Solutions to Axler, Linear Algebra Done Right 2nd Ed.

Edvard Fagerholm
edvard.fagerholm@helsinki. [gmail.com
Beware of errors. I read the book and solved the exercises during s
pring break (one
week), so the problems were solved in a hurry. However, if you do nd better or
interesting
solutions to the problems, Id still like to hear about them. Also please dont pu
t this on
the Internet to encourage copying homework solutions...
1 Vector Spaces
1. Assuming that C is a eld, write z = a + bi. Then we have 1/z
= z/zz = z/[z[
2
.
Plugging in the numbers we get 1/(a + bi) = a/(a
2
+ b
2
) bi/(a
2
+ b
2
) = c + di. A
straightforward calculation of (c+di)(a+bi) = 1 shows that this is indeed an inv
erse.
2. Just calculate ((1 +

3)/2)
3
.
3. We have v + (v) = 0, so by the uniqueness of the additive inv
erse (prop. 1.3)
(v) = v.
4. Choose a ,= 0 and v ,= 0. Then assuming av = 0 we get v = a
1
av = a
1
0 = 0.
Contradiction.
5. Denote the set in question by A in each part.
(a) Let v, w A, v = (x
1
, x
2
, x
3
), w = (y
1
, y
2
, y
3
). Then x
1
+ 2x
2
+ 3x
3

= 0 and
y
1
+ 2y
2
+ 3y
3
= 0, so that 0 = x
1
+ 2x
2
+ 3x
3
+ y
1
+ 2y
2
+ 3y
3
= (x
1
+ y
1
) +
2(x
2
+y
2
) +3(x
3
+y
3
), so v +w A. Similarly 0 = a0 = ax
1
+2ax
2
+3ay
3
, so
av A. Thus A is a subspace.
(b) This is not a subspace as 0 , A.
(c) We have that (1, 1, 0) A and (0, 0, 1) A, but (1, 1, 0)+(0, 0, 1) = (1, 1,
1) , A,
so A is not a subspace.
(d) Let (x
1
, x
2
, x
3
), (y
1
, y
2
, y
3
) A. If x
1
= 5x
3

and y
1
= 5y
3
,
1

then ax

= 5ax
3
,
so a(x
1
, x
2
, x
3
) A. Similarly x
1
+ y
1
= 5(x
3
+ y
3
), so that (x
1
, x
2
, x
3
) +
(y
1
, y
2
, y
3
) A. Thus A is a subspace.
1
6. Set U = Z
2
.
7. The set (x, x) R
2
[ x R (x, x) R
2
[ x R is closed under multiplication
but is trivially not a subspace ((x, x) + (x, x) = (2x, 0) doesnt belong to it unl
ess
x = 0).
8. Let V
i
be a collection of subspaces of V . Set U =
i
V
i
. Then if u, v U. We
have that u, v V
i
for all i because V
i

is a subspace.

Thus au V

i
for all i, so that
av U. Similarly u + v V
i
for all i, so u + v U.
9. Let U, W V be subspaces. Clearly if U W or W U, then U W is cle
arly a
subspace. Assume then that U , W and W , U. Then we can choose u U W
and w W U. Assuming that U W is a subspace we have u + w U W.
Assuming that u + w U we get w = u + w u U. Contradiction. Similarly for
u + w W. Thus U W is not a subspace.
10. Clearly U = U + U as U is closed under addition.
11. Yes and yes. Follows directly from commutativity and associativity of ve
ctor addition.
12. The zero subspace, 0, is clearly an additive identity. Assuming
that we have
inverses, then the whole space V should have an inverse U such that U + V
= 0.
Since V + U = V this is clearly impossible unless V is the trivial vector
space.
13. Let W = R
2
. Then for any two subspaces U
1
, U
2
of W we have U
1
+ W = U
2
+ W,
so the statement is clearly false in general.
14. Let W = p T(F) [ p =
n
i=0
a
i
x
i
, a
2
= a
5
= 0.
15. Let V = R
2
. Let W = (x, 0) R
2
[ x R. Set U
1
= (x, x) R
2
[ x R,
U
2
= (x, x) R
2
[ x R. Then its easy to see that U

1
+ W = U
2
+ W = R
2
,
but U
1
,= U
2
, so the statement is false.
2 Finite Dimensional Vector Spaces
1. Let u
n
= v
n
and u
i
= v
i
+v
i+1
, i = 1, . . . , n1. Now we see that v
i
=
n
j=i
u
i
. Thus
v
i
span(u
1
, . . . , u
n
), so V = span(v
1
, . . . , v
n
) span(u
1
, . . . , u
n
).
2. From the previous exercise we know that the span is V . As (v
1
, . . . , v
n
) is a linearly
independent spanning list of vectors we know that dimV = n. The claim now fol
lows
from proposition 2.16.
2
3. If (v
1
+ w, . . . , v
n
+ w) is linearly dependent, then we can write

0 = a
1
(v
1
+ w) + . . . + a
n
(v
n
+ w) =
n
i=1
a
i
v
i
+ w
n
i=1
a
i
,
where a
i
,= 0 for some i.

Now

n
i=1
a
i
,= 0 because otherwise we would get
0 =
n
i=1
a
i
v
i
+ w
n
i=1
a
i
=
n
i=1
a
i
v
i
and by the linear independence of
i
) we would get a
i
= 0, i contradicting our
assumption. Thus

(v

w =
_
n
i=1
a
i
_
1
n
i=1
a
i
v
i
,
so w span(v
1
, . . . , v
n
).
4. Yes, multiplying with a nonzero constant doesnt change the degree of a polyn
omial
and adding two polynomials either keeps the degree constant or makes
it the zero
polynomial.
5. (1, 0, . . .), (0, 1, 0, . . .), (0, 0, 1, 0, . . .), . . . is triviall
y linearly independent. Thus F

isnt nite dimensional.

6. Clearly T(F) is a subspace which is in nite dimensional.
7. Choose v
1
V . Assuming that V is in nite dimensional span(v
1
) ,= V . Thus we can
choose v
2
V span(v
1
). Now continue inductively.
8. (3, 1, 0, 0, 0), (0, 0, 7, 1, 0), (0, 0, 0, 0, 1) is clearly linearly indep
endent. Let (x
1
, x
2
, x
3
, x
4
, x
5
)
U. Then (x
1
, x
2
, x
3

, x
4
, x
5
) = (3x
2
, x
2
, 7x
4
, x
4
, x
5
) = x
2
(3, 1, 0, 0, 0)+x
4
(0, 0, 7, 1, 0)+
x
5
(0, 0, 0, 0, 1), so the vectors span U. Hence, they form a basis.
9. Choose the polynomials (1, x, x
2
x
3
, x
3
). They clearly span T
3
(F) and form a basis
by proposition 2.16.
10. Choose a basis (v
1
, . . . , v
n
). Let U
i
= span(v
i
). Now clearly V = U
1
. . . U
n
by
proposition 2.19.
11. Let dimU = n = dimV . Choose a basis (u
1
, . . . , u
n
) for U and extend it to a
basis (u
1
, . . . , u
n
, v
1
, . . . , v
k
) for V . By assumption a basis for V has length n, so

(u
1
, . . . , u
n
) spans V .
3
12. Let U = span(u
0
, . . . , u
m
). As U ,= T
m
(F) we have by the previous exercise that
dimU < dimT
m
(F) = m + 1. As (p
0
, . . . , p
m
) is a spanning list of vectors having
length m + 1 it is not linearly independent.
13. By theorem 2.18 we have 8 = dimR
8
= dimU + dimW dim(U W) = 4 + 4
dim(U W). Since U +W = R
8
, we have that dim(U W) = 0 and the claim follows
14. Assuming that U W = 0 we have by theorem 2.18 that
9 = dimR
9
= dimU + dimW dim(U W) = 5 + 5 0 = 10.
Contradiction.
15. Let U
1
= (x, 0) R
2
[ x R, U
2
= (0, x) R
2
[ x R, U
3
= (x, x) R
2
[ x
R. Then U
1
U
2
= 0, U
1
U
3
= 0 and U
2
U
3
= 0. Thus the left hand side
of the equation in the problem is 2 while the right hand side is 3.
hus we have a

counterexample.
16. Choose a basis (u
i
1
, . . . , u
i
n
i
) for each U
i
. Then the list of vectors (u
1
1
, . . . , u
m
nm
) has
length dimU
1
+ . . . + dimU
m
and clearly spans U
1
+ . . . + U
m
proving the claim.
17. By assumption the list of vectors (u
1
1
, . . . , u
m
nm
) from the proof of the previous exercise
is linearly independent. Since V = U
1
+. . .+U
n
= span(u
1
1
, . . . , u
m
nm
) the claim follows.
3 Linear Maps
1. Any v ,= 0 spans V . Thus, if w V , we have w = av for some a F. Now T(
v) = av
for some a F, so for w = bv V we have T(w) = T(bv) = bT(v) = bav = aw. Thus
T is multiplication by a scalar.
2. De ne f by e.g.
f(x, y) =
_
x, x = y
0, x ,= y
,
then clearly f satis es the condition, but is non linear.
3. To de ne a linear map its enough to de ne the image of the elements
of a basis.
Choose a basis (u
1

,
m
)
1
,
m
,
1
,
n
)
V
1
,
n

. . . , u
for U and extend it to a basis (u
. . . , u
v
. . . , v
of
. If S L(U, W).

Choose some vectors w

. . . , w

W. Now de ne T L(V, W)
by T(u
i
) = S(u
i
), i = 1, . . . , m and T(v
i
) = w
i
, i = 1, . . . , n. These relations de ne
the linear map and clearly T
|U
= S.
4. By theorem 3.4 dimV = dimnull T +dimrange T. If u , null T, then dimrang
e T >
0, but as dimrange T dimF = 1 we have dimrange T = 1 and it fol
lows that
4
dimnull T = dimV 1. Now choose a basis (u
1
, . . . , u
n
) for null T. By our assumption
(u
1
, . . . , u
n
, u) is linearly independent and has length dimV . Hence, its a basis for
V . This implies that
V = span(u
1
, . . . , u
n
, u) = span(u
1
, . . . , u
n
) span(u) = null T au [ a F.
5. By linearity 0 =
n
i=1
a
i
T(v

i
) =
n
i=1
T(a
i
v
i
) = T(
n
i=1
a
i
v
i
). By injectivity we
have
n
i=1
a
i
v
i
= 0, so that a
i
= 0 for all i. Hence (T(v
1
), . . . , T(v
n
)) is linearly
independent.
6. If n = 1 the claim is trivially true.
is true for n = k.
If S
1
, . . . , S
k+1
satisfy the assumptions, then S
1
S
k
is injective by the induction
hypothesis. Let T = S
1
S
k
. If u ,= 0, then by injectivity of S
k+1
we S
k+1
u ,= 0
and by injectivity of T we have TS
k+1
u ,= 0. Hence TS
k+1
is injective and the claim
follows.

