18.
06 PROBLEM SET #2 - SOLUTIONS
FALL 2014
1. Section 2.6, Problem 13, page 104.
Thefour conditions
are a 6=0, b 6= a, c 6= b, d 6=
c.
1000 a a a a
1 1 0 0 0 b−a b−a b−a
L= 1 1 1 0
U = 0 0 c−b c−b
1111 0 0 0 d−c
2. Section 2.6, Problem 16, page 105.
Solving Lc = b, leads to c = (4,1,1)T and then solving U x = c leads to
x = (3,0,1)T .
111
A = LU = 1 2 2
123
3. Section
2.6,Problem
19,
page 105.
110 100 110 100 100 110
A = 1 2 1 = 1 1 0 0 1 1 = 1 1 0 0 1 0 0 1 1
012 011 001 011 001 001
Assume
a,b,c 6
= 0. Then:
a a 0 100 aa0 100 a00 110
A = a a + b b = 1 1 0 0 b b = 1 1 0 0 b 0 0 1 1
0 b b+c 011 00c 011 00c 001
4. Section 2.7, Problem 15, page 117.
(a) If P sends row 1 to row 4, then P T sends row 4 to row 1.
Note: Here, P is not necessarily a row exchange matrix, it is just a permu-
tation matrix. Only when P T = P the row exchanges come in pairs with no
overlap.
0100
0 0 1 0
0 0 0 1 .
(b) Several answers are possible, one of them is: P =
1000
5. Section 2.7, Problem 17, page 117.
2x2 symmetric matrices, AT = A, many answers possible:
(a) A is not invertible:
ab 2 12
Any matrix of the form where b = ac works. For instance: .
bc 24
(b) A is invertible but cannot
be factored into LU (row exchanges needed):
0b 01
Any matrix of the form where b 6= 0 works. For instance: .
bc 11
T T
LDL but
(c) A can be factored into not into LL (because of negative D):
a ab
Any matrix of the form where a 6= 0 and c < 0 works.
ab ab2 + c
12 10 1 0 12
For instance: = .
23 2 1 0 −1 0 1
1
2 Solutions Pset2 18.06
6. Section 3.1, Problem 10, page 128.
(a) The plane of vectors (b1 ,b2 ,b3 ) with b1 = b2 .
Given a vector v = (x,x,z) on that plane and a scalar c, we see that
cv = (cx,cx,cz) lies on that plane because cx = cx. Given one more vector on
the plane w = (a,a,b), we see that v + w = (x + a,x + a,z + b) remains on that
plane because the first two components match. Thus YES, this plane is a
subspace of R3 .
(b) The plane of vectors (b1 ,b2 ,b3 ) with b1 = 1.
The zero vector (0,0,0) needs to be on the plane, for it to be a subspace of R3 .
Since this plane does not pass through the origin, it is NOT a subspace of R3 .
(c) The plane of vectors (b1 ,b2 ,b3 ) with b1 b2 b3 = 0.
Both (0,1,1) and (1,0,0) lie on this plane, but not their sum (1,1,1).
Therefore, this plane is NOT a subspace of R3 .
(d) All linear combinations of v = (1,4,0) and w = (2,2,2).
YES, this is definitely a subspace of R3 . In fact, this subspace is the plane
that contains both vectors and passes through the origin.
(e) All vectors (b1 ,b2 ,b3 ) that satisfy b1 + b2 + b3 = 0.
Actually, b1 + b2 + b3 = 0 is the equation of a plane that contains the origin.
Therefore, YES, the set of all those vectors (forming a plane) is a subspace
of R3 .
(f ) All vectors (b1 ,b2 ,b3 ) with b1 ≤ b2 ≤ b3 .
Pick a vector in that set, v = (b1 ,b2 ,b3 ) and consider −v = (−b1 ,−b2 ,−b3 )
(multiplying by the scalar c = −1). Here, −b1 ≥ −b2 ≥ −b3 , so −v does not
belong to the set. So, this set is NOT a subspace of R3 .
7. Recall that the column space consists of all linear combinations of the
columns. If the column space is a just line, then all columns need to be a
multiple of, say, the first column. We can also choose that first column to
be the given vector v = (1,2,3)T . Let us write one symmetric matrix that
satisfies the problem:
123
A=2 4 6
369
1 0 0 0 0 0
1 1 0 0 0 0
1 2 1 0 0 0
8. Pascal’s triangle 6 × 6: A =
1
3 3 1 0 0
1 4 6 4 1 0
1 5 10 10 5 1
For better visualization, we expand the matrix-vector product as a linear
combination of the columns.
