Tutorial 3

MA1513 – Linear Algebra with Differential Equations

Tutor: Christian Go | [Link]@[Link]

1. Determine whether the following sets are (i) linearly independent, and
(ii) bases for R3 .

(a) 𝑆 1 = {(1, 0, −1) , (−1, 2, 3)}

(b) 𝑆 2 = {(1, 0, −1) , (−1, 2, 3) , (0, 3, 0)}
(c) 𝑆 3 = {(1, 0, −1) , (−1, 2, 3) , (0, 3, 3)}
(d) 𝑆 4 = {(1, 0, −1) , (−1, 2, 3) , (0, 3, 0) , (1, −1, 1)}

To begin, recall that a basis for R3 is a set of three linearly independent

vectors or, equivalently, a set of three vectors that span R3 .

(a) The two vectors in the set are not scalar multiples and so 𝑆 1 is
linearly independent. 𝑆 1 is not a basis for R3 since it only contains
two vectors.
(b) We consider the vector equation
𝑐 1 (1, 0, −1) + 𝑐 2 (−1, 2, 3) + 𝑐 3 (0, 3, 0) = 0.
Then, 𝑆 2 is linearly independent if and only if the above vector
equation has only the trivial solution. Row-reducing,

© 1 −1 0 0 ª 𝑅3 +𝑅1 𝑅3 −𝑅2 © 1 −1 0 0 ª
­ 0
­ 2 3 0 ®® −−−−→−−−−→ ­­ 0 2 3 0 ®® .
« −1 3 0 0 ¬ « 0 0 −3 0 ¬
Hence, the vector equation has only the trivial solution, and 𝑆 2 is
linearly independent. Since 𝑆 2 is a set of three linearly independent
vectors in R3 , it must span R3 .
(c) We consider the vector equation
𝑐 1 (1, 0, −1) + 𝑐 2 (−1, 2, 3) + 𝑐 3 (0, 3, 3) = 0.
Notice that we may write the vector equation as a matrix equation,

© 1 −1 0 ª © 𝑐 1 ª © 0 ª
­ 0
­ 2 3 ®® ­­ 𝑐 2 ®® = ­­ 0 ®® .
« −1 3 3 ¬ « 𝑐 3 ¬ « 0 ¬
The homogeneous system will have only the trivial solution if and
only if the coefficient matrix is invertible. We check its determinant:
0
1 −1 0 >

2 3 0 3 0 2
0 2 3 =1 − (−1) +0  = 0.
3 3 −1 3 −1 3
−1 3 3


Since the determinant is zero, the homogeneous system has non-
trivial solutions; thus, 𝑆 3 is linearly dependent, and 𝑆 3 is not a basis
for R3 .
(d) 𝑆 4 is a set of four vectors in R3 : since any more than 3 vectors in
R3 cannot be linearly independent, 𝑆 4 must be linearly dependent,
and is hence not a basis for R3 .
 tutorial 3 2

2. Find a basis for and the dimension of the solution space of each of the
following homogeneous systems.
(
𝑥 1 + 3𝑥 2 − 𝑥 3 + 2𝑥 4 = 0
(a)
− 3𝑥 2 + 𝑥 3 = 0

 𝑥1


 + 3𝑥 2 − 𝑥3 + 2𝑥 4 = 0
(b)

− 3𝑥 2 + 𝑥3 = 0
0


 𝑥1
 − 𝑥4 =

(a) Setting up the augmented matrix of the homogeneous system,

! !
1 3 −1 2 0 𝑅1 +𝑅2 − 13 𝑅2 1 0 0 2 0
−−−−→−−−−→ .
0 −3 1 0 0 0 1 − 13 0 0

Assigning arbitrary parameters to 𝑥 3 and 𝑥 4 , say 𝑥 3 = 𝑠 and 𝑥 4 = 𝑡,

we have
1
𝑥 1 = −2𝑡, 𝑥 2 = 𝑠, 𝑥 3 = 𝑠, 𝑥 4 = 𝑡, 𝑠, 𝑡 ∈ R.
3
Then,
0 −2
© 1 ª
𝑥
© 1 ª ©
® ­ 0 ®
ª
­ 𝑥2 ®
® = 𝑠 ­ 3 ®+𝑡 ­
­ 1 ® ­ 0 ®, for 𝑠, 𝑡 ∈ R.
­
­ ®
­ 3 ®
­ 𝑥 ®
­ ® ­ ®
« 𝑥4 ¬ « 0 ¬ « 1 ¬
Thus, the set {(0, 1/3, 1, 0) , (−2, 0, 0, 1)} is a basis for the solution
space, and the dimension of the solution space is 2.
(b) Setting up the augmented matrix of the homogeneous system,