Assume that

the claim

7. Let w W. By surjectivity of T we can

nd a vector v V
that T(v) = w.
Writing v = a
1
v
1
+ . . . + a
n
v
n
we get w = T(v) = T(a
1
v
1
+ . . . + a
n
v
n
) = a
1
T(v
1
) +
. . . + a
n
T(v
n
) proving the claim.
8. Let (u
1
, . . . , u
n
) be a basis for null T and extend it to a basis (u
1
, . . . , u
n
, v
1
, . . . , v
k
)
of V . Let U := span(v
1
, . . . , v
k
). Then by construction null T U = 0 and for an
arbitrary v V we have that v = a
1
u
1
+ . . . + a
n
u
n
+ b
1
v
n
+ . . . + b
k

such

v
k
, so that
T(v) = b
1
T(v
1
) + . . . + b
k
T(v
k
).
Hence range T = T(U).
9. Its easy to see that (5, 1, 0, 0), (0, 0, 7, 1) is a basis for null T (see
exercise 2.8). Hence
dimrange T = dimF
4
dimnull T = 4 2 = 2. Thus range T = F
2
, so that T is
surjective.
10. Its again easy to see that (3, 1, 0, 0, 0), (0, 0, 1, 1, 1) is a basis of n
ull T. Hence, we get
dimrange T = dimF
5
dimnull T = 5 2 = 3 which is impossible.
11. This follows trivially from dimV = dimnull T + dimrange T.
12. From dimV = dimnull T +dimrange T it follows trivially that if we have a
surjective
linear map T L(V, W), then dimV dimW. Assume then that dimV dimW.
Choose a basis (v
1
, . . . , v
n
) for V and a basis (w
1
, . . . , w
m
) for W. We can de ne
a linear map T L(V, W) by letting T(v
i
) = w
i
, i = 1, . . . , m, T(v
i
) = 0, i =
m + 1, . . . , n. Clearly T is surjective.
5
13. We have that dimrange T dimW. Thus we get dimnull T = dimV dimrange T
dimV dimW. Choose an arbitrary subspace U of V of such that dimU dimV
dimW. Let (u
1
, . . . , u
n
) be a basis of U and extend it to a basis (u
1
, . . . , u
n
, v
1

, . . . , v
k
)
of V . Now we know that k dimW, so let
1
, . . . , w
m
) be a basis for W. De ne
a linear map T L(V, W) by T(u
i
) = 0, i = 1, . . . , n and T(v
i
) = w
i
, i = 1, . . . , k.
Clearly null T = U.
14. Clearly if we can nd such a S, then
T is injective.
Let (v
1
, . . . , v
n
) be a basis of V . By exercise 5
1
), . . . , T(v
n
)) is linearly inde
pendent so we can extend it to a basis (T(v
1
), . . . , T(v
n
), w
1
, . . . , w
m
) of W. De ne
S L(W, V ) by S(T(v
i
)) = v
i
, i = 1, . . . , n and S(w
i
) = 0, i = 1, . . . , m. Clearly
ST = I
V
.
15. Clearly if we can nd such a S, then
1
, . . . , v
n
) be a
basis of V . By assumption (T(w
1
), . . . , T(w
n
)) spans W. By the linear dependence
lemma we can make the list of vectors (T(w
1
), . . . , T(w
n

(w

T is injective.

Assume then that

(T(v

T is surjective.

Otherwise let (v

)) a basis by removing some

vectors. Without loss of generality we can assume that the rst m vectors form
the
basis (just permute the indices). Thus T(w
1
), . . . , T(w
m
)) is a basis of W. De ne
the map S L(W, V ) by S(T(w
i
)) = w
i
, i = 1, . . . , m. Now clearly TS = I
W
.
16. We have that dimU = dimnull T + dimrange T dimnull T + dimV . Substitut
ing
dimV = dimnull S + dimrange S and dimU = dimnull ST + dimrange ST we get
dimnull ST + dimrange ST dimnull T + dimnull S + dimrange S.
Clearly dimrange ST dimrange S, so our claim follows.
17. This is nothing but pencil pushing. Just take arbitrary matrices
satisfying the re
quired dimensions and calculate each expression and the equalities easily fall o
ut.
18. Ditto.
19. Let V = (x
1
, . . . ,
n
). From proposition 3.14 we have that
/(Tv) = /(T)/(v) =
_

_
a
1,1
a
1,n
.
.
.
.
.
.
.
.
.
a
m,1
a
m,n
_

_
_

_
x
1
.

.
.
x
n
_

_ =
_

_
a
1,1
x
1
+ . . . + a
1,n
x
n
.
.
.
a
m,1
x
1
+ . . . + a
m,n
x
n
_

_
which shows that Tv = (a
1,1
x
1
+ . . . + a
1,n
x
n
, . . . , a
m,1
x
1
+ . . . + a
m,n
x
n
).
20. Clearly dimMat(n, 1, F) = n = dimV . We have that Tv = 0 if
and only if v =
0v
1
+ . . . , 0v
n
= 0, so null T = 0. Thus T is injective and hence invertible.
6
21. Let e
i
denote the n 1 matrix with a 1 in the ith row and 0 everywhere else. Let T

be the linear map. De ne a matrix A := [T(e

1
), . . . , T(e
n
)]. Now its trivial to verify
that
Ae
i
= T(e
i
).
By distributivity of matrix multiplication (exercise 17) we get for an arbitrary
v =
a
1
e
1
+ . . . + a
n
e
n
that
Av = A(a
1
e
1
+ . . . + a
n
e
n
) = a
1
Ae
1
+ . . . + a
n
Ae
n
= a
1
T(e
1
) + . . . + a
n
T(e
n
) = T(a
1
e
1
+ . . . + a
n
e
n
),
so the claim follows.
22. From theorem 3.21 we have ST invertible ST bijective S and T bijec
tive S
and T invertible.
23. By symmetry its su cient to prove this in one direction only. Thus if TS =

I. Then
ST(Su) = S(TSu) = Su for all u. As S is bijective Su goes through the whole sp
ace
V as u varies, so ST = I.
24. Clearly TS = ST if T is a scalar multiple of the identity.
For the other direction
assume that TS = ST for every linear map S L(V ).
25. Let T L(F
2
) be the operator T(a, b) = (a, 0) and S L(F
2
) the operator S(a, b) =
(0, b). Then neither one is injective and hence invertible by 3.21. However,
T + S is
the identity operator which is trivially invertible. This can be triv
ially generalized
to spaces arbitrary spaces of dimension 2.
26. Write
A :=
_

_
a
1,1
a
1,n
.
.
.
.
.
.
.
.
.
a
n,1
a
n,n
_

_, x =
_

_
x
1
.
.
.
x
n
_

_.
Then the system of equations in a) reduces to Ax = 0. Now A de nes a linear ma
p
from Mat(n, 1, F) to Mat(n, 1, F). What a) states is now that the ma
p is injective
while b) states that it is surjective. By theorem 3.21 these are equivalent.

4 Polynomials
1. Let
1
, . . . ,
m
be m distinct numbers and k
1
, . . . , k
m
> 0 such that their sum is n.
Then
m
i=1
(x
i
)
k
i
is c
ear
y such a po
ynomia
.
7
2. Let p
i
(x) =
j=i
(x z
j
), so that deg p
i
= m. Now we have that p
i
(z
j
) ,= 0 if and
on
y if i = j. Let c
i
= p
i
(z
i
). De ne
p(x) =
m+1
i=1
w
i
c
1
i
p
i
(x).
Now deg p = m and p c
ear
y p(z
i
) = w
i
.
3. We on
y need to prove uniqueness as existence is theorem 4.5.

Let s

, r
be other such
po
ynomia
s. Then we get that
0 = (s s
)p + (r r
) r
r = (s s
)p
We know that for any po
ynomia
s
deg p + deg q.
Assuming that s ,= s

p ,= 0 ,= q

we have deg pq

we have deg(r
r) < deg(s s
)p which is impossib
e.
s = s

Thus

imp
ying r = r
.
4.
p

Let be a root of p.

Thus we can write p(x) = (x )q(x).

Now we get

(x) = q(x) + (x )q
(x).
Thus is a root of
if and on
y if
statement fo

ows.
5. Let p(x) =

is a root of

q i.e.

is a mu
tip
e root.

The

n
i=0
a
i
x
i
, where a
i
R. By the fundamenta

theorem of ca
cu
us we
have a comp
ex root z. By proposition 4.10 z is a
so a root of p. Then
z and z are
roots of equa
mu
tip
icity (divide by (x z)(x z) which is rea
). Thus if we h
ave
no rea
roots we have an even number of roots counting mu
tip
icity, but the num
ber
of roots counting mu
tip
icity is deg p hence odd. Thus p has a rea
root.
5 Eigenva
ues and Eigenvectors
1. An arbitrary e
ement of U
1
+. . . +U

n
is of the form u
1
+. . . +u
n
where u
i
U
i
. Thus
we get T(u
1
+ . . . + u
n
) = T(u
1
) + . . . + T(u
n
) U
1
+ . . . + U
n
by the assumption
that T(u
i
) U
i
for a

i.
2. Let V =
i
U
i
where U
i
is invariant under T for a

i. Let v V , so that v U
i
for a


i. Now T(v) U
i
for a

i by assumption, so T(v) U
i
= V . Thus V is invariant
under T.
3. The c
ame is c
ear
y true for U = 0 or U = V . Assume that 0 ,= U ,=
V . Let
(u
1
, . . . , u
n
) be a basis for U and extend it to a basis (u
1
, . . . , u
n
, v
1
, . . . , v
m
). By
our assumption m 1. De ne a
inear operator by T(u
i

) = v
1
, i = 1, . . . , n and
T(v
i
) = v
1
, i = 1, . . . , m. Then c
ear
y U is not invariant under T.
8
4. Let u nu

(T I), so that T(u) u = 0. ST = TS gives us
0 = S(Tu u) = STu Su = TSu Su = (T I)Su,
so that Su nu

(T I).
5. C
ear
y T(1, 1) = (1, 1), so that 1 is an eigenva
ue. A
so
T(1, 1) = (1, 1) =
(1, 1) so 1 is another eigenva
ue. By coro

ary 5.9 these are a

eigenva
ues.
6. We easi
y see that T(0,0,1)=(0,0,5), so that 5 is an eigenva
ue.
A
so T(1, 0, 0) =
(0, 0, 0) so 0 is an eigenva
ue. Assume that ,= 0 and T(z
1
, z
2
, z
3
) = (2z
2
, 0, 5z
3
) =
(z
1
, z
2
, z
3
). From the assumption ,= 0 we get z
2
= 0, so the equation is of the
form (0, 0, 5z
3
) = (z
1
, 0, z
3
). Again we see that z
1
= 0 so we get the equation
(0, 0, 5z
3
) = (0, 0, z
3
). Thus 5 is the on
y non zero eigenva
ue.
7. Notice that the range of T is the subspace (x, . . . , x) F
n
[ x F and has dimension
1. Thus dimrange T = 1, so dimnu

T = n 1. Assume that T has tw
o distinct
eigenva
ues
1
,
2

and assume that

1
,= 0 ,=
2
. Let v
1
, v
2
be the corresponding
eigenvectors, so by theorem 5.6 they are
inear
y independent. Then v
1
, v
2
, nu

