Linear Algebra Problem Set 4

Homework 4
Christopher Hillenbrand
18.700 Linear Algebra, Fall 2015
Collaborators: Diptarka Hait, David Vaccaro.
17 (5.A); 6, 15 (5.B); 3, 6, 10 (5.C); 1, 4, 7, 12 (6.A).
Problem 5.A.17. Give an example of an operator T 2 L(R4 ) such that T has no real
eigenvalues.
Solution. Let
T (x1 , x2 , x3 , x4 ) = (3x3 , 4x4 , x1 , 2x2 ).
Then
0
6 0
M(T ) = 6
4 1
0
I = 0 implies
det( I
which has no real roots as b2
T) =
4ac = 9
0
0
0
2
3
0
0
0
3
3
0
4 7
7.
0 5
0
+ 24 = 0
96 < 0. Hence T has no real eigenvalues.
Problem 5.B.6. Suppose T 2 L(V ) and U is a subspace of V invariant under T . Prove

that U is invariant under p(T ) for every polynomial p 2 P(F).
Proof. Let p 2 P(F) be a polynomial of degree m such that
p(z) =
m
X
ci z i .
i=0
Then
p(T ) = c0 I + c1 T + + cm T m .
If u 2 U , then we can prove by induction that T j u 2 U for any j 2 N0 . If j = 0, this

equation is trivially true. Now assume T j 1 u 2 U . Then
T j u = T T j 1u 2 U
because U is invariant under T . Therefore
[p(T )] (u) = c0 u + c1 T u + + cm T m u 2 U
because each cj T j u 2 U by closure. Hence U is invariant under p(T ).
1
Problem 5.B.15. Give an example of an operator whose matrix with respect to some basis
contains only nonzero numbers on the diagonal, but the operator is not invertible.
Solution. Let T 2 L (F2 ) be the operator with matrix
1 1
M(T ) =
.
1 1
Then det T = 0 so T is not invertible, but T has no zeroes on the diagonal.
Problem 5.C.3. Suppose V is finite-dimensional and T 2 L(V ). Prove that the following
are equivalent:
(a)
V = null T
range T .
(b)
V = null T + range T .
(c)
null T \ range T = {0}.
Proof. Clearly (a) () ((b) ^ (c)) by definition. We shall prove (c) () (b), thereby
proving the theorem. Assume (c) is true. We know from the dimension formula that
dim V = dim null T + dim range T.
Since null T \ range T = {0}, we also know that
dim (null T + range T ) = dim null T + dim range T.
Thus
dim V = dim (null T + range T ) .
Since null T + range T V , we must have
null T + range T = V
as desired. Now assume (b) is true. Then
dim (null T + range T ) = dim V.
There is a theorem that says
dim (null T + range T ) = dim null T + dim range T + dim (null T \ range T ) .
Combining these two equations, we obtain
dim V = dim null T + dim range T + dim (null T \ range T ) .
But dim V = dim null T + dim range T , so
dim (null T \ range T ) = 0
=) null T \ range T = {0}
proving the theorem.

2
Problem 5.C.6. Suppose V is finite-dimensional, T 2 L(V ) has dim V distinct eigenvalues,
and S 2 L(V ) has the same eigenvectors as T (not necessarily with the same eigenvalues).
Prove that ST = T S.
Proof. Let n = dim V . Then T has distinct eigenvalues 1 , . . . , n . We know that S has
eigenvalues 1 , . . . , n , not necessarily unique, but such that T v = j v implies Sv = j v. Let
v1 , . . . , vn be nonzero eigenvectors such that T vj = j v, j = 1, 2, 3, . . . , n. These are linearly
independent since the eigenvalues are distinct. Since there are dim V of them, they form a
basis for V . Thus if v = a1 v1 + + an vn , we have
ST (v) = S ( 1 v1 + + n vn )
= 1 1 v 1 + + n n v n
= T (1 v1 + + n vn )
= T S(v)
as desired.
Problem 5.C.10. Suppose that V is finite-dimensional and T 2 L(V ). Let
denote the distinct nonzero eigenvalues of T . Prove that
dim E ( 1 , T ) + + dim E (
m, T )
1, . . . ,
dim range T.
Proof. Similarly to the proof of theorem 5.38, we first show that E ( 1 , T ) + + E (

is a direct sum. Suppose
u1 + + um = 0
m, T )
where each uj is in E ( j , T ) . Because eigenvectors corresponding to distinct eigenvalues are