© 1 3 −1 2 0
ª 𝑅3 −𝑅1 𝑅1 +𝑅2 − 13 𝑅2 𝑅1 −2𝑅3 ©
1 0 0 0 0
­ 0 −3 1 0 0 ® −−−−→−−−−→−−−−→−−−−−→ ­ 0 1 − 1 0 0 ®.
ª
­ ® 𝑅3 −𝑅2 − 1 𝑅3 ­ 3 ®
« 1 0 0 −1 0 0 0 0 1 0
3
¬ « ¬
Assigning an arbitrary parameter 𝑥 3 = 𝑡, we have

0 0
© 1 ª © 1 ª
𝑥
© 1 ª
­ 𝑥2 ® ­ 𝑡 ®
® ­ 3 ® ­ 3 ®
­ 𝑥 ® = ­ 𝑡 ® = 𝑡 ­ 1 ®, for 𝑡 ∈ R.
­ ®
­
­ 3 ® ­ ® ­ ®
« 𝑥4 ¬ « 0 ¬ « 0 ¬
Thus, {(0, 1/3, 1, 0)} is a basis for the solution space, and the dimen-
sion of the solution space is 1.
 tutorial 3 3

3. The regular coordinate axes for the 𝑥𝑦𝑧-space is rotated about the
𝑧-axis through 45◦ counterclockwise, viewing from the positive 𝑧-axis.

(a) Find a basis 𝑆 consisting of unit vectors that determine the new
coordinate axes.
(b) State the coordinates of the vector 𝒗 = (1, 1, 1) relative to the new
axes.
(c) Find a matrix 𝑴 such that 𝑴𝒗 = (𝒗)𝑆 .

Since we rotate about the 𝑧-axis, it remains fixed. To visualize what

𝑦
happens on the 𝑥𝑦-plane, we take a top view of 𝑥𝑦𝑧-space, so that the 𝑦0 𝑥0
𝑧-axis points out of the page.
45◦
(a) In order to find a basis that determines the coordinate axes, it suf-
45◦
fices to consider how the standard R3 basis vectors are transformed 𝑥
by the rotation. Observe that while (0, 0, 1) is fixed,
√ √ ! √ √ !
2 2 2 2
(1, 0, 0) ↦→ , , 0 , (0, 1, 0) ↦→ − , ,0 .
2 2 2 2

Note that the unit vector along the 𝑧-axis remains unchanged by the
rotation. Hence, we have a basis
( √ √ ! √ √ ! )
2 2 2 2
𝑆= , ,0 , − , , 0 , (0, 0, 1) .
2 2 2 2

(b) We wish to express 𝒗 in terms of the vectors in 𝑆; that is,

√ √
2 2
© −√ 2
√2 © 0 ª © 1 ª
2 2 ®+𝑧 ­ 0 ® = ­ 1 ®.
© ª ª
𝑥­ ® +𝑦 ­ 2
­ ® ­
2 ® ­ ® ­ ®
« 0 ¬ « 0 ¬ « 1 ¬ « 1 ¬
Solving this system,
√ √ √ √
2 2
− 0 1 0 01 2 © 𝑥 ª © 2 ª
√2 √2 ª GJE ©
2 2 ® =⇒ ­ 𝑦 ® = ­ 0 ® .
©
­ 0 1 0 0
ª
­
­ 2 0 1 ®
2
® −−→ ­
® ­ ® ­ ®
« 0 0 1 1 ¬ « 0 0 1 1 ¬ « 𝑧 ¬ « 1 ¬
√ 
Hence, (𝒗)𝑆 = 2, 0, 1 .
(c) Observe that our linear combination in the previous part can be
expressed as the matrix product 𝑨𝒙 = 𝒃, where
√ √ √
2 2
− 0 2 1
√2 √2
2 2
0 ® ­ 0 ® = ­ 1 ®® .
© ª© ª © ª
­ ® ­ ® ­
­ 2 2
« 0 0 1 ¬« 1 ¬ « 1 ¬

