Beruflich Dokumente
Kultur Dokumente
Solutions
Linear Algebra
Question 1. Determinant calculations (Webassign Q1-Q4) 5 points
(b) Prove that if A is SPD, then all aii > 0 (1 i n). (2 subpts)
Note: Do not use eigenvalues or eigenvectors to answer this question, just use the defini-
tions given above.
Answer
MATH1115, Assignment 5 2
So X T AX is positive definite.
Hence X T AX is SPD.
Now consider the case where X T AX is SPD.
T T
AT = X 1 X T AXX 1
T T 1 T T
= X 1 X T AX X
T 1 T T
= X 1 X T AX X
T
= X 1 X T AXX 1
= A.
So A is symmetric.
Consider x 6= 0 Rn . As X is nonsingular we can set y = X 1 x 6=
0. So
T
xT Ax = xT X 1 X T AXX 1 x = y T X T AXy > 0.
(c) Firstly
bji = aji = aij = bij
for 1 i, j m. So B T = B.
Consider y 6= 0 Rm . Define x 6= 0 Rn by
(
yi if 1 i m
xi =
0 otherwise
Then
m
X
T
y By = yi bij yj
i,j=1
X m
= yi aij yj since bij = aij for 1 , j m
i,j=1
X n
= xi aij xj by construction of x
i,j=1
= xT Ax
>0 since A is positive definite
(b) 1pt for noting that we need to use the ith unit vector.
1pt for show how the unit vector can be used to get the result.
(a) If a 44 matrix A has det(A) = 1/2, find det(2A), det(A2 ) and det(A1 ). (2.5 subpts)
(b) Suppose CD = DC, and find the flaw in the following argument: (1.5 subpts)
Answer
(a) Essentially, right or wrong grading. 0.5pts for each correct answer, 0pt
for a wrong answer. Add additional 1pt if students justified answer (in
some vague way).
(b) 1pt for picking up det(DC) = (1)n det(DC). 0.5pt picking up on the fact
that answer depends on whether n is even or not.
We want to calculate
2
x + y2 + z2 x y z 1
6 10 0 1
62 + 102 6 10 0 1
= x2 + y 2 + z 2 13 3 0 1
132 + 32 13 3 0 1
2
1 + 32 + 122 1
1 3 12 1
3 12 1
2
4 + 22 + 122 4 2 12 1
4 2 12 1
136 10 0 1 136 6 0 1
178 3 0 1 178 13 0 1
x
+ y 154 1 12 1
154 3 12 1
164 2 12 1 164 4 12 1
136 6 10 1 136 6 10 0
178 13 3 1
+ 1 178 13 3 0
z
154 1 3 1
154 1
3 12
164 4 2 1 164 4 2 12
= 2856 x2 + y 2 + z 2 5712x + 11424y 0z 468384
which simplifies to
(x 1)2 + (y + 2)2 + z 169 = 0.
Check:
(6 1)2 + (10 + 2)2 + 02 169 = 0
(13 1)2 + (3 + 2)2 + 02 169 = 0
(1 1)2 + (3 + 2)2 + 122 169 = 0
(4 1)2 + (2 + 2)2 + 122 169 = 0
1pt for writing out correct cofactor expansion into a collection of the
determinants of 4 4 matrices
1pt for correctly writing out equation of sphere. Subtract 0.5pts if they
do not simplify the answer.
Calculus
Question 5. Proof from the Axioms 1 (WebAssign Q5) 4 points
Justify each line of the following proof by quoting either a real number axiom, or a
previous line, or one of either Theorem 1.2.2 statement (1) or Theorem 1.2.2 statement
(2) from Calculus Lecture 11 (30 March). Note that the first part of the proof follows the
lines discussed in that lecture.
Theorem 1. Suppose a, b R. Then (ab) = (a)b.
Proof. We will first aim to show that (1)x = x, for all x R.
Let x R be given. Then
(1)x + x = (1)x + 1x (by )
= x(1) + x1 (by )
= x((1) + 1) (by )
= x0 (by )
= 0. (by )
So we have that (1)x + x = 0. (?)