T,
but dimnu

T = n 1, so this is impossib
e. Hence T has at most o
ne non zero
eigenva
ue hence at most two eigenva
ues.
Because T is not injective we know that 0 is an eigenva
ue.
We a
so see that
T(1, . . . , 1) = n(1, . . . , 1), so n is another eigenva
ue. By the
previous paragraph,
these are a

eigenva
ues of T.
8. Let a F. Now we have T(a, a
2
, a
3
, . . .) = (a
2
, a
3
, . . .) = a(a, a
2
, . . .), so every a F
is an eigenva
ue.
9. Assume that T has k +2 distinct eigenva
ues
1
, . . . ,
k+2
with corresponding eiven
vectors v
1
, . . . , v
k+2
. By theorem 5.6 these eigenvectors are
inear
y independent. Now
Tv
i
=
i
v
i
and dimspan(Tv
1
, . . . , Tv
k+2
) = dimspan(
1
v
1
, . . . ,
k+2

v
k+2
) k + 1
(its k + 2 if a

i
are non zero, otherwise k + 1). This is a contradiction as
dimrange T = k and span(
1
v
1
, . . . ,
k+2
v
k+2
) range T.
10. As T = (T
1
)
1
we only need to show this in one direction. If T is invertible, then
0 is not an eigenvalue. Now let be an eigenva
ue of T and v the
corresponding
eigenvector. From Tv = v we get that T
1
v = T
1
v = v, so that T
1
v =
1
v.
11. Let be an eigenva
ue of TS and v the corresponding eigenvecto
r. Then we get
STSv = Sv = Sv, so if Sv ,= 0, then it is is an eigenvector for the eigenva
ue .
If
Sv = 0, then TSv = 0, so = 0. As Sv = 0 we know that S is not injective,
so ST
is not injective and it has eigenva
ue 0. Thus if is an eigenva
ue of TS, then
its an
eigenva
ue of ST. The other imp
ication fo

ows by symmetry.
9
12. Let (v
1
, . . . , v
n
) be a basis of V . By assumption (Tv
1
, . . . , Tv
n
) = (
1
v, . . . ,
n
v
n
).
We need to show that
i
=
j

for a

v
i

i, j.

Choose i ,= j,

+ v
j
is an eigenvector, so T(v
i
+ v
j
) =
i
v
i
+
j
v
j
= (v
i
+ v
j
) = v
i
+ v
j
. This
means that (
i
)v
i
+ (
j
)v
j
= 0. Because (v
i
, v
j
) is
inear
y independent we
get that
i
= 0 =
j
i.e.
i
=
j
.
13. Let v
1
V and extend it to a basis (v
1
, . . . , v
n
) of V . Now Tv
1
= a
1
v
1

then by our assumption

+ . . . + a
n
v
n
.
Let U
i
be the subspace generated by the vectors v
j
, j ,= i. By our assumption each
U
i
is an invariant subspace. Let Tv
1
= a
1
v
1
+ . . . + a
n
v
n
. Now v
1
U
i
for i > 1. So

et j > 1, then Tv U
j
imp
es a
j
= 0. Thus Tv
1
= a
1
v
1
. We see that v
1
was an
eigenvector. The resu
t now fo

ows from the previous exercise.
14. C
ear
y (STS
1
)
n
= ST
n
S
1
, so
n
i=0
a
i
(STS
1
)
i
=

n
i=0
a
i
ST
i
S
1
= S
_
n
i=0
a
i
T
i
_
S
1
.
15.
or.

Let
Let

be an eigenva
ue of
p(x) =

T and v V

a corresponding eigenvect

n
i=0
a
i
x
i
. Then we have
p(T)v =
n
i=0
a
i
T
i
v =
n
i=0
a
i

i
v = p()v.
Thus p() is an eigenva
ue of p(T). Then
et a be an eigenva
ue of p(T) and
v V
the corresponding eigenvector. Let q(x) = p(x) a. By the fundamenta
theorem
of
a
gebra we can write q(x) = c
n
i=1
(x
i
). Now q(T)v = 0 and q(T) = c

n
i=1
(T

i
I). As q(T) is non injective we have that T
i
I is non injective for some i. Hence

i
is an eigenva
ue of T. Thus we get 0 = q(
i
) = p(
i
) a, so that a = p(
i
).
16. Let T L(R
2
) be the map T(x, y) = (y, x). On page 78 it was shown that it has
no eigenva
ue. However, T
2
(x, y) = T(y, x) = (x, y), so 1 it has an eigenva
ue.
17. By theorem 5.13 T has an upper triangu
ar matrix with respect to some basi
s (v
1
, . . . , v
n
).
The c
aim now fo

ows from proposition 5.12.
18. Let T L(F
2
) be the operator T(a, b) = (b, a) with respect to the standard basis
.
Now T
2
= I, so T is c
ear
y invertib
e. However, T has the matrix
_
0 1
1 0
_
.
10
19. Take the operator T in exercise 7. It is c
ear
y not invertib
e, but h
as the matrix
_

_
1
1
.
.
.
.
.
.
.
.
.
1
1

_.
20. By theorem 5.6 we can choose basis (v
1
, . . . , v
n
) for V where v
i
is an eigenvector of
T corresponding to the eigenva
ue
i
. Let be the eigenva
ue of S corresponding to
v
i
. Then we get
STv
i
= S
i
v
i
=
i
Sv
i
=
i
v
i
=
i
v
i
= Tv
i
= Tv
i
= TSv
i
,
so ST and TS agree on a basis of V . Hence they are equa
.
21. C
ear
y 0 is an eigenva
ue and the corresponding eigenvectors are nu

T
= W 0.
Now assume ,= 0 is an eigenva
ue and v = u + w is a corresponding e
igenvector.
Then P
U,W
(u + w) = u = u + w (1 )u = w. Thus w U, so w = 0, but
,= 0, so we have w = 0 which imp
ies (1 )u = 0. Because v is an eigenvector we
have v = u + w = u ,= 0, so that 1 = 0. Hence = 1. As we can choose free

y
our u U, so that its non zero we see that the eigenvectors corresponding to 1 ar
e
u = U 0.
22. As dimV = dimnu

P + dimrange P, its c
ear
y su cient to prove t
hat nu

P
range P = 0. Let v nu

P range P. As v range P, we can nd a u V such
that Pu = v. Thus we have that v = Pu = P
2

u = Pv, but v nu

P, so Pv = 0.

Thus v = 0.
23. Let T(a, b, c, d) = (b, a, d, c), then T is c
ear
y injective, so 0 is no
t an eigenva
ue.
If ,= 0 is an eigenva
ue and (a, b, c, d) = (b, a, d, c). We c
ear
y
see that a ,=
0 b ,= 0 and simi
ar
y with c, d. By symmetry we can assume that
a ,= 0 (an
eigenvector is non zero). Then we have a = b and b = a. Substituting w
e get

2
b = b (
2
+ 1)b = 0. As b ,= 0 we have
2
+ 1 = 0, but this equation has no
so
utions in R. Hence T has no eigenva
ue.
24. If U is an invariant subspace of odd degree, then by theorem 5.26 T
|U
has an eigenva
ue
with eigenvector v. Then is an eigenva
ue of T with eigenvector v
against our
assumption. Thus T has no subspace of odd degree.
6 Inner Product Spaces
1. From the
aw of cosines we get |x y|
2
= |x|
2
+ |y|
2
2|x||y| cos . Solving we
get
|x||y| cos =
|x|
2
+|y|
2
|x y|
2
2
.
11
Now let x = (x
1
, x2), y = (y
1
, y
2
). A straight calculation shows that
|x|
2
+|y|
2
|xy|
2
= x
2
1
+x

2
2
+y
2
1
+y
2
2
(x
1
y
1
)
2
(x
2
y
2
)
2
= 2(x
1
y
1
+x
2
y
2
),
so that
|x||y| cos =
|x|
2
+|y|
2
|x y|
2
2
=
2(x
1
y
1
+ x
2
y
2
)
2
= x, y) .
2. If u, v) = 0, then u, av) = 0, so by the Pythagorean theorem |u| |u| + |av| =
|u+av|. Then assume that |u| |u+av| for all a F. We have |u|
2
|u+av|
2

u, u) u + av, u + av). Thus we get

u, u) u, u) +u, av) +av, u) +av, av) ,
so that
2Re a u, v) [a[

2
|v|
2
.
Choose a = t u, v) with t > 0, so that
2t u, v)
2
t
2
u, v)
2
|v|
2
2 u, v)
2
t u, v)
2
|v|
2
.
If v = 0, then clearly u, v) = 0. If not choose t = 1/|v|
2
, so that we get
2 u, v)
2
u, v)
2
.
Thus u, v) = 0.
3. Let a = (a
1
,

2a
2
, . . . ,

na
n
) R
n
and b = (b
1
, b
2
/

2, . . . , b
n
/

n) R
n
. This
euality is then simply a, b)
2
|a|
2
|b|
2

which follows directly from the Cauchy

Schwarz ineuality.
4. From the parallelogram euality we have |u + v|
2
+ |u v|
2
= 2(|u|
2
+ |v|
2
).
Solving for |v| we get |v| =

17.
5. Set e.g. u = (1, 0), v = (0, 1). Then |u| = 1, |v| = 1, |u + v| = 2
, |u v| = 2.
Assuming that the norm is induced by an inner product, we would ha
ve by the
parallelogram ineuality
8 = 2
2
+ 2
2
= 2(1
2
+ 1
2
) = 4,
which is clearly false.
6. Just use |u| = u, u) and simplify.
7. See previous exercise.
12
8. This exercise is a lot trickier than it might seem. Ill prove it for R, th
e proof for C
is almost identical except for the calculations that are longer and more tedious
. All
norms on a nite dimensional real vector space are euivalent. This gives us
lim
n
|r
n
x + y| = |x + y|
when r
n
. This probab
y doesnt make any sense, so check a book on topo
ogy
or functiona
ana
ysis or just assume the resu
t.
De ne u, v) by
u, v) =
|u + v|
2
|u v|
2
4
.
Trivia

y we have |u|
2
= u, u), so positivity de niteness fo

ows. Now
4(u + v, w) u, w) v, w)) = |u + v + w|
2
|u + v w|

2
(|u + w|
2
|u w|
2
)
(|v + w|
2
|v w|
2
)
= |u + v + w|
2
|u + v w|
2
+(|u w|
2
+|v w|
2
)
(|u + w|
2
+|v + w|
2
)
We can app
y the para

e
ogram equa
ity to the two parenthesis getting
4(u + v, w) u, w) v, w)) = |u + v + w|
2
|u + v w|
2
+
1
2
(|u + w 2w|
2
+[u v[
2
)

1
2
(|u + v + 2w|
2
+|u v|
2
)
= |u + v + w|
2
|u + v w|
2
+
1
2
|u + w 2w|
2

1
2
|u + v + 2w|
2

= (|u + v + w|
2
+|w|
2
) +
1
2
|u + v 2w|
2
(|u + v w|
2
+|w|
2
)
1
2
|u + v + 2w|
2
App
ying the para

e
ogram equa
ity to the two parenthesis we and simp

ifying we
are
eft with 0. Hence we get additivity in the
rst s
ot. Its e
asy to see that
u, v) = u, v), so we get nu, v) = nu, v) for a

n Z. Now we get
u, v) = nu/n, v) = nu/n, v) ,
13
so that u/n, v) = 1/nu, v). Thus we have nu/m, v) = n/mu, v). It fo

ows that
we have homogeneity in the rst s
ot when the sca
ar is rationa
. Now
et R and
choose a sequence (r
n
) of rationa
numbers such that r
n
. This gives us
u, v) =
im
n
r
n
u, v) =
im
n
r
n
u, v)
=
im
n
1
4
(|r
n
u + v|
2
|r
n
u v|
2
)
=
1
4
(|u + v|
2
|u v|