Since the columns of 𝑨 form a basis, it follows that 𝑨 is a square

matrix whose columns are linearly independent, and hence, 𝑨 must
be invertible. Hence,
√ √ √ −1 √ √
2 2 2 2
2 − 0 1 0 1
√2 √2 √2 √2
2 2 2 2
0 ® ­ 1 ®® .
© ª © © ª © ª© ª
0
ª
­
­
®=­
® ­ 2 2 0 ®® ­
­ 1 ® = ­−
® ­ 2 2
® ­

« 1 ¬ « 0 0 1 ¬ « 1 ¬ « 0 0 1 ¬« 1 ¬
 tutorial 3 4

4. Let
1 −1 0 1
­ 0 1 −1 ®® © 𝑥 ª ­ 1 ®
© ª © ª
𝑨 = ­­ , 𝒙 = ­­ 𝑦 ®® , 𝒃 = ­­ ®® .
­ −1 0 1 ®® ­ 1 ®
« 𝑧 ¬
« 1 1 1 ¬ « 1 ¬
(a) Show that the linear system 𝑨𝒙 = 𝒃 has no solution.
(b) Find the least squares solution to 𝑨𝒙 = 𝒃.
(c) Find the projection of 𝒃 onto span {(1, 0, −1, 1) , (−1, 1, 0, 1) , (0, −1, 1, 1)} .
(d) Find a vector that is orthogonal to all three vectors (1, 0, −1, 1) , (−1, 1, 0, 1) , (0, −1, 1, 1) .

(a) Performing Gaussian Elimination on (𝑨 | 𝒃) ,

1 −1 0 1 1 −1 0 1
­ 0 1 1 ® 𝑅3 +𝑅1 𝑅3 +𝑅2 𝑅2 ↔𝑅3 ­ 0 1 −1 1
© ª © ª
­ −1 ® −−−−→−−−−−→−−−−−→ ­ ®.
®
­ −1 0 1 1 ® 𝑅4 −𝑅1 𝑅4 −2𝑅2 ­ 0 0 3 −2 ®
­ ® ­ ®
« 1 1 1 1 ¬ « 0 0 0 1 ¬
Observe that the last row corresponds to the equation 0 = 1, imply-
ing that the system is inconsistent and 𝑨𝒙 = 𝒃 has no solution.
(b) The least squares solution to an inconsistent system is simply the
solution to the consistent system 𝑨𝑇 𝑨𝒙 = 𝑨𝑇 𝒃. Hence,

1 −1 0 1
© 1 0 −1 1 © 1 0 −1 1 © ª
ª 𝑥
ª­ 0 1 −1 ®© ª © ª­ 1 ®
­ −1 1 0 1 ®­ ® ­ 𝑦 ® = ­ −1 1 0 1 ®® ­­ ®®
­ ® ­ −1 0 1 ®® ­ ® ­ 1
0 −1 1 1 ­
¬ 1 𝑧 ¬ « 0 −1 1 1 ¬ ­ ®
«
« 1 1 ¬« « 1 ¬

© 3 0 0 𝑥 1
®­ 𝑦 ® = ­ 1 ®,
ª© ª © ª
∴ ­­ 0 3 0 ®­ ® ­ ®
« 0 0 3 ¬« 𝑧 ¬ « 1 ¬
and hence, we have the solution 𝑥 = 𝑦 = 𝑧 = 1/3.
(c) Observe that the vectors in linear span form the columns of the
matrix 𝑨; that is, this linear span is precisely the column space of
𝑨. Hence, the projection of 𝒃 onto the column space of 𝑨 is simply
𝒑 = 𝑨𝒖, where 𝒖 is a least square solution. Then,

1 −1 0 1 0
­ 0 1 −1 ®® ©­ 3 ª ­ 0 ®
© ª © ª
1 ® = ­ ®.
𝒑 = 𝑨𝒖 = ­­ 3
­ −1 0 1 ®® ­ ® ­ 0 ®
1 ­ ®
3
« 1 1 1 ¬« « 1 ¬
¬

(d) Since 𝒑 is the projection of 𝒃 onto the span of the three vectors,
their difference must be orthogonal to the span of those vectors (and
hence the vectors themselves):

