Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Quiz 5 Linear Algebra ⋅ Spring 2019

Name:

Rules: Textbooks, notes, calculators, cell phones or headphones are not allowed.
All work must be shown for full credit. Partial credit may be awarded.

1. Consider the following matrix


3 2
B=[ ].
2 3
Is B positive definite? Explain your answer.
Yes, it is. We simply check that the eigenvalues of B are positive

(3 − λ)2 − 4 = 0 Ô⇒ λ = 1, λ = 5

2. Compute the SVD of the following matrix

2 −1
A=[ ]
−2 1

Hint: write A = U ΣV T .
To compute the SVD, we find the eigenvalues of AT A to find an orthonormal basis of the row
space (which we put as columns in the matrix V ). Observe that

8 −4
AT A = [ ] Ô⇒ λ = 0, 10
−4 2

which is singular, since A is singular. Then the corresponding eigenvalues are

−2 −4 −2
[ ] v 1 = 0 Ô⇒ v 1 = [ ]
−4 −8 1

So we have only one vector for the row space. To find an orthogonal vector to this one, we
find the basis for the nullspace, which we know is orthogonal

8 −4 1
[ ] v 2 = 0 Ô⇒ v 2 = [ ]
−4 2 2

Then, we normalize these vectors and put them the matrix V with their corresponding eigen-
values in Σ,
⎡− √2 √1 ⎤ √
⎢ 5
⎢ 5

⎥ 10 0
V =⎢ 1 2 ⎥, Σ=[
√ √
]
⎢ 5 5⎥
0 0
⎣ ⎦
To find u1 , we can find the eigenvalues of AAT , or we can just solve
⎡− √5 ⎤ √ ⎡− √1 ⎤
⎢ 5⎥ ⎢ ⎥
Av 1 = σ1 u1 Ô⇒ ⎢ 5 ⎥ = 10u1 Ô⇒ u1 = ⎢⎢ 1 2 ⎥⎥
⎢ ⎥
⎢ √5 ⎥ ⎢ √2 ⎥
⎣ ⎦ ⎣ ⎦
Then, to find u2 , we need an additional vector that is orthogonal to u1 in the column space,
so we find the basis for the left nullspace:

2 −2 1 1
AT u2 = 0 Ô⇒ [ ] u2 = 0 Ô⇒ u2 = √ [ ]
−1 1 2 1
Finally,
⎡− √1 √1 ⎥ 0 ⎢⎢− √5 √1 ⎥
⎤ √ ⎡ 2 ⎤
2 ⎥ [ 10
T ⎢
A = U ΣV = ⎢⎢ 1 2 5⎥
⎢ √2 √1 ⎥ 0 ⎢ √5 √2 ⎥
]⎢ 1
2⎥
0 5⎥
⎣ ⎦ ⎣ ⎦

You might also like