2
)
= u, v)
Thus we have homogeneity in the rst s
ot. We trivia

y a
so have symmetry, so w
e
have an inner product. The proof for F = C can be done by de ning the inner prod
uct
from the comp
ex po
arizing identity. And by using the identities
u, v)
0
=
1
4
(|u + v|
2
|u v|
2
) u, v) = u, v)
0
+u, iv)
0
i.
and using the properties just proved for , )
0
.
9. This is rea

y an exercise in ca
cu
us. We have integra
s of the types
sinnx, sinmx) =
_

sinnxsinmxdx
cos nx, cos mx) =
_

cos nxcos mxdx

sinnx, cos mx) =
_

sinnxcos mxdx
and they can be evaluated using the trigonometric identities
sinnxsinmx =
cos((n m)x) cos((n + m)x)
2
cos nxcos mx =
cos((n m)x) + cos((n + m)x)
2
cos nxsinmx =
sin((n m)x) sin((n + m)x)
2
.
10. e
1
= 1, then e
2
=
xx,11
xx,11
. Where

x, 1) =
_
1
0
x = 1/2,
14
so that
|x 1/2| = x 1/2, x 1/2) =

_
1
0
(x 1/2)
2
dx =
_
1/12,
which gives e
2
=

12(x 1/2) =

3(2x 1). Then to continue encil ushing we

have

x
2
, 1
_
= 1/3,
_
x
2
,

3(2x 1)
_
=

3/6,
so that x
2

x
2
, 1
_
1

x
2
,

3(2x 1)
_
3(2x 1) = x
2
1/3 1/2(2x 1). Now

|x
2
1/3 1/2(2x 1)| = 1/(6

5)
giving e
3
=

5(6x
2
6x+1). The orthonormal basis is thus (1,

3(2x1),

5(6x
2

6x + 1).
11. Let (v
1
, . . . , v
n
) be a linearly deendent list. We can assume that (v
1
, . . . , v
n1
) is
linearly indeendent and v
n
san(v
1
, . . . , v
n
). Let P denote the rojection on the
subsace sanned by (v
1
, . . . , v
n1
). Then calculating e
n
with the Gram Schmidt
algorithm gives us
e
n
=
v
n
P(v
n
)
|v
n
Pv
n
|
,
but v
n
= Pv

n
n

by our assumtion,

so e

= 0. Thus the resulting list will simly

contain zero vectors. Its easy to see that we can extend the algorithm to work
for
linearly deendent lists by tossing away resulting zero vectors.
12. Let (e
1
, . . . , e
n1
) be an orthogonal list of vectors. Assume that (e
1
, . . . , e
n
) and
(e
1
, . . . , e
n1
, e
n
) are orthogonal
e

and san the same subsace.

n
= a
1
e
1
+ . . . + a
n
e
n
.

Now we have e

n
, e
i
) = 0 for all i < n, so that a
i
= 0 for
i < n. Thus we have e
n
= a
n
e
n
and from |e
n
| = 1 we have a
n
= 1.
Let (e
1
, . . . , e
n

Then we can write

) be the orthonormal base roduced from (v

1
, . . . , v
n
) by Gram Schmidt.
Then if (e
1
, . . . , e
n
) satis es the hyothesis from the roblem we have by the revious
aragrah that e
i
= e
i
. Thus we have 2
n
ossible such orthonormal lists.
13. Extend (e
1
, . . . , e
m
) to a basis (e
1
, . . . , e
n
) of V . By theorem 6.17
v = v, e
1
) e
1
+ . . . +v, e
n
) e
n
,
so that
|v| = | v, e
1
) e
1
+ . . . +v, e
m
) e
n
|
= | v, e
1
) e
1
| + . . . +| v, e
m
) e
n
|
= [ v, e
1
) [ + . . . +[ v, e

n
) [.
Thus we have |v| = [ v, e
1
) [ + . . . +[ v, e
m
) [ if and only if v, e
i
) = 0 for i > m i.e.
v san(e
1
, . . . , e
m
).
15
14. Its easy to see that the di erentiation oerator has a uer triangular matri
x in the
orthonormal basis calculated in exercise 10.
15. This follows directly from V = U U

.
16. This follows directly from the revious exercise.
17. From exercise 5.21 we have that V = range P null P. Let U := range P
, then by
assumtion null P U

. From exercise 15, we have dimU = dimV dimU =

dimnull P, so that null P = U

. An arbitrary v V can be written as v = Pv +(v

Pv). From P
2
= P we have that P(v Pv) = 0, so v Pv null P = U

. Hence
the decomposition v = Pv + (v Pv) is the unique decomposition in U U

. Now
Pv = P(Pv +(v Pv)) = Pv, so that P is the identity on U. By de nition P = P
U
.
18. Let u range P, then u = Pv for some v V hence Pu = P
2
v = Pv = u, so P is
the identity on range P. Let w null P. Then for a F
|u|
2
= |P(u + aw)|
2
|u + aw|
2
.
By exercise 2 we have u, w) = 0. Thus null P (range P)

and from the dimension

equality null P = (range P)

. Hence the claim follows.

19. If TP
U

= P
U
TP
U
clearly U is invariant. Now assume that U is invariant.
u U we have P
U
TP
U
u = P
U
Tu = Tu = TP
U
u.
20. Let u U, then we have Tu = TP
U
u = P
U
Tu U. Thus U is invariant. Then let
w U

. Now we can write Tw = u+u

Then for

where u U and u
U

. Now P
U
Tw = u,
but u = P
U
Tw = TP
U
w = T0 = 0, so Tw U

. Thus U

is also invariant.
21. First we need to nd an orthonormal basis for U.
e
1
=
(1/

2, 1/

2, 0, 0), e
2
= (0, 0, 1/

5, 2/

5). Let U = span(e

1
, e
2
). Then we have
u = P
U

With Gram Schmidt we get

(1, 2, 3, 4) = (1, 2, 3, 4), e

1
) e
1
+(1, 2, 3, 4), e
2
) e
2
= (3/2, 3/2, 11/5, 22/5).
22. If p(0) = 0 and p
(0) = 0, then p(x) = ax
2
+ bx
3
. Thus we want to
nd the
projection of 2 + 3x to the subspace U := span(x
2
, x
3
). With Gram Schmidt we get
the orthonormal basis (

3x
2
,
_
420/11(x
3

1
2
x
2
)). We get
P
U
(2+3x) =
_
2 + 3x,

3x
2
_

3x
2
+
_
2 + 3x,
_
420/11(x
3

1
2
x
2
)
_

_
420/11(x
3

1
2
x
2
).
Here
_
2 + 3x,

3x
2
_
=
17
12

3,
_
2 + 3x,
_
420/11(x
3

1
2
x
2
)
_
=
47
660

1155,
so that
P
U
(2 + 3x) =
71
22
x
2
+
329
22
x
3
.
16
23. Theres is nothing special with this exercise, compared to the pre
vious two, except
that it takes ages to calculate.
24. We see that the map T : T
2
(R) R de ned by p p(1/2) is linear. From exercise
10 we get that (1,


3(2x 1),

5(6x
2
6x + 1) is an orthonormal basis for T
2
(R).
From theorem 6.45 we see that
q(x) = 1 +

3(2 1/2 1)

3(2x 1) +

5(6 (1/2)
2
6 1/2 + 1)

5(6x
2
6x + 1)
= 3/2 + 15x 15x
2
.
25. The map T : T
2
(R) R de ned by p
_
1
0
p(x) cos(x)dx is clearly linear. Again
we have the orthonormal basis (1,

3(2x 1),

5(6x
2
6x + 1), so that
q(x) =
_
1
0
cos(x)dx +
_
1
0

3(2x 1) cos(x)dx

3(2x 1)
+
_
1
0

5(6x
2
6x + 1) cos(x)dx

5(6x
2
6x + 1)
= 0
4

3(2x 1) + 0 =
12

2
(2x 1).
26. Choose an orthonormal basis (e
1
, . . . , e
n
) for V and let F have the usual basis.
by roosition 6.47 we have
/(T, (e
1
, . . . , e
n
)) = [e
1
, v)
e
n
, v)] ,
so that
/(T
, (e
1
, . . . , e
n
)) =
_

_
e
1
, v)
.
.
.
e
n
, v)
_

_
and nally T
a = (ae
1
, v), . . . , ae
n
, v)).

Then

27. with the usual basis for F

n
we get
/(T) =
_

_
0 1 0
.
.
.
.
.
.
.
.
. 1
0 0
_

_
so by roosition 6.47
/(T
) =
_

_
0 0
1
.
.
.
.
.
.
.
.
.
0 1 0
_

_
.
17
Clearly T
(z

1
, . . . , z
n
) = (z
2
, . . . , z
n
, 0).
28. By additivity and conjugate homogeneity we have (T I)
= T
I. From exercise
31 we see that T
I is non injective if and on
y if T I is non injective.
is an eigenva
ue of T

That is

if and on
y if is an eigenva
ue of T.

29. As (U

= U and (T
)
= T its enough to show on
y one of the imp
ications.
U be invariant under T and choose w U

. Now we can write T

w
u

,
0

Let

= u + v, where
U and v U
so that
= Tu, w) = u, T

w) = u, u + v) = u, u) = |u|
2
.
Thus u = 0 which comp
etes the proof.
30. As (T
)
= T its su cient to prove the rst part.
by exercise 31 we have
dimV = dimrange T = dimrange T

Assume that T is injective, then

,
but range T is a subspace of V , so that range T = V .
31. From proposition 6.46 and exercise 15 we get
dimnu

T = dim(range T)

= dimW dimrange T
= dimnu

T + dimW dimV.
For the second part we have
dimrangeT

dimW dimnu

T

= dimV dimnu

T
= dimrange T,
where the second equa
ity fo

ows from the rst part.
32. Let T be the operator induced by A. The co
umns are the images of the b
asis vectors
under T, so the generate range T. Hence the dimension of the span
of the co
umn
vectors equa

dimrange T. By proposition 6.47 the span of the row v
ectors equa
s
range T
.

By the previous exercise dimrange T = dimrange T

, so the c
aim fo

ows.