1 0 1
­ 1 ® ­ 0 ® ­ 1 ®
© ª © ª © ª
𝒃 − 𝒑 = ­­ ®® − ­­ ® = ­ ®.
­ 1 ® ­ 0
® ­ 1 ®
® ­ ®
« 1 ¬ « 1 ¬ « 0 ¬
 tutorial 3 5

5. To measure the takeoff performance of an airplane, the horizontal

𝑡 𝑦
position of the plane was measured every second, from 𝑡 = 0 to 𝑡 = 12.
0 0
The positions (in meters) are given in the table. 1 2.9
2 9.8
(a) Find the least squares cubic curve: 𝑦 = 𝑎 + 𝑏𝑡 + 𝑐𝑡 2 + 𝑑𝑡 3 for this data 3 20.1
set. 4 34.3
5 52.8
(b) Use the result from the previous part to estimate the velocity of the 6 74.0
plane when 𝑡 = 4.5 seconds. 7 98.5
8 126.7
9 157.2
(a) We wish to find the curve 𝑦 = 𝑎 + 𝑏𝑡 + 𝑐𝑡 2 + 𝑑𝑡 3 that best fits the 10 190.3
13 data points (𝑡, 𝑦) . Notice that this yields a linear system with 13 11 228.6
12 269.0
equations (one for each data point), in the unknowns 𝑎, 𝑏, 𝑐, 𝑑.
Let us write the system in the form 𝑻 𝒙 = 𝒚, where each row in the
matrix 𝑻 correspond to the coefficients 1, 𝑡, 𝑡 2 , 𝑡 3 of each linear equa-
tion, 𝒙 is the vector of variables, and 𝒚 is the vector of constants;
that is, the measured positions at each time 𝑡.
To begin, notice that the first column of 𝑻 consists of all ones, while
the second column of 𝑻 consists of the numbers one through 12.
>‌> t1 = ones(13,1)
>‌> t2 = [0:12]’
The third and fourth columns of 𝑡 are precisely the squares and
cubes of the numbers 0 through 12. We can use the “.^” command
to compute element-wise powers of the vector t2.
>‌> t3 = t2.^2
>‌> t4 = t2.^3
Hence, we have the matrix 𝑻 , as well as the column matrix 𝒚.
>‌> T = [t1 t2 t3 t4]
>‌> y = [0 2.9 9.8 20.1 34.3 52.8 74.0 98.5 126.7
157.2 190.3 228.6 269.0]’
To find the least squares solution to this linear system, it suffices to
solve the normal equation 𝑻 𝑇 𝑻 𝒙 = 𝑻 𝑇 𝒚.
>‌> rref([T’*T T’*y])

1 0 0 0 −0.1758 𝑎 ≈ −0.1758
  ©­ 0 1 0 0 1.1716
ª
𝑏 ≈ 1.1716
rref 𝑻 𝑇 𝑻 | 𝑻 𝑇 𝒚 = ­­ .
®
® =⇒
­ 0 0 1 0 1.9453 𝑐 ≈ 1.9453
®
®
« 0 0 0 1 −0.0147 ¬ 𝑑 ≈ −0.0147
We thus have a unique least squares solution, and the cubic curve of
best fit is given by

𝑦 = −0.1758 + 1.1716𝑡 + 1.9453𝑡 2 − 0.0147𝑡 3 .

(b) The velocity function 𝑣 (𝑡) of the plane is given by the first deriva-
tive of its distance function; hence,

𝑣 (𝑡) = 𝑦 0 (𝑡) = 1.1716 + 2 × 1.9453𝑡 − 3 × 0.0147𝑡 2 .

Thus, at 𝑡 = 4.5 s, the velocity is approximately 𝑣 (4.5) ≈ 17.7863

meters per second.
 tutorial 3 6

6. For each of the following matrices

© 1 2 0 ª 2 1 4 1 2
­ 0 1 1 ®
4 2 2 3 2 ®® © 1 4 5 8 ª
­ ® © ª
𝑨 = ­ −1 3 6 ® , 𝑩 = ­­ , ­ −1 4 3 0 ® ,
­
𝑪 =
­ ®
­ 2 1 −2 2 0 ®
® ­ ®
­ 2 1 0 ® 2 0 2 1
­ ®
« 6 3 6 4 4 ¬
­ ® « ¬
« 3 1 −1 ¬
find a basis for the row space and a basis for the column space; find
a basis for the nullspace; and find the nullity of the matrix. For each
matrix, verify the Dimension Theorem for Matrices.
We have previously already found the reduced row-echelon forms of
these matrices.
© 1 0 0 ª
­ 0 1 0 ®
­ ®
rref (𝑨) = ­ 0 0 1 ®
­ ®
­ 0 0 0 ®
­ ®
­ ®
« 0 0 0 ¬
Consider the matrix 𝑨, which we have established has rank 3.

