Scribd
Upload
32 views2 pages

Probability Practice Problems With Solutions 5

probability practice problems with solutions

Uploaded by

getasew
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
32 views2 pages

Probability Practice Problems With Solutions 5

probability practice problems with solutions

Uploaded by

getasew
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

STAT 714 MATRIX ALGEBRA REVIEW 5

TERMINOLOGY : A square matrix P is idempotent if P2 = P.

TERMINOLOGY : A square matrix P is a projection matrix onto the vector space S


if and only if
1. P is idempotent
2. Px ∈ S, for any x
3. z ∈ S =⇒ Pz = z (projection).

Result MAR5.1. The matrix P = AA− projects onto C(A).


Proof. Clearly, AA− is a square matrix. Note that AA− AA− = AA− . Note that
AA− x = A(A− x) ∈ C(A). Finally, if z ∈ C(A), then z = Ax, for some x. Thus,
AA− z = AA− Ax = Ax = z. 

NOTE : In general, projection matrices are not unique. However, if we add the require-
ment that Pz = 0, for any z⊥S, then P is called a perpendicular projection matrix,
which is unique. These matrices are important in linear models.

Result MAR5.2. The matrix I − A− A projects onto N (A).


Proof. Clearly, I − A− A is a square matrix. Note that
(I − A− A)(I − A− A) = I − 2A− A + A− A = I − A− A.
For any x, note that (I − A− A)x ∈ N (A) because A(I − A− A)x = 0. Finally, if
z ∈ N (A), then Az = 0. Thus, (I − A− A)z = z − A− Az = z. 

Example. Consider the (linear) subspace of R2 defined by


   
2a
S= z:z= , for a ∈ R
a
and take  
0.8 0.4
P= .
0.4 0.2
Exercise. Show that P is a projection matrix onto S. Show that I − P is a projection
matrix onto S ⊥ , the orthogonal complement of S.

Result MAR5.3. The matrix M is a perpendicular projection matrix onto C(M) if and
only if M is symmetric and idempotent.
Proof. (=⇒) Suppose that M is a perpendicular projection matrix onto C(M) and write
v = v1 + v2 , where v1 ∈ C(M) and v2 ⊥C(M). Also, let w = w1 + w2 , where w1 ∈ C(M)
and w2 ⊥C(M). Since (I − M)v = (I − M)v2 and Mw = Mw1 = w1 , we get
w0 M0 (I − M)v = w10 M0 (I − M)v2 = w10 v2 = 0.
This is true for any v and w, so it must be true that M0 (I − M) = 0 =⇒ M0 = M0 M.
Since M0 M is symmetric, so is M0 , and this, in turn, implies that M = M2 . (⇐=) Now,

PAGE 1
STAT 714 MATRIX ALGEBRA REVIEW 5

suppose that M is symmetric and idempotent. If M = M2 and v ∈ C(M), then since


v = Mb, for some b, we have that Mv = MMb = Mb = v (this establishes that M is
a projection matrix). To establish perpendicularity, note that if M0 = M and w⊥C(M),
then Mw = M0 w = 0, because the columns of M are in C(M). 

Result MAR5.4. If M is a perpendicular projection matrix onto C(X), then C(M) =


C(X).
Proof. We need to show that C(M) ⊆ C(X) and C(X) ⊆ C(M). Suppose that v ∈ C(M).
Then, v = Mb, for some b. Now, write b = b1 + b2 , where b1 ∈ C(X) and b2 ⊥C(X).
Thus, v = Mb = M(b1 + b2 ) = Mb1 + Mb2 = b1 ∈ C(X). Thus, C(M) ⊆ C(X). Now
suppose that v ∈ C(X). Since M is a perpendicular projection matrix onto C(X), we
know that v = Mv = M(v1 + v2 ), where v1 ∈ C(X) and v2 ⊥C(X). But, M(v1 + v2 ) =
Mv1 , showing that v ∈ C(M). Thus, C(X) ⊆ C(M) and the result follows. 

Result MAR5.5. Perpendicular projection matrices are unique.


Proof. Suppose that M1 and M2 are both perpendicular projection matrices onto any
arbitrary subspace S ⊆ Rn . Let v ∈ Rn and write v = v1 + v2 , where v1 ∈ S and v2 ⊥S.
Since v is arbitrary and M1 v = v1 = M2 v, we have M1 = M2 . 

Result MAR5.6. If M is the perpendicular projection matrix onto C(X), then I − M


is the perpendicular projection matrix onto N (X0 ).
Sketch of Proof. I − M is symmetric and idempotent so I − M is the perpendicular
projection matrix onto C(I − M). Show that C(I − M) = N (X0 ); use Result MAR5.5. 

PAGE 2

You might also like