7 Operators on Inner Product Spaces
1. For the rst part,
et p(x) = x and q(x) = 1. Then c
ear
y 1/2 = Tp, q) ,
= p, Tq) =
p, 0) = 0. For the second part its not a contradiction since the basis is not or
thog
ona
.
18
2. Choose the standard basis for R
2
. Let T and S the the operators de ned by the
matrices
_
0 1
1 0
_
,
_
0 0
0 1
_
.
Then T and S as c
ear
y se
f adjoint, but TS has the matrix
_
0 1
0 0
_
.
so TS is not se
f adjoint.
3. If T and S are se
f adjoint, then (aT +bS)
= (aT)
+(bS)
= aT +bS for any a, b R,
so se
f adjoint operators form a subspace.
oint, but
c
ear
y (iI)
= iI, so iI is not se
f adjoint.
4. Assume that P is se
f adjoint.
= nu

P
=

The identity operator I is se
f adj

Then from proposition 6.46 we have nu

P

(range P)

. Now P is c
ear
y a projection to range P. Then assume that P i

s a
projection. Let (u
1
, . . . , u
n
) an orthogona
basis for range P and extend it to a basis
of V . Then c
ear
y P has a diagona

matrix with respect to thi
s basis, so P is
se
f adjoint.
5. Choose the operators corresponding to the matrices
A =
_
0 1
1 0
_
, B =
_
1 1
1 0
_
.
Then we easi
y see that A
A = AA
and B
B = BB
. However, A + B doesnt
de ne a norma
operator which is easy to check.
6. From proposition 7.6 we get nu

T = nu

T
. It fo

ows from proposition 6.46 that
range T = (nu

T
)

= (nu

T)

= range T
.
7.
k
.
k
.
_
T
T
k1
v, T
T
k1

C
ear
y nu

T null T

Let v null T
Then we have

v
_
=
_
T
T
k
v, T
k1
v
_
= 0,
so that T
T
k1
v = 0.
_
T
k1
v, T
k1
v
_
=
_
T

Now

T
k1
v, T
k2
v
_
= 0,
so that v null T
k1
. Thus null T
k
null T
k1
and continuing we get null T
k

null T.
Now let u range T
k
, then we can nd a v V such that u = T
k
v = T(T
k1
v), so
that range T
k
range T. From the rst part we get dimrange T
k
= dimrange T, so
range T
k

= range T.
19
8. The vectors u = (1, 2, 3) and v = (2, 5, 7) are both eigenvecto
rs corresponding to
di erent eigenvalues. A self adjoint operator is normal, so by corollar
y 7.8 if T is
normal that would imply orthogonality of u and v. Clearly u, v) ,= 0, so T
cant be
even normal much less self adjoint.
9. If T is normal, we can choose a basis of V consisting of eigenvectors o
f T. Let A be
the matrix of T corresponding to the basis. Now T is self adjoint if and o
nly if the
conjugate transpose of A equals A that is the eigenvalues of T are real.
10. For any v V we get 0 = T
9
v T
8
v = T
8
(Tv v). Thus Tv v null T
8
= null T
by the normality of T. Hence T(Tv v) = T
2
v Tv = 0, so T
2
= T. By the spectral
theorem we can choose a basis of eigenvectors for T such that T has a diagonal
matrix
with
1
, . . . ,
n
on the diagona
. Now T
2
has the diagona
matrix with
2
1
, . . . ,
2
n
on
the diagona
and from T
2
= T we must have
2
i
=
i
for a

i. Hence
i
= 0 or
i
= 1,
so the matrix of T equa
s its conjugate transpose. Hence T is se
f adjoint
.
11. By the spectra

theorem we can choose a basis of T such th
at the matrix of T
corresponding to the basis is a diagona

matrix. Let

1
, . . . ,
n
be the diagona

e
ements. Let S be the operator corresponding to the diagona

matrix
having the
e
ements

1
, . . . ,

n
on the diagona
. C
ear
y S
2
= T.
12. Let T the operator corresponding to the matrix
_
0 1
1 0
_
. Then
T
2
+ I = 0.
13. By the spectra
theorem we can choose a basis consisting of eigenvectors o
f T. Then
T has a diagona

matrix with respect to the basis. Let
1
, . . . ,
n
be the diago
na

e
ements. Let S be the operator corresponding to the diagona

matrix having
3

1
, . . . ,
3

n
on the diagona
. Then c
ear
y S
3
= T.
14. By the spectra
theorem we can choose a basis (v
1
, . . . , v
n
) consisting of eigenvectors
of T. Let v V be such that |v| = 1. If v = a
1
v
1
+ . . . + a
n
v
n

this imp
ies that

n
i=1
[a
i
[ = 1. Assume that |Tv v| < .
i
[ for a

i, thn
|Tv v| = |
n
i=1
(a
i
Tv
i
v
i
)| = |
n
i=1
(a
i

i
v
i
v
i
)|
= |
n
i=1
a
i
(
i
)v
i
| =
n
i=1
[a
i
[[
i
[|v
1
|
=
n
i=1
[a
i
[[

If [

i
[
n
i=1
[a
i
[ = ,
which is a contradiction. Thus w can nd an ignva
u
i
such that [
i
[ < .
20
15. If such an innr product xists w hav a basis consisting o
f ignvctors by th
spctra
thorm. Assum thn that (v
1
, . . . , v
n
) is basis such that th matrix of T
is a matrix consisting of ignvctors of T. D n an innr product by
v
i
, v
j
) =
_
0, i ,= j
1, i = j
.
Extnd this to an innr product by bi
inarity and homognity. Thn
w hav an
innr product and c
ar
y T is s
f adjoint as (v
1
, . . . , v
n
) is an orthonorma
basis of
U.
16. Lt T L(R
2
) b th oprator T(a, b) = (a, a + b). Thn span((1, 0)) is invarian
t
undr T, but its orthogona
comp
mnt span((0, 1)) is c
ar
y not.
17. Lt T, S b two positiv oprators. Thn thy ar s
f adjoint
and (S + T)
=
S
+T
= S +T, so S +T is s
f adjoint. A
so (S + T)v, v) = Sv, v) +Tv, v) 0.
Thus, S + T is positiv.
18. C
ar
y T
k
is s
f adjoint for vry positiv intgr. For k = 2 w hav

T
2

v, v
_
=
Tv, Tv) 0. Assum that th rsu
t is tru for a

positiv k < n and n 2. T
hn
T
n
v, v) =

T
n2
Tv, Tv
_
0,
by hypothsis and s
f adjointnss. Hnc th rsu
t fo

ows by induction.
19. C
ar
y Tv, v) > 0 for a


v V 0 imp
is that T is injctiv hnc
 invrtib
.
So assum that T is invrtib
. Sinc T is s
f adjoint w can choos an o
rthonorma

basis (v
1
, . . . , v
n
) of ignvctors such that T has a diagona
matrix with rspct to
th basis. Lt th 
mnts on th diagona
b
1
, . . . ,
n
and by assumption
i
> 0
for a

i. Lt v = a
1
v
1
+ . . . + a
n
v
n
V 0, so that
Tv, v) = [a
1
[
2

1
+ . . . +[a
n
[
2

n
> 0.
20.
_
sin cos
cos sin
_
is a suar root for vry R, which shows that it has in nit
y
many suar roots.

21. Lt S L(R

2
) b d nd by S(a, b) = (a + b, 0). Thn |S(1, 0)| = |(1, 0)| = 1 and
|S(0, 1)| = 1, but S is c
ar
y not an isomtry.
22. R
3
is an odd dimnsiona

ra

vctor spac. Hnc
S has an ignva
u
and a
corrsponding ignvctor v. Sinc |Sv| = [[|v| = |v| w hav
2
= 1. Hnc
S
2
v =
2
v = v.
21
23. Th matrics corrsponding to T and T
ar
/(T) =
_
_
0 0 1
2 0 0
0 3 0
_
_
, /(T
) =
_
_
0 2
0 0
1 0
_
_
.
Thus w
/(T

0
3
0

gt

T)
_
_
4
0
0
_
_
,
so

T
1
,
2
,
3
)

is th

0
9
0

0
0
1

that T

z
z
(4z

oprator (z

1
, 9z
2
, z
3
). Hnc

T
T(z
1
, z
2
, z
3
) =
(2z
1
, 3z
2
, z
3
). Now w just nd to prmut
isomtry S(z
1
, z
2
, z
3
) = (z
3
, z
1
, z
2
).
24. C
ar
y T

th

indics which corrsponds to th

T is positiv, so by proposition 7.26 it has a uniu positiv suar root.

Thus its su cint to show that R
2
= T
Now w hav T

T.

= (SR)
= R
S
= RS
by s
f adjointnss of R.
2
= RIR = RS
SR = T
T.

Thus R

Assum that T is invrtib
.

25.

Thn

T must b invrtib
 (po
ar dcomposition).

Hnc
S = T

T
T
1
, so S is uniu
y dtrmind.
invrtib
. Thn rang

Now assum

that

is not

T is not invrtib
 (hnc not surjctiv).

Now assum that T is not invrtib
, so that

T
T is not invrtib
. By th spctra

thorm w can choos a basis (v
1
, . . . , v
n
) of ignvctors of

T
T and w can
assum that v
1
corrsponds to th ignva
u

T
T.
by
1
=
1
,
i
=
i
,

D n
Uv

0.

Now
t T = S

Sv
Uv
Sv
i > 1.

C
ar
y U ,= S and T = U

T. Now choos
u, v V . Thn its asy to vrify that Uu, Uv) = Su, Sv) = u, v),
is an
isomtry.
26. Choos a basis of ignvctors for T such that T has a diagona
matrix w
ith ignva

us
1
, . . . ,
n

so U

on th diagona
.

Now T

T = T
2
corrsponds to th diagona
matrix
with
2
1
, . . . ,
2
n
on th diagona
and th suar root

T
T corrsponds to th diag
ona
matrix having [
1
[, . . . , [
n
[ on th diagona
. Ths ar th singu
ar va
us, so
th c
aim fo

ows.
27. Lt T L(R
2
) b th map T(a, b) = (0, a). Thn from th matrix rprsntation
w s that T
T(a, 0) = (a, 0), hnc

T
T = T
T and 1 is c
ar
y a singu
ar va
u.
Howvr, T
2
= 0, so
_
(T
2
)
T
2
= 0, so 1
2
= 1 is not a singu
ar va
u of T
2
.
28. Th composition of two bijctions is a bijction.

If T=S

T, thn sinc S is an
isomtry w hav that T is bijctiv hnc invrtib
 if and on
y if

T
T is bijctiv.
But


T
T is injctiv hnc bijctiv if and on
y if 0 is not an ignva
u. Thus T
is
bijctiv if and on
y if 0 is not a singu
ar va
u.
22
29. From th po
ar dcomposition thorm w s that dimrang T = dimrang

T
T.
Sinc

T
T is s
f adjoint w can choos a basis of ignvctors of

T
T such that
th matrix of

T
T is a diagona
matrix.

C
ar
y dimrang

T ua
s th numbr

of non zro 
mnts on th diagona
i..
s.
30. If S is an isomtry, thn c
ar
y

th numbr of non zro singu
ar va
u

S = I, so a

singu
ar va
us ar 1. Now assum

that a

singu
ar va
us ar 1. Thn

S
S is a s
f adjoint (positiv) oprator with
a

ignva
us ua
to 1. By th spctra
thorm w can choos a basis of
V such
that

S
S has a diagona

at th
matrix of

As a

ignva
us ar

matrix.

S is th idntity matrix.

Thus

S = I, so S is an isomtry.
31. Lt T

on,

this mans th

1
= S
1
_
T
1
T
1
and T
2
= S
2
_
T
2
T
2
.
1

Assum

that

and T
2

hav th sam

singu
ar va
us s
1
, . . . , s
n
. Thn w can choos bass of ignvctors of
_
T
1
T
1
and
_
T
2
T
2

such that thy hav th sam matrix.