• By considering the reduced row-echelon form of 𝑨, we find that the

row space of 𝑨 is given by the basis 𝑅𝐴 = {(1, 0, 0) , (0, 1, 0) , (0, 0, 1)} .
• The pivot columns in rref (𝑨) correspond to linearly independent
columns in 𝑨. Hence, a basis for the column space of 𝑨 is given by


 © 1 ª © 2 ª © 0 
ª

0 1 1

 ­ ® ­ ® ­ ®


 ®

 ­­ ® ­ ® ­
®,­ ®,­ ® .
 ®

𝐶𝐴 = ­ −1 3 6
® ­ ® ­
2 1 0

 ­ ® ­ ® ­ ®
­
 ® ­ ® ­ ®

 ­ ® ­ ® ­ ®

3 1
 

« ¬ « ¬ « −1 
¬
• The nullspace of 𝑨 is the solution space of the homogeneous system
𝑨𝒙 = 0. The system only has the trivial solution, and the basis for
the nullspace is the empty set 𝑁𝐴 = {} .
• The nullity is the dimension of the nullspace: nullity (𝑨) = 0.
• Indeed, the sum of the rank and nullity of 𝑨 is 3 + 0 = 3, equal to the
number of columns of 𝑨.
1 5 1
1 2 0 6 3
­ 0 0 1 − 16 1
© ª
rref (𝑩) = ­­ 3
®
®
­ 0 0 0 0 0 ®®
« 0 0 0 0 0 ¬
Consider the matrix 𝑩, which we have established has rank 2.

• By considering the reduced row-echelon form of 𝑩, we find that the

row space of 𝑩 is given by the basis 𝑅𝐵 = 1, 12 , 0, 56 , 13 , 0, 0, 1, − 16 , 13 .


• The pivot columns in rref (𝑩) correspond to linearly independent

columns in 𝑩. Hence, a basis for the column space of 𝑩 is given by


 2 4  
ª
4 ® ­ 2 ®

 ©­ ª © 
®,­ ® .

 

𝐶𝐵 = ­­

 ­ 2 ® ­ −2 ® 
® ­ ®

6 6
 

 
« ¬ « ¬
 tutorial 3 7

• The nullspace of 𝑩 is the solution space of the homogeneous system

𝑩𝒙 = 0. Assigning arbitrary parameters to non-pivot columns, the
reduced row-echelon form of 𝑩 gives us the solution

© 𝑥1 ª © − 12 ª © − 56 ª © − 31 ª
­
­ 𝑥2 ®
®
­
­ 1 ®
®
­
­ 0 ®® ­­ 0 ®
®
1 ®
®=𝑟­ 0 ®+𝑠 ­ +𝑡 ­ − 13 ®.
­ ® ­ ® ­ ­ ®
­ 𝑥3 6 ®
0 1 ®® ­­ 0
­ ® ­ ® ­ ® ­ ®
­
­ 𝑥4 ®
®
­
­
®
®
­
­
®
®
« 𝑥5 ¬ « 0 ¬ « 0 ¬ « 1 ¬
Thus, a basis for the nullspace is given by



 © − 12 ª © − 56 ª © − 13 ª 

1 0 ®® ­­ 0 ®® 

 ­ ® ­ 

 

 ­­ ® ­
1 ® ­ 1 ®
®,­ , .
 

𝑁𝐵 = ­ 0 −
® ­
6 ® ­
® ­ 3 ®
®
0 1 ®® ­­ 0 ®® 

 ­ ® ­
­
 ® ­ 

 ­ ® ­ 
0 0 ¬ « 1 ¬

 
« ¬ « 

• The dimension of the nullspace of 𝑩 is nullity (𝑩) = 3.

• Indeed, the sum of the rank and nullity of 𝑩 is 2 + 3 = 5 equal to the
number of columns of 𝑩.

© 1 0 1 0 ª
rref (𝑪) = ­­ 0 1 1 0 ®®
« 0 0 0 1 ¬
Consider the matrix 𝑪, which we have established has rank 3.

