both C(A) and N (A) are subsets of R3 ,
 
1 0 a13
we assume A to be A = 1 3 a23 .
5 1 a33
 
1
Now, solving Ax = 0 for x = 1 where x ∈ N (A),
2
   
  1 0
A1 A2 A3 1 = 0 (Column-wise multiplication)
2 0
       
1 0 a13 0
1 ∗ 1 + 3 ∗ 1 + a23  ∗ 2 = 0
5 1 a33 0
 −1 
2
∴ A3 = −2
−3
 −1 
1 0 2
Hence, our matrix A is 1 3 −2.
5 1 −3

12
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 17
Construct a matrix whose column space contains (1, 1, 1) and whose nullspace is the line of multiples of (1,
1, 1, 1).

Solution
As C(A) ⊂ R3 and N (A) ⊂ R4 , we can say that the matrix A is a 3 × 4 matrix. Moreover, as (1, 1, 1) ∈ C(A),
1
 
 
1 a12 a13 a14 1
we can write matrix A as 1 a22 a23 a24 . Solving Ax = 0 for x =  1,

1 a32 a33 a34
1

1
 
 
  1 0
A1 A2 A3 A4   = 0
   (Column-wise multiplication)
1
0
1
         
1 a12 a13 a14 0
1 ∗ 1 + a22  ∗ 1 + a23  ∗ 1 + a24  ∗ 1 = 0
1 a32 a33 a34 0
∴ a12 + a13 + a14 = −1
∴ a22 + a23 + a24 = −1
∴ a32 + a33 + a34 = −1

Clearly the above system of equations has infinitely many solutions. But for all those values, the rref of
1 0 0 −1
 

matrix will always reduce down to rref (A) = 0 1 0 −1 as the rank of the matrix is 3 (using rank
0 0 1 −1
nullity theorem with nullity = 1 and number of columns = 4).
1 1 2 −4
 

One such matrix is A = 1 7 8 −16

1 9 2 −12

13
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 18
 
If A is a 4 × 4 invertible matrix, describe all vectors in N (B), where B = A A .

Solution
a11 a12 a13 a14
 
a21 a22 a23 a24 
Given A = 
a31
,
a32 a33 a34 
a41 a42 a43 a44
 
B= A A
a11 a12 a13 a14 a11 a12 a13 a14
 
a21 a22 a23 a24 a21 a22 a23 a24 
B=
a31 a32

a33 a34 a31 a32 a33 a34 
a41 a42 a43 a44 a41 a42 a43 a44

Performing row operations to reduce first four columns of B to I will automatically reduce the last four
columns to I as well.
 
∴ B = I4×4 I4×4

Comparing B to [I F], we can see that F = I.

   
−F −I
∴ N (B) = =
I I
Hence, the vectors in null space of B can be described as,
        

 −1 0 0 0  


  0  −1  0   0  

         


  0   0  −1  0  


         
        
 0   0   0  −1

N (B) =   ,   ,   ,  

  1   0   0   0 
         


  0   1   0   0  
         
0 0 1 0 

         

 

 
0 0 0 1

14
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 19
 
1 2 3 4
Consider the matrix A = 2 4 6
 8.
3 5 7 9
Write down a basis for the Column space of A and basis for the Row space of A.

Solution
Performing Gaussian elimination on matrix A,
 
1 2 3 4
A ∼ 0 0 0 0  (R2 → R2 − 2R1 , R3 → R3 − 3R1 )
0 −1 −2 −3
 
1 2 3 4
∼ 0 −1 −2 −3 (swap R2 and R3 )
0 0 0 0
 
1 2 3 4
∼ 0
 1 2 3 (R2 → −R2 )
0 0 0 0
−1 −2
 
1 0
∼ 0
 1 2 3  (R1 → R1 − 2R2 )
0 0 0 0
   
 1 0 
   

0  , 1

Now, the non-zero rows from rref (A) form a basis for the row space of A. ∴ R(A) = −1 2

 
 
−2 3
The pivot columns from theoriginal matrix form a basis for the column space of A which in this case are
   
 1 2 
columns 1 and 2. ∴ C(A) = 2 , 4
 
3 5

Problem 20
     
1 1 2
Construct a matrix such that its nullspace contains 1 and row space contains  2  and −3.
1 −3 5

Solution
The property of every matrix A is that N (A) ⊥ R(A). Here, we first try to verify that
 
1  
1. 1* 1 2 −3 = 1 ∗ 1 + 1 ∗ 2 + 1 ∗ (−3) = 0
1
 
1  
2. 1* 2
 −3 5 = 1 ∗ 2 + 1 ∗ (−3) + 1 ∗ 5 = 4 6= 0.
1
This implies that the property does not hold true and hence, no such matrix is possible.