1
, . . . , v
n
) and
(w
1
, . . . , w
n
). Lt S th oprator d nd by S(w
i
) = v
i
, thn c
ar
y S is an isomtry
and
_
T

Lt ths

bass b

(v

2
T
2
= S
1
_
T
1
T
1
S. Thus w gt T
2
= S
2
_
T
2
T
2
= S
2
S
1
_
T
1
T
1
S. Writing
S
3
= S
2
S
1
S
1
1
, which is c
ar
y an isomtry, w gt
T
2
= S
3
S
1
_
T
1
T
1
S = S
3
T
1
S.
Now
t

2
= S
1
T
1
S
2
.

Thn w

hav

2
T
2
= (S
1
T
1
S
2
)
(S
1
T
1
S
2
) = S
1
T
1
T
1
S
1
=
S
1
1
T
1
T
1
S
1
.
1
T
1

Lt

b

an ignva
u

of

and v a corrsponding ignvctor.

Bcaus S
1
is bijctiv w hav a u V such that v = S
1
u. This givs us
T

2
T
2
u = S
1
1
T
1
T
1
S
1
u = S
1
1
T
1
T
1
v = S
1
1
v = u,
so that is an ignva
u of T
2
T
2
. Lt
1
, . . .
n
) b an
vctors

(v
, v
orthonorma
basis of ign
of T

1
T
1
, thn (S
1
1
v
1
, . . . , S
1
v
n
) is an orthonorma
basis for V of ignvctors
of T
2
T
2
.
2
T
2

Hnc

and T
1
T
1

hav th sam ignva
us. Th singu
ar va
us ar

simp
y th positiv suar roots of ths ignva
us, so th c
aim fo

ows.
32. For th rst part, dnot by S th map d nd by th formu
a. St
A := /(S, (f
1
, . . . , f
n
), (
1
, . . . , 
n
)), B := /(T, (
1
, . . . , 
n
), (f
1
, . . . , f
n
)).
C
ar
y A = B and is a diagona
matrix with s
1
, . . . , s
n
on th diagona
. A

th s
i
ar positiv ra
numbrs, so that B
= B = A.
that S = T
.
For th
maps
i
to s
1
i

i
which
i
. Thus
xrcis
23
33. By

It fo

ows from proposition 6.47

scond part obsrv

f

that a
inar map S d nd by th

is mappd by T to f
TS = I.
3.23.

Th

c
aim now fo

ows from

d nition

T is positiv. Thus a

singu
ar va
us of T ar positiv. From th

singu
ar va
u dcomposition thorm w can choos bass of V such that
|Tv| = |
n
i=1

formu
a

s
i
v, 
i
) f
i
| =
n
i=1
s
i
[ v, 
i
) [
sinc a

s
i
ar positiv.
s|v| = s
n

Now c
ar
y

i=1
[ v, 
i
) [
n
i=1
s
i
[ v, 
i
) [ = |Tv|
and simi
ar
y for s|v|.
34. From th triang
 ua
ity and th prvious xrcis w gt
|(T
+ T
)v| |T
v| +|T
v| (s
+ s
)|v|.
Now
t v b th ignvctor of
_
(T
+ T
)
(T
+ T

) corrsponding to th singu
ar

va
u s. Lt T + T
= S
_
(T
+ T
)
(T
+ T
), so that
|(T
+ T
)v| = |S
_
(T
+ T
)
(T
+ T
)v| = |Ssv| = s|v| (s
+ s
)|v|,
so that s s
+ s
.
8 Oprators on Comp
x Vctor Spacs
1. T is not injctiv, so 0 is an ignva
u of T. W s that T
2
= 0, so that any v V
is a gnra
izd ignvctor corrsponding to th ignva
u 0.
2. On pag 78 its shown that i ar th ignva
us of T. W hav dimnu

(T iI)
2
+
dimnu

(T + iI)
2
= dimC
2
= 2, so that dimnu

(T iI)
2
= dimnu

(T + iI)
2
= 1.
Bcaus (1, i) is an ignvctor corrsponding to th ignva
u i and

(1, i) is an
ignvctor corrsponding to th ignva
u i. Th st of a

gnra
izd ign
vctors
ar simp
y th spans of ths corrsponding ignvctors.
3. Lt
m1
i=0
a
i
T
i
v = 0, thn 0 = T
m1
(
m1
i=0
a
i
T
i
v = a
0
T
m1
v, so that a
0
= 0. Ap
p
ying rpatd
y T
mi
for i = 2, . . . w s that a
i
= 0 for a

i. Hnc (v, . . . , T
m1
v)
is
inar
y indpndnt.
4. W s that T
3
= 0, but T
2
,= 0. Assum that S is a suar root of T,
is ni
potnt, so by coro

ary 8.8 w hav S
dimV
= S
3
= 0, so that 0 = S
4
= T
2
.
Contradiction.
5. If ST is ni
potnt, thn n N such that (ST)
n
= 0, so (TS)
n+1
= T(ST)
n
S = 0.
24

thn S

6. Assum that ,= 0 is an ignva
u of N with corrsponding ign

vctor v. Thn
N
dimV
v =
dimV
v ,= 0. This contradicts coro

ary 8.8.
7. By
mma 8.26 w can nd a basis of V such that th matrix of N is uppr tr
iangu
ar
with zros on th diagona
. Now th matrix of N
is th conjugat transpos.

Sinc

= N, th matrix of N must b zro, so th c
aim fo

ows.

8. If N
dimV 1
,= N
dimV
, thn dimnu

N
i1
< dimnu

N
i
for a

i dimV by propo
sition 8.5. By coro

ary dimnu

N
dimV
= dimV , so th c
aim c
ar
y fo

ows.
9. rang T
m
= rang T
m+1
imp
is dimnu

T
m
= dimnu

T
m+1
i.. nu

T
m
= nu

T
m+1
.
From proposition 8.5 w gt dimnu

T
m
= dimnu

T
m+k
for a


k 1. Again this
imp
is that dimrang T
m
= dimrang T
m+k
for a

k 1, so th c
aim fo

ows.
10. Lt T b th oprator d nd in xrcis 1. C
ar
y nu

T rang T ,= 0,
so th
c
aim is fa
s.
11. W hav dimV = dimnu

T
n
+dimrang T
n
, so its su cint to prov that nu

T
n

rang T
n
= 0. Lt v nu

T
n
rang T
n
. Thn w can
nd a u V such that
T
n
u = v. From 0 = Tv = T
n+1
u w s that u nu

T
n+1
= nu

T
n
which imp
is
that v = T
n
u = 0.
12. From thorm 8.23 w hav V = nu

T
dimV
, so that T
dimV
= 0. Thn
t T L(R
3
)
b th oprator T(a, b, c) = (b, a, 0). Thn c
ar
y 0 is th on
y ignva
u,
but T is
not ni
potnt.
13. From nu

T
n2
,= nu

T
n1
and from proposition 8.5 w s that dimnu

T
n
n1.
Assum that T has thr di rnt ignva
us 0,
1
,
2
. Thn dimnu

(T
i
I)
n
1,
so from thorm 8.23
n dimnu

T
n
+dimnu

(T
1
I)
n
+dimnu

(T
2
I)
n
n 1 +1 +1 = n +1,
which is impossib
, so T has at most two di rnt ignva
us.
14. Lt T L(C
4

) b d nd by T(a, b, c, d) = (7a, 7b, 8c, 8d) from th matrix of T its
asy to s that th charactristic po
ynomia
is (z 7)
2
(z 8)
2
.
15. Lt d
1
b th mu
tip
icity of th ignva
u 5 and d
2
th mu
tip
icity of th ignva
u
6. Thn d
1
, d
2
1 and d
1
+ d
2
= n. It fo

ows that d
1
, d
2
n 1, so that
(z 5)
d
1
(z 6)
d
2
divids (z 5)
n1
(z 6)
n1
. By th Cay
y Hami
ton thorm
(T 5I)
d
1
(T 6I)
d
2
= 0, so that (T 5I)
n1
(T 6I)
n1
= 0.
16. If vry gnra
izd ignvctor is an ignvctor, thn by thorm 8.23 T
has a basis
of ignvctors. If thrs a gnra
izd ignvctor that is not an ignvctor
, thn
w hav an ignva
u such that dimnu

(T I)
dimV
,= dimnu

(T I). Thus if
25

1
, . . . ,
m
ar th ignva
us of T, thn

m
i=1
dimnu

(T
i
I) < dimV , so thr
dosnt xists a basis of ignvctors.
17. By
mma 8.26, choos a basis (v
1
, . . . , v
n
) such that N has an uppr triangu
ar
matrix. App
y Gram Schmidt orthogona
ization to th
1
= 0 and
assum that N
i
span(
1
, . . . , 
i
), so that
N
i+1
= N
_
v
i+1
v
i+1
, 
1
) 
1
. . . v
i+1
, 
i
) 
i
|v
i+1
v
i+1
, 
1
) 
1
. . . v
i+1
, 
i
) 
i
|
_
span(
1
, . . . , 
i+1
).

basis.

Thn N

Thus N has an uppr triangu
ar matrix in th basis (

1
, . . . , 
n
).
18. Continuing in th proof of
mma 8.30 up to j = 4 w s that a
4
= 5/128. So that

I + N = I +
1
2
N
1
8
N
2
+
1
16
N
3

5
128
N
4
in
mma 8.30 with th
19. Just rp
ac th Tay
or po
ynomia

r po
ynomia

of
3

1 + x and copy th proof of th
mma and thorm 8.32.

20. Lt p(x) =
m
i=0
a
i
x
i
b th minima

po
ynomia

0
,= 0, thn p(x) =
x

of

T.

m
i=1
a
i
x
i1
. Sinc T is invrtib
 w must hav
m
i=1
a
i
T
i1
= 0 which contradicts

If

Tay
o

th minima
ity of p. Hnc a

0
,= 0, so so
ving for I in p(T) w gt
a
1
0
(a
1
+ . . . + a
m
T
m1
)T = I.
Hnc stting (x) = a
1
0
(a
1
+ . . . + a
m
x
m1
) w hav (T) = T
1
.
21. Th oprator d nd by th matrix
_
_
0 0 1
0 0 0
0 0 0
_
_
is c
ar
y an xamp
.
22. Choos th matrix
A =
_

_
1 0 0 0
0 1 1 0
0 0 1 0
0 0 0 0
_

_
.
Its asy to s that A(AI)
2
= 0. Thus, th minima
po
ynomia
divids z(z 1)
2
.
Howvr, th who
 po
ynomia
is th on
y factor that annihi
ats A, so z(z 1)
2
is
th minima
po
ynomia
.
26
23. Lt

1
, . . . ,
m
b th ignva
us of T with mu
tip
icity d
1
, . . . , d
m
. Assum that V
has a basis of ignvctors (v
1
, . . . , v
n
). Lt p(z) =
m
i=1
(z
i
). Thn w gt
p(T)v
i
=
_
_
j=i
(T
j
I)
_
_
(T
i
I)v
i
= 0,
so that th minima
po
ynomia
divids p and hnc has no doub
 root.
Now assum that th minima

po
ynomia

has no doub
 roots. Lt th
 minima

po
ynomia

b p and
t (v
1
, . . . , v
n
) b a Jordan basis of T. Lt A b th
argst
Jordan b
ock of T. Now c
ar
y p(A) = 0. Howvr, by xrcis 29,
th minima

po
ynomia

of (A I) is z
m+1
whr m is th
ngth of th
ongst conscutiv
string of 1s that appar just abov th diagona
, so th minima

po
of A

ynomia

is (z )
m+1
. Hnc
m = 0, so that A is a 1 1 matrix and th basis vcto
r
corrsponding to A is an ignva
u. Sinc A was th
argst Jordan b
ock it
fo

ows
that a

basis vctors ar ignva
us (corrsponding to a 1 1 matrix).
24. If V is a comp
x vctor spac, thn V has a basis of ignvctors. No
w
t

1
, . . . ,
m
b th distinct ignva
us.