• By considering the reduced row-echelon form of 𝑪, we find that the

row space of 𝑪 is given by the basis 𝑅𝐶 = {(1, 0, 1, 0) , (01, 1, 0) , (0, 0, 0, 1)} .
• The pivot columns in rref (𝑪) correspond to linearly independent
columns in 𝑪. Hence, a basis for the column space of 𝑪 is given by

© 1 ª © 5 ª © 8 ª

 


𝐶𝐶 = ­ −1 ® , ­ 3 ® , ­ 0 ® .
 

­ ® ­ ® ­ ®
 2
¬ « 2 ¬ « 1 ¬

 


«
• The nullspace of 𝑪 is the solution space of the homogeneous system
𝑪𝒙 = 0. Assigning arbitrary parameters to non-pivot columns, the
reduced row-echelon form of 𝑪 gives us the solution

−1 −1
© 1 ª
𝑥 




© ª 
 ©­ ª

­ 𝑥2 ® ­ −1 ® −1 ®
® .

 

­ ®=𝑡­ ® =⇒ 𝑁𝐶 = ­­
­ 1 ® 1
­ 3 ®
­ 𝑥 ®  ®
­ ®  ­ ®
0

0
 

𝑥 4
 
« ¬ « ¬ « ¬

• The dimension of the nullspace of 𝑪 is nullity (𝑪) = 1.

• Indeed, the sum of the rank and nullity of 𝑪 is 3 + 1 = 4, equal to the
number of columns of 𝑪.
 tutorial 3 8

7. Consider two 3 × 4 matries 𝑨 and 𝑩 with respective row-echelon forms

© 1 0 1 2 © 0 1 0 1
𝑹 𝑨 = ­­ 0 0 1 1 ®, ®.
ª
𝑹 𝑩 = ­­ 0 0 1
ª
® −1 ®
« 0 0 0 0 ¬ « 0 0 0 1 ¬
Determine if we have enough information to find:

(a) the matrices 𝑨 and 𝑩,

(b) the row spaces of 𝑨 and 𝑩,
(c) the column spaces of 𝑨 and 𝑩.

If possible, write down a basis for each of the row spaces and column
spaces of 𝑨 and 𝑩.

(a) We are unable to recover the original matrices 𝑨 and 𝑩 without

knowing the elementary row operations taken to obtain the corre-
sponding row-echelon forms.
(b) The row space of a matrix is preserved by elementary row opera-
tions; thus,

row (𝑨) = span {(1, 0, 1, 2) (0, 0, 1, 1) (0, 0, 0, 0)} ,

row (𝑩) = span {(0, 1, 0, 1) (0, 0, 1, −1) , (0, 0, 0, 1)} .

In particular, the linearly independent rows of 𝑹 𝑨 and 𝑹 𝑩 form

bases for the row spaces of 𝑨 and 𝑩, respectively, so a basis for
row (𝑨) is given by {(1, 0, 1, 2) (0, 0, 1, 1)} , and a basis for row (𝑩) is
given by {(0, 1, 0, 1) (0, 0, 1, −1) , (0, 0, 0, 1)} .
(c) Linear independence of columns is preserved by elementary row
operations. Thus, the linearly independent columns of 𝑹 𝑨 and 𝑹 𝑩
determine which columns of 𝑨 and 𝑩 form a basis for the column
space of each matrix.
Consider 𝑹 𝑨 : only the first and third columns are linearly inde-
pendent; thus, the first and third columns of the original matrix 𝑨
would form a basis for col (𝑨) ⊆ R3 . Since we have no information
on the original matrix 𝑨, we are unable to determine col (𝑨) .
Now, consider 𝑹 𝑩 : the second, third, and fourth columns are lin-
early independent; hence, the second, third, and fourth columns
of 𝑩 form a basis for col (𝑩) ⊆ R3 . Note that, in this case, three
linearly independent vectors in R3 span col (𝑩): thus, col (𝑩) must
be all of R3 . Though we do not have the original columns of 𝑩, we
know that they form a basis for all of R3 . Thus, we may take any
basis for R3 as our basis for col (𝑩) , for instance, the standard basis:

© 1 ª © 0 ª © 0 ª
­ 0 ®, ­ 1 ®, ­ 0 ®.
­ ® ­ ® ­ ®
« 0 ¬ « 0 ¬ « 1 ¬