15
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 21
Construct a matrix such that every row of the matrix is orthogonal to every column (the matrix is not the
zero matrix).

Solution
As every row is orthogonal to every column, matrix A must be a square matrix and its square must be 0,
i.e. A2 = 0.  
0 1
One such matrix is A =
0 0

Problem 22
Solve system of equations

x + y + 5z = 18
2x + 3y + 12z = 44
3x − 4y + z = −2

Is it possible to make one change to the RHS so that the resulting system of equations has a unique solution?
Is it possible to make one change to the RHS so that the resulting equations do not have any solution? Give
reasons.

Solution
The augmented matrix for this system is,
 
1 1 5 18
∼ 2 3 12 44 
3 −4 1 −2
 
1 1 5 18
∼ 0 1 2 8  (R2 → R2 − 2R1 , R3 → R3 − 3R1 )
0 −7 −14 −56
 
1 1 5 18
∼ 0 1 2 8  (R3 → R3 + 7R2 )
0 0 0 0

Clearly, the system has infinitely many solutions due to existence of one free variable.
No such change is possible for an unique solution to exist as no change in RHS can convert the free variables
to non-free variables which is a must for an unique solution to exist.
For Ax = b, if b ∈/ Rm , then no solution exists. Hence, such a change is in fact possible. For eg. change
RHS of third equation to -3 and no solution will exist for the system.

16
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 23
Solve the system of equation

x+y+z =4

Solution
The augmented matrix for the system is,


1 1 1 4
Clearly, the system has infinitely many solutions due to both y and z being free variables.
 
4
Particular solution: taking y = z = 0, x = 4, ∴ xparticular = 0.
0
 
Null space solution: Comparing matrix A = 1 1 1 to [I F], we can write the null space as
−1 −1
 
 
−F
N (A) = = 1 0  So two solutions from the null space for Ax = 0.
I
0 1
For the system Ax = b, the complete solution is

xcomplete = xparticular + c ∗ xnull

−1 −1
     
4
∴ xcomplete = 0 + c1 ∗ 1 + c2 ∗ 0 
    
0 0 1

where c1 , c2 ∈ R.

Problem 24
Let A be a column vector of length n and B be a row vector of length n.

(a) If you can form the product AB, write down the same; if you cannot give reason.

(b) In case you can form the product, what will be the rank of AB? Give reason.

Solution
(a) As A is a n × 1 matrix and B is a 1 × n matrix, the product AB is possible.
 
a11 ∗ b11 a11 ∗ b12 · · · a11 ∗ b1n
That will be ABn×n = 
 .. .. .. .. 
. . . . 
an1 ∗ b11 an1 ∗ b12 · · · an1 ∗ b1n

(b) By property, rank(AB) ≤ rank(A), rank(B)

17
Prayag Savsani Linear Algebra : Assignment #2 AU1841035

Problem 25
A is an n × n matrix. The elements of A are now known, but all its cofactors are known. Explain how you
will find the matrix A.

Solution
Given that cofactors are known, we know adjA. We know,
adjA
A−1 = (1)
det A
adjA = A−1 · det A (2)
−1
A · adjA = AA · det A (Multiplying both sides by A) (3)
A · adjA = det A · In (4)
det(A · adjA) = det(det A · In ) (Taking determinant both sides) (5)
n
(det A) · det(adjA) = (det A) (6)
∴ det(adjA) = (det A)n−1 (7)

NOTE: In the above equation, if n is odd (det A)n−1 = (det −A)n−1 . Hence, matrix A will be determined
upto a factor of -1. Also det A 6= 0, so matrix A is an invertible square matrix.
Now,
adjA
A−1 = (8)
det A
adjA = A−1 · det A (9)
−1 −1 −1 −1
(adjA) · adjA = A · det A · (adjA) (Multiplying both sides by (adjA) ) (10)
−1 −1
I = AA · det A · (adjA) (11)
−1 −1
IA = AA · det A · (adjA) (Multiplying both sides by A) (12)
A
(adjA)−1 = (13)
det A
Now, using the general inverse formula,

adj(adjA)
(adjA)−1 = (14)
det(adjA)
A adj(adjA)
= (Since (13)) (15)
det A det(adjA)
adj(adjA)
A= · det A (16)
det(adjA)
adj(adjA) 1
A= · (det(adjA)) n−1 (Since (7)) (17)
det(adjA)
2−n
∴ A = adj(adjA) · (det(adjA)) n−1 (18)

18