Now c
ar
y p =

m
i=1
(z
i
) annihi
ats T, so that th
minima
po
ynomia
bing a factor of p dosnt hav a doub
 root.
Now assum that V is a ra
vctor spac. By thorm 7.25 w can nd a basis o
f T
such that T has a b
ock diagona
matrix whr ach b
ock is a 1 1 or 2 2 matri
x.
Lt
1
, . . . ,
m
b th distinct ignva
us corrsponding to th 1 1 b
ocks. Thn
p(z) =
m
i=1
(z
i
) annihi
ats a


but th 2 2 b
ocks of th matrix. Now its
su cint to show that ach 2 2 b
ock is annihi
atd by a po
ynomia
which dosnt
hav ra
roots.
By thorm 7.25 w can choos th 2 2 b
ocks to b of th form
_
a b
b a
_
whr b > 0 and its asy to s that
_
a b
b a
_
2a
_
a b
b a
_
+ (a
2
+ b
2
)
_
1 0
0 1
_
= 0.
C
ar
y this po
ynomia
has a ngativ discriminant, so th c
aim fo

ows.
25. Lt  b th minima
po
ynomia
. Writ  = sp + r, whr dg
r < dg p, so that
0 = (T)v = s(T)p(T)v +r(T)v = r(T)v. Assuming r ,= 0, w can mu
tip
y th bot
h
sids with th invrs of th highst co cint of r yi
ding a monic po
ynomia

r
2
of dgr
ss than p such that r
2
(T)v = 0 contradicting th minima
ity of p. Hnc
r = 0 and p divids .
27
26. Its asy to s that no propr factor of z(z 1)
2
(z 3) annihi
ats th matrix
A =
_

_
3 1 1 1
0 1 1 0
0 0 1 0
0 0 0 0
_

_
,
so th its minima

po
ynomia

is z(z 1)
2
(z 3) which by d nition is a
so th
charactristic po
ynomia
.
27. Its asy to s that no propr factor of z(z 1)(z 3) annihi
ats th matrix
A =
_

_
3 1 1 1
0 1 1 0
0 0 1 0
0 0 0 0
_

_
,
but A(AI)(A3I) = 0, so th c
aim fo

ows.
28. W s that T(
i
) = 
i+1
for i < n. Hnc (
1
, T
1
, . . . , T
n1

1
) = (
1
, . . . , 
n

) is
p of dgr
ss than

inar
y indpndnt. Thus for any non zro po
ynomia

n w
hav p(
i
) ,= 0. Hnc th minima

po
ynomia

has dgr n, so that th m
inima

po
ynomia
ua
s th charactristic po
ynomia
.
Now from th matrix of T w s that T
n
(
i
) = a
0
a
1
T(
1
) . . . a
n1
T
n1
. St
p(z) = a
0
+a
1
z +. . . +a
n1
z
n1
+z
n
. Now p(T)(
1
) = 0, so by xrcis 25 p divids
th minima
po
ynomia
. Howvr, th minima
po
ynomia
is monic and has dgr

n, so p is th minima
po
ynomia
.
29. Th biggst Jordan b
ock of N is of th form
_

_
0 1
0 1
.
.
.
.
.
.
0 1
0
_

_
.
Now c
ar
y N
m+1
= 0, so th minima
po
ynomia
divids z
m+1
. Lt v
i
, . . . , v
i+m
b
th basis 
mnts corrsponding to th biggst Jordan b
ock. Thn N
m

i+m
= 
i
,=
0, so th minima
po
ynomia
is z
m+1
.
30. Assum that V cant b dcomposd into two propr subspacs. Th
n T has on
y
on ignva
u. If thr is mor than on Jordan b
ock, thn Lt (
v
1
, . . . , v
m
) b
th vctors corrsponding to a


but th
ast Jordan b
ock and (v
m+1
, . . . , v
n
) th
28
vctors corrsponding to th
ast Jordan b
ock. Thn c
ar
y V = span(v
1
, . . . , v
m
)
span(v
m+1
, . . . , v
n
). Thus, w hav on
y on Jordan b
ock and T I is ni
potnt
and has minima
po
ynomia

z
dimV
by th prvious xrcis. Hnc T has minima

po
ynomia
(z )
dimV
.
Now assum that T has minima
po
ynomia
(z )
dimV
. If V = U W,
t p
1
b th
minima
po
ynomia
of T

|U
and p
2

th minima
po
ynomia
of T

|W
. Now (p
1
p
2
)(T) =
0, so that (z )
dimV
divids p
1
p
2
. Hnc dg p
1
p
2
= dg p
1
+ dg p
2
dimV , but
dg p
1
+dg p
2
dimU +dimW = dimV . Thus dg p
1
p
2
= dimV . This mans that
p
1
(z) = (z )
dimU
and p
2
(z) = (z )
dimW
. Now if n = maxdimU, dimW w
hav (T
|U
I)
n
= (T
|W
I)
n
= 0. This mans that (T I)
n
= 0 contradicting
th fact that (z )
dimV
is th minima
po
ynomia
.
31. Rvrsing th Jordan basis simp
y rvrss th ordr of th Jordan b
ocks
and ach
b
ock nds to b rp
acd by its transpos.

9 Oprators on Ra
Vctor Spacs

1. Lt a b th va
u in th uppr
ft cornr. Thn th matrix must hav t
h form
_
a 1 a
1 a a
_
and its asy to s that
_
1
1
_
is an ignvctor corrsponding to th ignva
u 1.
2. Th charactristic po
ynomia
of th matrix is p(x) = (x a)(x d) bc. Now
th
discriminant ua
s (a + d)
2
4(ad bc) = (a d)
2
+ 4bc. Hnc th charactristic
po
ynomia
has a root if and on
y if (a d)
2
+ 4bc 0. If p(x) = (x
1
)(x
2
),
thn p(T) = 0 imp
is that ithr A
1
I or A
2
I is not invrtib
 hnc A has
an ignva
u. If p dosnt hav a root thn assuming that Av = v w gt
0 = p(A)v = p()v,
so that v = 0. Hnc A has no ignva
us. Th c
aim fo

ows.
3. S th proof of th nxt prob
m, though in this spcia
cas th proof is
trivia
.
4. First
t b an ignva
u of A. Assum that is not an ignva
u of
A
1
, . . . , A
m
,
so that A
i
I is invrtib
 for ach i. Now
t B b th matrix
_

_
(A
1
I)
1
.
.
.
0
m

(A
I)

1
_

_.
29
Its now asy to vrify that B(AI) is a diagona
matrix with a

1s on th diagona
,
hnc its invrtib
. Howvr, this is impossib
, sinc A I is not. Thus, i
s an
ignva
u of som A
i
.
Now
t b an ignva
u of A
i
. Lt T b th oprator corrsponding to A I
in L(F
n
). Lt A
i
corrspond to th co
umns j, . . . , j + k and
t (
1
, . . . , 
n
) b th
standard basis. Lt (a
1
, . . . , a
k
) b th ignvctor of A
i
corrsponding to . St v =
a
1

j
+. . . +a
k

j+k
. Thn its asy to s that T maps th spac span(
1
, . . . , 
j1
, v)
to th spac span(
1
, . . . , 
j1
). Hnc T is not injctiv, so that is an ignva
u
of A.
5. Th proof is idntica
to th argumnt usd in th so
ution of xrcis 2.
6. Lt (v
1
, . . . , v
n
) b th basis with th rspct to which T has th givn matri
x.
App
ying Gram Schmidt to th basis th matrix trivia

y has th sam form.
7. Choos a basis (v
1
, . . . , v

n
) such that T has a matrix of th form in thorm 9.10.
Now if v
j
corrsponds to th scond vctor in a pair corrsponding to a 2 2 matrix
or corrsponds to a 1 1 matrix, thn span(v
1
, . . . , v
j
) is an invariant subscap. Th
scond posibi
ity mans that v
j
corrsponds to th rst vctor in a pair corrsponding
to a 2 2 matrix, so that span(v
1
, . . . , v
j+1
) is an invariant subspac.
8. Assuming that such an oprator xistd w wou
d hav basis (v
1
, . . . , v
7
) such that
th matrix of T wou
d hav a matrix with ignpair (1, 1)
dimnu

(T
2
+ T + I)
dimV
2
= 7/2
tims on th diagona
. This wou
d contradict thorm 9.9 as 7/2 is
not a who

numbr. It fo

ows that such an oprator dosnt xist.
9. Th uation x
2
+ x + 1 = 0 has a so
ution in C. Lt b a so
ution. Thn th
oprator corrsponding to th matrix I c
ar
y is an xamp
.
10. Lt T b th oprator in L(R
2k
) corrsponding to th b
ock diagona
matrix whr
ach b
ock has charactristic po
ynomia
x
2
+x +. Then y theorem 9.9 we hve
k =
dimnull(T
2
+ T + I)
2k
2
,
so tht dimnull(T
2
+ T + I)
2k
is even. However, the miniml polynomil of T
divides (T
2
+T +I)
2k
nd hs degree less thn 2k, so tht (T

2
+T +I)
k
= 0.
It follows tht dimnull(T
2
+ T + I)
2k
= dimnull(T
2
+ T + I)
k
nd the clim
follows.
30
11. We see tht (, ) is n eigenpir nd from the nilpotency of T
2
+ T + I nd
theorem 9.9 we hve
dimnull(T
2
+ T + I)
dimV
2
= (dimV )/2
it follows tht dimV is even. For the second prt we know from the Cyley-Hm
ilton
theorem tht the miniml polynomil p(x) of T hs degree less thn dimV . Thu
s we
hve p(x) = (x
2
+ x + )
k
nd from deg p dimV we get k (dimV )/2. Hence
(T + T + I)
(dimV )/2
= 0.
12. By theorem 9.9 we cn choose  sis such tht the mtrix of T hs the fo
rm of 9.10.
Now we hve the mtrices [5] nd [7] t lest once on the digonl. Assuming t
ht
T hs n eigenpir we would hve t lest one 2 2 mtrix on the digonl. Thi
s is
impossile s the digonl hs only length 3.
13. By proposition 8.5 dimnull T
n1
n1. Lik in th prvious xrcis w know that
th matrix [0] is at
ast n1 tims on th diagona
and its impossib
 to t a 2 2
matrix in th on
y p
ac
ft.
14. W s that A satis s th po
ynomia

p(z) = (z a)(z d) bc no mattr if F
is
R or C. If F = C, thn bcaus p is monic, has dgr 2 and annihi
ats A it
fo

ows
that p is th charactristic po
ynomia
.
Assum thn that F = R. Now if A has no ignva
us, thn p is th charactr
istic
po
ynomia
by d nition. Assum thn that p(z) = (z
1
)(z
2

