2DRR00 20240126
2DRR00 20240126
2DRR00 20240126
(3) Let y = x1 + x2 t be the line that best fits the 3 points (0, −1), (1, 3), and (2, 1) in
the least squares sense. Then the first coordinate x1 is equal to:
1 1 3
(a) − (b) 0 (c) (d) 1 (e)
2 2 2
1 −1
(4) Let A = . The square of the 2-norm kAk2 (= kAk22 ) is equal to:
0 1
√ √ √ √
(a) 12 (3 + 3) (b) 12 (3 + 5) (c) 12 (3 + 7) (d) 1 (e) 3
α β
(5) Let α, β ∈ R, and suppose that P = is a projection.
0 0
Consider the following two statements:
(i) Then α2 = α (ii) Then α β = α
(a) Both statements are true (b) Statement (i) is true; (ii) is false
(c) Statement (i) is false; (ii) is true (d) Both statements are false (e) —–
(18) Let A, B, C be 2 × 2 matrices, with A = BC. Consider the following two state-
ments:
(i) The rows of A are linear combinations of the rows of B
(ii) The columns of A are linear combinations of the columns of B
(a) Both statements are true (b) Statement (i) is true; (ii) is false
(c) Statement (i) is false; (ii) is true (d) Both statements are false (e) —–
Formulas: SVD: singular values and left and right singular vectors
SVD (A : m × n) A = U ΣV T , U : m × m, Σ : m × n, V : n × n
Economical SVD A = U ΣV T , U : m × n, Σ : n × n, V : n × n (m ≥ n)
T
Singular values/vectors A Avj = σj2 vj , T
Avj = σj uj , A uj = σj vj
σ1
..
on block level A [v1 . . . vn ] = [u1 . . . un ]
. , or short: AV = U Σ
σn
also AT = V ΣU T and AT U = V Σ
Can also be written as A = σ1 u1 v1T + · · · + σn un vnT
This is used in Class 14 for an TSVD: A ≈ σ1 u1 v1T (rank-1)
vvT
Orthogonal projection in R2 I− v ⊥ axis
vT v
uvT
Oblique projection in R2 I− v ⊥ axis, u “along”
vT u
Axis of projection in R2 Px = x (onto)
Direction of projection in R2 Pu = 0 (along)
vvT
Reflection in R2 I −2 v ⊥ axis
vT v
Axis of reflection in R2 Rx = x (across)
————–END OF EXAM————–
5