Asgn2 LST
Asgn2 LST
SOLUTION
2p 3 2 3
1=p2 2=3 p 3
2
U x D 4 1= 2 2=3 5 D 4 15
3
0 1=3 1
p p
kU xk D 9 C 1 C 1 D 11
p p
kxk D 2 C 9 D 11
Theorems 6 and 7 are particularly useful when applied to square matrices. An
orthogonal matrix is a square invertible matrix U such that U 1 D U T . By Theo-
rem 6, such a matrix has orthonormal columns.1 It is easy to see that any square matrix
with orthonormal columns is an orthogonal matrix. Surprisingly, such a matrix must
have orthonormal rows, too. See Exercises 27 and 28. Orthogonal matrices will appear
frequently in Chapter 7.
PRACTICE PROBLEMS
p p
1=p5 2=p5
1. Let u1 D and u2 D . Show that fu1 ; u2 g is an orthonormal basis
2= 5 1= 5
for R2 .
2. Let y and L be as in Example
3 and Figure 3. Compute the orthogonal projection yO
2
of y onto L using u D instead of the u in Example 3.
1
p
3 2
3. Let U and x be as in Example 6, and let y D . Verify that U x U y D x y.
6
4. Let U be an n n matrix with orthonormal columns. Show that det U D ±1.
6.2 EXERCISES
2 3 2 3 2 3 2 3 2 3 2 3
In Exercises 1–6, determine which sets of vectors are orthogonal. 3 1 3 5 4 3
2 3 2 3 2 3 2 3 2 3 2 3 6 27 6 37 687 6 47 6 17 6 37
1 5 3 1 0 5 5. 6 7 6
4 1 5, 4
7, 6 7 6. 6 7 6
4 0 5, 4
7, 6 7
35 475 3 4
5 55
1. 4 4 5, 4 2 5, 4 4 5 2. 4 2 5, 4 1 5, 4 2 5 3 4 0 3 8 1
3 1 7 1 2 1
In Exercises 7–10, show that fu1 ; u2 g or fu1 ; u2 ; u3 g is an orthog-
onal basis for R2 or R3 , respectively. Then express x as a linear
2 3 2 3 2 3 2 3 2 3 2 3 combination of the u’s.
2 6 3 2 0 4
3. 4 7 5, 4 3 5, 4 1 5 4. 4 5 5, 4 0 5, 4 2 5 2 6 9
7. u1 D , u2 D , and x D
1 9 1 3 0 6 3 4 7
1A better name might be orthonormal matrix, and this term is found in some statistics texts. However,
orthogonal matrix is the standard term in linear algebra.
3 2 6 b. If y is a linear combination of nonzero vectors from an
8. u1 D , u2 D , and x D
1 6 3 orthogonal set, then the weights in the linear combination
2 3 2 3 2 3 2 3 can be computed without row operations on a matrix.
1 1 2 8
9. u1 D 4 0 5, u2 D 4 4 5, u3 D 4 1 5, and x D 4 4 5 c. If the vectors in an orthogonal set of nonzero vectors are
1 1 2 3 normalized, then some of the new vectors may not be
2 3 2 3 2 3 2 3 orthogonal.
3 2 1 5 d. A matrix with orthonormal columns is an orthogonal
10. u1 D 4 3 5, u2 D 4 2 5, u3 D 4 1 5, and x D 4 3 5 matrix.
0 1 4 1
e. If L is a line through 0 and if yO is the orthogonal projection
1 of y onto L, then kOyk gives the distance from y to L.
11. Compute the orthogonal projection of onto the line
7
4 24. a. Not every orthogonal set in Rn is linearly independent.
through and the origin.
2 b. If a set S D fu1 ; : : : ; up g has the property that ui uj D 0
whenever i ¤ j , then S is an orthonormal set.
1
12. Compute the orthogonal projection of onto the line c. If the columns of an m n matrix A are orthonormal, then
1
the linear mapping x 7! Ax preserves lengths.
1
through and the origin. d. The orthogonal projection of y onto v is the same as the
3
orthogonal projection of y onto c v whenever c ¤ 0.
2 4
13. Let y D and u D . Write y as the sum of two e. An orthogonal matrix is invertible.
3 7
orthogonal vectors, one in Span fug and one orthogonal to u. 25. Prove Theorem 7. [Hint: For (a), compute kU xk2 , or prove
(b) first.]
