Professional Documents
Culture Documents
HU Wei
October 21, 2019
Solution 1. We are going to prove that − inf A is the least upper bound of −A.
1. First we prove that it is an upper bound. For every y ∈ −A, we have, −y ∈ A. Then,
−y ≥ inf A as inf A is a lower bound of A. Therefore, y ≤ − inf A.
2. Next, we show that it is the least. Let l < − inf A, we are going to show that l is not an
upper bound of −A. This means −l > inf A. As inf A is the greatest lower bound, then
−l is not a lower bound, which means, there is a ∈ A such that a < −l, or −a > l.
This means l is not a upper bound of −A. As l < − inf A is arbitrary, this proves least.
On the other hand, max{sup(A), sup(B)} is a upper bound of both A and B. There-
fore, it is a upper bound of A ∪ B. Since sup(A ∪ B) is the least upper bound, we
have max{sup(A), sup(B)} ≥ sup(A ∪ B). Together, we provemax{sup(A), sup(B)} =
sup(A ∪ B).
3. We have sup A · sup B = sup(A · B). Replace the addition operation in above argument
by multiplication. Note that
Here, both (sup A + sup B), 2 can be arbitrarily small since sup A + sup B < ∞. (A, B
are bounded)
1
(b) inf(A + B) = inf A + inf B.
(c) inf(A · B) = inf A · inf B, provided elements of A, B are positive.
Solution
√ 3. In previous tutorial, we have shown that there exists positive irrational number
( 2). Denote this irrational number by d.
1. An easy proof goes like: find a rational number q ∈ (a − d, b − d), then d + q ∈ (a, b).
d + q is obvious irrational.
2. For any a < b, let’s construct an irrational number that lies in (a, b). By the density
of rational number, there is r ∈ (a, b), r ∈ Q. By Archimedean property of R, there is
N ∈ N such that N1 < (b − r)/2. (2/(b − r) > N ).
(a) First we show we can find K ∈ N such that Kd < N1 . Actually, this is again implied
by Archimedean property of R. Kd < N1 ⇐⇒ Nd < K.
d 1 b−r b+r
(b) Now we have a < r + K
<r+ N
<r+ 2
< 2
< b.
(c) To finish the construction, we left to prove r + Kd is irrational.
Suppose d = (xk ),
d xk
now if r + K is rational, then there is q ∈ Q such that r + K − q → 0. This
implies, xk − (q − r)K → 0, which implies d = (xk ) = (q − r)K is a rational
number contradicting our assumption that d is irrational.
|akj − a| <
This tells limn akn = lim bn = a∗ . One stil needs to make akn into a true subsequence of {an },
namely, force kn to be increasing in n. This can be done easily by choosing a subsequence as
kn ≥ n → ∞.
Solution 7. Let bn = supk≥n ak . b > lim bn implies there is N such that bn < b for n ≥ N.
(By definition of convergence, take r < b−lim
2
bn
) Especially, bN < b. Namely,
sup ak < b,
k≥N