) which imp
is
that p(T) = (T
1
I)(T
2
I) = 0. If nithr T
1
I or T
2
I is injctiv, thn
by d nition p is th minima
po
ynomia
. Assum thn that T
2
I is invrtib
.
Thn dimnu

(T
1
I) = 2, so that T
1
I = 0. Hnc w gt c = b = 0 and
a = d =
1
. Hnc p(z) = (z
1
)
2
, so that p is th charactristic po
ynomia
by
d nition.
15. S is norma
, so thrs an orthonorma
basis of V such that S has a b
o
ck diagona

matrix with rspct to th basis and ach b
ock is a 1 1 or 2 2 matrix and th
22 b
ocks hav no ignva
u (thorm 7.25). From S
S = I w s that ach b
ock
A
i
satisfy A
i
A
i

= I, so that thy ar isomtris. Hnc a 2 2 b
ock is of th form

_
cos sin
sin cos
_
.
W s that T
2
+T +I is not injective if nd only if A
2
i
+A
i
+I is not injective
for some 22 mtrix A
i
. Hence x
2
+x+ must equl the chrcteristic polynomil
of some A
i

, so y the previous exercise its of the form

(x cos )(z cos ) + sin
2
= x
2
2xcos + cos
2
+ sin
2

so that = cos
2
+ sin
2
= 1.
31
10 Trac and Dtrminant
1. Th map T /(T, (v
1
, . . . , v
n
)) is bijctiv and satis s ST /(ST, (v
1
, . . . , v
n
)) =
/(S, (v
1
, . . . , v
n
))/(T, (v
1
, . . . , v
n
)). Hnc
ST = I /(ST, (v
1
, . . . , v
n
)) = /(S, (v
1
, . . . , v
n
))/(T, (v
1
, . . . , v
n
)) = I.
Th c
aim now fo

ows trivia

y.
2. Both matrics rprsnt an oprator in L(F
n
). Th c
aim now fo

ows from xrcis
3.23.
3. Choos a basis (v
1
, . . . , v
n
) and
t A = (a
ij
) b th matrix corrsponding to th

basis. Thn Tv
1
= a
11
v
1
+ . . . + a
n1
v
n
. Now (v
1
, 2v
2
, . . . , 2v
n
) is a
so a basis and w
hav by our assumption Tv = a
11
v
1
+ 2(a
21
v
2
+ . . . + a
n1
v
n
). W thus gt
a
21
v
2
+ . . . + a
n1
v
n
= 2(a
21
v
2
+ . . . + a
n1
v
n
),
which imp
is (a
21
v
2
+ . . . + a
n1
v
n
= 0. By
inar indpndnc w gt a
21
= . . . =
a
n1

= 0, so that Tv
1
= a
11
v. Th c
aim c
ar
y fo

ows.
4. Fo

ows dirct
y from th d nition of /(T, (u
1
, . . . , u
n
), (v
1
, . . . , v
n
)).
5. Lt (
1
, . . . , 
n
) b th standard basis for C
n
. Thn w can
nd an oprator T
L(C
n
) such that /(T, (
1
, . . . , 
n
)) = B. Now T has an uppr triangu
ar matrix
corrsponding to som basis (v
1
, . . . , v
n
) of V . Lt A = /((v
1
, . . . , v
n
), (
1
, . . . , 
n
)).
Thn
A
1
BA = /(T, (v
1
, . . . , v
n
))
which is uppr triangu
ar. C
ar
y A is an invrtib
 suar matrix.
6. Lt T b th oprator corrsponding to th matrix
_
0 1
1 0
_
,
_
0 1
1 0
_

2
=
_
1 0
0 1
_
.
By thorm 10.11 w hav trac(T
2
) = 2 < 0.
7. Lt (v
1
, . . . , v
n
) b a basis of ignvctors of T and
1
, . . . ,
n
b th corrsponding
ignva
us. Thn T has th matrix
_

1
0
.
.
.
0
n
_

_.
C
ar
y trac(T
2
) =
2
1
+ . . . +
2
n
0 by thorm 10.11.
32
8. Extnd v/|v| to an orthonorma
basis (v/|v|, 
1
, . . . , 
n
) of V and
t A = /(T, (v/|v|, 
1
, . . . , 
n
)).
Lt a, a
1
, . . . , a
n
dnot th diagona

mnts of A. Thn w hav
a
i

= T
i
, 
i
) = 
i
, v) w, 
i
) = 0, 
i
) = 0,
so that
trac(T) = a = Tv/|v|, v/|v|) = v/|v|, v) w, v/|v|) = w, v) .
9. From xrcis 5.21 w hav V = nu

P rang P. Now
t v rang P. Thn
v = Pu
for som u V . Hnc Pv = P
2
w = Pw = v, so that P is th idntity on rang P.
Now choos a basis (v
1
, . . . , v
m
) for rang P and xtnd it with a basis (u
1
, . . . , u
n
)
of nu

P to gt a basis for V . Thn c
ar
y th matrix for P in
this basis consists
of a diagona

matrix with 1s on part of th diagona

corrsponding
to th vctors
(v
1
, . . . , v
m
) and 0s on th rst of th diagona
. Hnc trac P = dimrang P 0.
10. Lt (v
1
, . . . , v
n
) b som basis of V and
t A = /(T, (v
1
, . . . , v
n
)). Lt a
1
, . . . , a
n
b
th diagona

mnts of A. By proposition 6.47 T
has th matrix A
, so that th
diagona

mnts of A
ar a
1
, . . . , a
n

. By thorm 10.11 w hav

trac(T
) = a
1
+ . . . + a
n
= a
1
+ . . . + a
n
= trac(T).
11. A positiv oprator is s
f adjoint. By th spctra

thorm
w can
nd a basis
(v
1
, . . . , v
n
) of ignvctors of T. Thn A = /(T, (v
1
, . . . , v
n
)) is a diagona
matrix
and by positivity of T a

th diagona

mnts a
1
, . . . , a
n
ar positiv. Hnc a
1
+
. . . + a
n
= 0 imp
is a
1
= . . . = a
n
= 0, so that A = 0 and hnc T = 0.
12. Th trac of T is th sum of th ignva
us. Hnc 48 + 24 = 51 40 +
1, so
that = 36.
13. Choos a basis of (v
1
, . . . , v
n
) of V and
t A = /(T, (v
1
, . . . , v
n
)). Thn th matrix
of cT is cA. Lt a
1
, . . . , a
n
b th diagona

mnts of A. Thn w hav
trac(cT) = ca
1
+ . . . + ca
n
= c(a
1

+ . . . + a
n
) = ctrac(T).
14. Th xamp
 in xrcis 6 shows that this is fa
s. For anoth
r xamp
 tak S =
T = I.
15. Choos a basis (v
1
, . . . , v
n
) for V and
t A = /(T, (v
1
, . . . , v
n
)). Its su cint to
prov that A = 0. Now
t a
i,j
b th 
mnt in row i, co
umn j of A. Lt B b
th matrix with 1 in row j co
umn i and 0 
swhr. Thn its asy to s th
at in
BA th on
y non zro diagona

mnt is a
i,j
. Thus trac(BA) = a
i,j
. Lt S b th
oprator corrsponding to B. It fo

ows that a
i,j
= trac(ST) = 0, so that A = 0.
33
16. Lt T
T
i
= a
1

1
+ . . . + a
n

n
.
i

Thn w

hav

that a

= T
T
i
, 
i
) = |T
i
|
2
by
th ortogona
ity of (
1
, . . . , 
n
). C
ar
y a
i

is th
/(T

ith diagona

mnt in th matrix

T, (
1
, . . . , 
n
)), so that
trac(T
T) = |T
1
|
2
+ . . . +|T
n
|
2
.
Th scond assrtion fo

ows immdiat
y.
17. Choos a basis (v
1
, . . . , v
n
) such that T has an uppr triangu
ar matrix B = (b
ij
)
with ignva
us
1
, . . . ,
n
on th diagona
. Now th ith 
mnt on th diagona
of
B
B is
n
j=1
b
ji
b
ji
=
n
j=1
[b
ji
[
2
, whr [b
ii
[
2
= [
i
[
2
. Hnc
n

i=1
[
i
[
2

n
i=1
n
j=1
[b
ji
[
2
= trac(B
B) = trac(A
A) =
n
k=1
n
j=1
[a
jk
[
2
.
18. Positivity and d nitnss fo

ows from xrcis 16 and additivity
in th
rst s
ot
from coro

ary 10.12. Homognity in th rst s
ot is xrcis 13 and by xrcis
 10
S, T) = trac(ST
) = trac((ST
)
) = trac(TS
) = T, S).
It fo

ows that th formu
a d ns an innr product.
19. W hav 0 Tv, Tv) T
v, T
v) = (T
T TT
)v, v).
T TT
is a

Hnc

positiv

oprator (its c
ar
y s
f adjoint),

but trac(T

T TT
) = 0, so a


ignva
us ar 0.

th
Hnc

T TT
= 0 by th spctra
thorm, so T is norma
.
20. Lt dimV = n and writ A = /(T). Thn w hav
dt cT = dt cA =

ca
(1),1
ca
(n),n
= c
n
det A = c
n
det T
21. Let S = I and T = I. Then we have 0 = det(S + T) ,= det S + det T =
1 + 1 = 2.
22. We can ue the reult of the next exercie which mean that we only need t
o prove
thi for the complex cae. Let T L(C
n
) be the operator correponding to the
matrix A. Now each block A
j
on the diagonal correpond to ome bai vector
(e
i
, . . . , e
i+k
). Thee can be replaced with another et of vector, (v
i
, . . . , v
i+k
), pan
ning the ame ubpace uch that A
j
in thi new bai i upper triangular. We have
T(pan(v
i
, . . . , v
i+k
)) span(v
i
, . . . , v
i+k
), so after doing this for all blocks we can
assume that A is upper triangular. The claim follows immediately.
23. This follows trivially from the formula of trace and determinant for a mat
rix, because
the formulas only depends on the elements of the matrix.
34
24. Let dimV = n and let A = /(T), so that B = /(T) equals the conjugate tran

spose
of A i.e. b
ij
= a
ji
. Then we have
det T = det A =

a
(1),1
a
(n),n
=

a
1,(1)
a
n,(n)
=

a
1,(1)
a
n,(n)
=

b
(1),1
b
(n),n
= det B = det T
.
The rt equality on the econd line follow eaily that every term in the upper 
um
i repreented by a term in the lower um and vice vera. The econd claim fol
low
immediately from theorem 10.31.
25. Let = (x, y, z) R
3
[ x
2
+ y
2
+ z
2
< 1 and let T be the operator T(x, y, z) =
(ax, by, cz). Its easy to see that T() is the ellipsoid. Now is a circle with
radius
1. Hence
[ det T[volume() = abc
4
3
=
4
3

abc,
which is the volume of an ellisoid.
35