linearly independent, this implies that each uj equals 0. Therefore E ( 1 , T )+ +E ( m , T )
is a direct sum. Furthermore, any vector in the space formed by summing the eigenspaces
is contained within the range of T . Therefore
dim E ( 1 , T ) + + dim E (
m, T )
= dim (E ( 1 , T )
dim range T
E(
m , T ))
where the equality follows from 2.C.16.
Problem 6.A.1. Show that the function which takes ((x1 , x2 ), (y1 , y2 )) 2 R2 R2 to |x1 y1 |+
|x2 y2 | is not an inner product on R2 .
Proof. Let x = (1, 1). Then h x, xi = 2 6= hx, xi =
the first argument, it cannot be an inner product.
2. Since the function is not linear in
Problem 6.A.4. Suppose V is a real inner product space.

vi = kuk2
kvk2 for every u, v 2 V .
(a)
Show that hu + v, u
(b)
Show that if u, v 2 V have the same norm, then u + v is orthogonal to u

3
v.
(c)
Use part (b) to show that the diagonals of a rhombus are perpendicular to each
other.
Solution.
(a)
hu + v, u
vi = hu, u
= hu
vi + hv, u
v, ui + hu
vi
v, vi
= hu, ui hv, ui + hu, vi

= hu, ui hv, vi
= kuk2 kvk2 .
(b)
Suppose kuk = kvk. Then

hu + v, u
so that u + v is orthogonal to u
(c)
hv, vi
vi = kuk2
kvk2 = 0
v by definition of orthogonality.
A rhombus is a parallelogram with congruent sides. If we draw a rhombus starting

from the origin using vectors u and v of the same length, then one of the diagonals
is parallel to u + v and the other is parallel to u v. By part (b), these vectors
are orthogonal. Therefore the diagonals of a rhombus are perpendicular.
Problem 6.A.7. Suppose u, v 2 V . Prove that kau + bvk = kbu + avk for all a, b 2 R if
and only if kuk = kvk.
Proof. Suppose that kau + bvk = kbu + avk for all a, b 2 R. Set a = 1, b = 0. Then we have
kuk = kvk. We now prove the converse. Suppose that kuk = kvk. Then
hau + bv, au + bvi = ahu, au + bvi + bhv, au + bvi
= a2 hu, ui + abhu, vi + abhv, ui + b2 hv, vi
= a2 kuk2 + b2 kvk2 + 2ab Rehu, vi.
Similarly,
hbu + av, bu + avi = bhu, bu + avi + ahv, bu + avi
= b2 hu, ui + abhu, vi + abhv, ui + a2 hv, vi
= b2 kuk2 + a2 kvk2 + 2ab Rehu, vi.
But kuk = kvk, so these two expressions are equal, as desired.
Problem 6.A.12. Prove that
(x1 + + xn )2 n x21 + + x2n
for all positive integers n and all real numbers x1 , . . . , xn .
4
Proof. Let x = (x1 , . . . , xn ). First note that
2
(x1 + + xn ) =
n
X
i=1
xi (x1 + + xn )
n
X
i=1
yi
where the yi are cyclic permutations of (x1 , . . . , xn ). The Cauchy-Schwarz inequality tells us
that
yi x kyi k kxk
= kxk2
= (x1 + + xn )2 .
Therefore
2
(x1 + . . . + xn ) =
n
X
i=1
n
X
i=1
yi x
(x21 + + x2n )
= n(x21 + + x2n )
as desired.

Linear Algebra Problem Set 4

Uploaded by

Copyright:

Available Formats

Linear Algebra Problem Set 4

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Linear Algebra Problem Set 4

Uploaded by

Copyright:

Available Formats

Homework 4

96 < 0. Hence T has no real eigenvalues.

Problem 5.B.6. Suppose T 2 L(V ) and U is a subspace of V invariant under T . Prove

If u 2 U , then we can prove by induction that T j u 2 U for any j 2 N0 . If j = 0, this

null T \ range T = {0}.

proving the theorem.

Proof. Similarly to the proof of theorem 5.38, we first show that E ( 1 , T ) + + E (

where each uj is in E ( j , T ) . Because eigenvectors corresponding to distinct eigenvalues are

where the equality follows from 2.C.16.

2. Since the function is not linear in

Problem 6.A.4. Suppose V is a real inner product space.

kvk2 for every u, v 2 V .

Show that if u, v 2 V have the same norm, then u + v is orthogonal to u

= hu, ui hv, ui + hu, vi

Suppose kuk = kvk. Then

A rhombus is a parallelogram with congruent sides. If we draw a rhombus starting

You might also like