3
1 0 0 0 00 1
1 1 0 0 0 0 1
1 2 1 0 0 0 1 =
A(1,1,1,1,1,1)T =
1
3 3 1 0 0
1
1 4 6 4 10 1
1 5 10 10 51 1
1 0 0 0 0 0 1
1 1 0 0 0 0 2
1
+1× 2 +1× 1 +1× 0 +1× 0 +1× 0 = 4
=1×
1 3 3 1 0 0 8
1 4 6 4 1 0 16
1 5 10 10 5 1 32
Note that the components of the last vector
are powersof 2.
10 0 0 00 1
1 1 0 0 0 0 2
1 2 1 0 0 0 4
A2 (1,1,1,1,1,1)T = A(1,2,4,8,16,32)T =
1 3 3 1 0 0 8 =
1 4 6 4 1 0 16
1 5 10 10 5 1 32
1 0 0 0 0 0 1
1 1 0 0 0 0 3
1
+ 2 × + 4 × + 8 × + 16 × + 32 × 0 = 9
2 1 0 0
=1×
1 3 3 1 0 0 27
1 4 6 4 1 0 81
1 5 10 10 5 1 243
Note that the components of the resulting vector for m = 2 are powers of 3.
The same pattern occurs
for m = 3.
10 0 0 00 1 1
1 1 0 0 0 0 3 4
3 T
1 2 1 0 0 0 9 16
A (1,1,1,1,1,1) =
=
1 3 3 1 0 0 27 64
1 4 6 4 1 0 81 256
1 5 10 10 5 1 243 1024
Based on these observations, we can guess what the result would be for any
positive integer m − 1, namely:
1
m
2
m
Am−1 (1,1,1,1,1,1)T =
m3
4
m
m5
If this is true for a given m − 1, the next step, m, Am (1,1,1,1,1,1)T is
2
obtained
nbynthe product A(1,m,m ,m3 ,m4 ,m5 )T . The n + 1-th row of
n
A is 0 1 ··· n 0··· 0 The n + 1-th component of the desired vec-
tor is the dot product between the corresponding n + 1-th row of A and
(1,m,m2 ,m3 ,m4 ,m5 )T , which gives:
4 Solutions Pset2 18.06
n
X n
mk = (m + 1)n
k
k=0
by Newton’s Binomial Formula.
9. Several possible answers,just showing one for each case as reference.
10000
0 1 0 0 0
Order 1: P = 0 0 1 0 0
0 0 0 1 0
00001
Only possible answer, P = I:
00010
0 0 1 0 0
Order 2: P = 0 1 0 0 0
1 0 0 0 0
00001
Note that P 6= I but P 2 = I.
10000
0 0 0 1 0
Order 3: P = 0 0 1 0 0
0 0 0 0 1
01000
We can check that this permutation matrix is such that P 6= I, P 2 6= I but
P 3 = I.
00001
0 0 0 1 0
Order 4: P = 0 0 1 0 0
1 0 0 0 0
01000
We can check that this permutation matrix is such that P 6= I, P 2 6= I, P 3 6= I
but P 4 = I.
00100
0 0 0 1 0
Order 5: P = 0 1 0 0 0
0 0 0 0 1
10000
We can check that this permutation matrix is such that P 6= I, P 2 6= I,
P 3 6= I, P 4 6= I but P 5 = I.
01000
0 0 1 0 0
Order 6: P = 1 0 0 0 0
0 0 0 0 1
00010
We can check that this permutation matrix is such that P 6= I, P 2 6= I,
P 3 6= I, P 4 6= I, P 5 6= I but P 6 = I.
5
You can see that the larger the order, the naive way of checking every power
is not efficient. Let us illustrate how you can speed up the determination of
the order by using disjoint cycles. The last permutation matrix above can be
represented as the permutation {1,2,3,4,5} → {2,3,1,5,4}. If we start by 1,
one step of the permutation takes us to 2, from 2 one more step takes us to
3, and finally back to 1. The cycle (1,2,3) is disjoint of the cycle (4,5). The
cycle (1,2,3) is of order 3, because we need 3 permutations to return to the
identity, and the cycle (4,5) is of order 2. Since these cycles are disjoint, the
order of the original permutation is the least common multiple of the orders
of the individual cycles. Thus, the order of the last permutation matrix is 6.
b)10 × 10 permutation matrix, order 30: we can construct this case by
choosing as many disjoint cycles as we need such that the least common
multiple of their orders is 30, say three disjoint cycles of orders 2,3, and
5. For example, let us choose the cycles (4,5), (3,10,8), and (1,2,9,6,7)
forming the permutation {1,2,3,4,5,6,7,8,9,10} → {2,9,10,5,4,7,1,3,6,8}.
The corresponding permutation matrix is:
0 1 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 0 1
0 0 0 0 1 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0
P =
0
0 0 0 0 0 1 0 0 0
1 0 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0 0
0 0 0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 1 0 0
10. See separate file for the computational part solutions.