2 7
14. Let y D and u D . Write y as the sum of a vector 26. Suppose W is a subspace of Rn spanned by n nonzero
6 1
in Span fug and a vector orthogonal to u. orthogonal vectors. Explain why W D Rn .
3 8 27. Let U be a square matrix with orthonormal columns. Explain
15. Let y D and u D . Compute the distance from y to why U is invertible. (Mention the theorems you use.)
1 6
the line through u and the origin. 28. Let U be an n n orthogonal matrix. Show that the rows of
U form an orthonormal basis of Rn .
3 1
16. Let y D and u D . Compute the distance from y
9 2 29. Let U and V be n n orthogonal matrices. Explain why
to the line through u and the origin. UV is an orthogonal matrix. [That is, explain why UV is
In Exercises 17–22, determine which sets of vectors are orthonor- invertible and its inverse is .UV /T .]
mal. If a set is only orthogonal, normalize the vectors to produce 30. Let U be an orthogonal matrix, and construct V by inter-
an orthonormal set. changing some of the columns of U . Explain why V is an
2 3 2 3 2 3 2 3 orthogonal matrix.
1=3 1=2 0 0
17. 4 1=3 5, 4 0 5 18. 4 1 5, 4 1 5 31. Show that the orthogonal projection of a vector y onto a line
1=3 1=2 0 0 L through the origin in R2 does not depend on the choice
2 3 2 3 of the nonzero u in L used in the formula for yO . To do this,
2=3 1=3
:6 :8 suppose y and u are given and yO has been computed by
19. , 20. 4 1=3 5, 4 2=3 5
:8 :6 formula (2) in this section. Replace u in that formula by c u,
2=3 0
where c is an unspecified nonzero scalar. Show that the new
2 p 3 2 p 3 2 3 formula gives the same yO .
1=p10 3=p10 0p
21. 4 3=p20 5, 4 1=p20 5, 4 1=p2 5 32. Let fv1 ; v2 g be an orthogonal set of nonzero vectors, and let
3= 20 1= 20 1= 2 c1 , c2 be any nonzero scalars. Show that fc1 v1 ; c2 v2 g is also
2 p 3 2 p 3 2 3 an orthogonal set. Since orthogonality of a set is defined in
1=p18 1= 2 2=3 terms of pairs of vectors, this shows that if the vectors in
22. 4 4=p18 5, 4 0p 5, 4 1=3 5 an orthogonal set are normalized, the new set will still be
1= 18 1= 2 2=3 orthogonal.
In Exercises 23 and 24, all vectors are in Rn . Mark each statement 33. Given u ¤ 0 in Rn , let L D Span fug. Show that the mapping
True or False. Justify each answer. x 7! projL x is a linear transformation.
23. a. Not every linearly independent set in Rn is an orthogonal 34. Given u ¤ 0 in Rn , let L D Span fug. For y in Rn , the
set. reflection of y in L is the point reflL y defined by
2 3
reflL y D 2 projL y y 6 3 6 1
6 1 2 1 67
See the figure, which shows that reflL y is the sum of 6 7
6 3 6 3 27
yO D projL y and yO y. Show that the mapping y 7! reflL y 6 7
6 6 3 6 17
is a linear transformation. AD6
6
7
6 2 1 2 377
x2 6
y 6 3 6 3 277
4 2 1 2 35
L = Span{u}
1 2 1 6
ŷ
ref l L y 36. [M] In parts (a)–(d), let U be the matrix formed by normal-
y – yˆ u izing each column of the matrix A in Exercise 35.
a. Compute U TU and U U T . How do they differ?
x1
b. Generate a random vector y in R8 , and compute
p D U U Ty and z D y p. Explain why p is in Col A.
yˆ – y Verify that z is orthogonal to p.
The reflection of y in a line through the origin. c. Verify that z is orthogonal to each column of U .
35. [M] Show that the columns of the matrix A are orthogonal d. Notice that y D p C z, with p in Col A. Explain why z is
by making an appropriate matrix calculation. State the calcu- in .Col A/? . (The significance of this decomposition of
lation you use. y will be explained in the next section.)