Therefore,
(1)x = (1)x + 0 (by )
= (1)x + (x + (x)) (by )
= ((1)x + x) + (x) (by )
= 0 + (x) (by )
= x . (by )
Because x R was arbitrary, we have that (1)x = x for all x R. ()
Finally,
(ab) = (1)(ab) (by )
= ((1)a)b (by )
= (a)b , (by )
as required.
(b) If a, b, c R then a < b and c < 0 implies that ac > bc. (4 subpts)
(a) Let I R have the property that if x, z I, y R, and x < y < z, then y I.
Prove that I is an interval in R. (5 subpts)
[Note: the definition of an interval was given in Calculus Lecture 11 (30 March).]
[Hint: the question has already given you a fixed subset I of real numbers. That subset
is not known to satisfy the definition of an interval, but it does satisfy the following
property:
(x, z I)(y R) [x < y y < z] = y I .
Based on the fact I is known to have this property, you must show that it satisfies
the definition of an interval as given in lectures.
That is, you must show that I must fit with one of the (long) list of options given in
the lecture.
Note: for each a R this list includes the possibility of the singleton set {a} = [a, a]
and the empty set = (a, a).
The following gives an outline of how your proof might go, with some important details
removed and replaced by . . . . These instances of . . . might represent many or
few missing lines. You may copy what I have written below, filling in the missing
lines as appropriate, or else come up with your own way of expressing the proof.
Firstly, note that if I is the empty set or a singleton set, then I is an interval and
there is nothing left to prove. Hence for the remainder of this proof we may assume
that I has at least two distinct elements. We consider four cases:
The fact that I is an interval in this case follows by a similar argument to case 2,
with (, r), (, r], and replaced by (r, ), [r, ], and
, respectively.
Case 4: the set I has both an upper bound and a lower bound.
...
Therefore, I = [a, b], I = [a, b), I = (a, b] or I = (a, b), and hence I is an interval.
Clearly the above cases cover all possibilities for I. Since we have shown that I is an
interval in each case, the proof is complete.
Answer
case 2, with (, r), (, r], upper bounds and sup I replaced by (r, ),
[r, ], lower bounds and inf I, respectively.
Case 4: the set I has both an upper bound and a lower bound.
Since I has an upper bound, by the completeness axiom I has a least upper
bound, sup I.
Let b = sup I.
Since I has a lower bound, by the completeness axiom I has a greatest lower
bound, inf I.
Let a = inf I.
If w > b then w / I, and if w < a then w / I, so we have that I [a, b].
Now suppose w is such that a < w < b.
Because b = sup I, there is z I such that w < z < b. (Apply the theorem
from Slide 5 of Analysis Lecture 7, with = b w.)
Similarly, because a = inf I, there is x I such that a < x < w.
Hence we have x, z I, w R, and x < w < z.
Therefore, w I.
So we have shown that I [a, b] and also that if a < w < r then w I.
Therefore, I = [a, b], I = [a, b), I = (a, b] or I = (a, b), and hence I is an
interval.
Clearly the above cases cover all possibilities for I. Since we have shown
that I is an interval in each case, the proof is complete.
Due to my hint, you will get a lot of nearly correct proofs here. Please
check to see what has been written actually makes sense. If they get Case
1 out properly, but the rest doesnt really make sense, award 2 marks.
Similarly, if they dont explain Case 1 well but do well on another case,
award at least 2 marks. However, they need to reference the completeness
axiom (axiom 14) at a correct point in their proof in order to get full
marks. For cases 3 and 4, be lenient if they dont express the lower
bound / infimum side of things well. I didnt discuss infima and lower
bounds in lectures, so they will be relying on a conceptual symmetric
understanding.
I = (, b) Q or I = (, b] Q ,
Mathematical Writing
2 pt if the assignment is neatly set out and the logic is easy to follow.
1 pt if the logical steps are not so well structured. For example, there are big jumps
in the steps or steps do not proceed in a sequential order.