0% found this document useful (0 votes)
31 views

Sets: Compact, Bounded, Closed: Open, If R 0 S.T. B (X, R) B X M, M V (X)

The document defines and discusses concepts related to sets in Rn such as open, closed, compact, bounded, and covering sets. It provides characterizations of closed, compact, and open sets. It then presents three exercises: 1) Proving that the set A consisting of the limit of a convergent sequence in Rn together with the sequence terms is compact. 2) Proving properties of the sum of a closed set and a compact set. 3) Proving properties about the distance between two sets, including cases where the infimum distance is attained by a minimum.

Uploaded by

catalin
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
31 views

Sets: Compact, Bounded, Closed: Open, If R 0 S.T. B (X, R) B X M, M V (X)

The document defines and discusses concepts related to sets in Rn such as open, closed, compact, bounded, and covering sets. It provides characterizations of closed, compact, and open sets. It then presents three exercises: 1) Proving that the set A consisting of the limit of a convergent sequence in Rn together with the sequence terms is compact. 2) Proving properties of the sum of a closed set and a compact set. 3) Proving properties about the distance between two sets, including cases where the infimum distance is attained by a minimum.

Uploaded by

catalin
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Calculus on Rn

Seminar 3 2020

Sets: compact, bounded, closed

To begin with, let us recall some main concepts.

According to the definition a set M ⊆ Rn is said to be

• open, if

∃r > 0 s.t. B(x, r) ⊆ B ⇐⇒ ∀x ∈ M, M ∈ V(x).

• closed, if
Rn \M is open.

If (xk ) ⊆ Rn is a sequence, then x0 ∈ Rn is its limit if

∀V ∈ V(x0 ) ∃k0 ∈ N s.t. ∀k ≥ k0 , xk ∈ V ⇐⇒

⇐⇒ ∀ε > 0 ∃k0 ∈ N s.t. ∀k ≥ k0 , kxk − x0 k < ε.

According to a characterization theorem

A set M ⊆ Rn is closed if and only if for all sequences (xk ) ⊆ M ,


which are convergent (so ∃x0 ∈ Rn such that limk→∞ xk = x0 ), the
limit must be in M (thus x0 ∈ M .)

Let I ⊆ N, let Ai ⊆ Rn , ∀i ∈ I be a collection of sets. Then the


set of all those sets {Ai : i ∈ I} forms a covering of M if
[
M⊆ Ai .
i∈I

According to the definition

1
A set M ⊆ Rn is said to be compact, if from every covering, we
may emphasize a finite covering, thus if {Ai : i ∈ I} is a covering,
exists J ⊆ I, J with a finite number of elements, such that
[
M⊆ Aj .
j∈J

In practice, for certain exercises, we use the following characterization for compact
sets:
The set M ⊆ Rn is compact, if each sequence (xk ) ⊆ M , has
a convergent subsequence in M (which actually means that the-
re exits also x0 ∈ M , which is the limit of that subsequence).
Mathematically, this is expressed as:

∀(xk ) ⊆ M, ∃(xkj )j≥1 , ∃x0 ∈ M s.t. lim xkj = x0 .


j→∞

Exercise 1:Let (xk ) ⊆ Rn be a sequence, and let

x = lim xk ∈ Rn
k→∞

be its limit. Prove that the set


[
A = {x} {xk : k ∈ N}

is compact.
Solution: We prove that the set A is compact, by showing that from an open cover,
we can determine a finite subcover.
Let (Ai )i∈I be an open cover of A. This actually means that each set Ai is open, and
[
A⊆ Ai .
i∈I

Let i0 ∈ I be such that x ∈ Ai0 . Since Ai0 is open, it follows that

Ai0 ∈ V(x).

Recall now that x = limk→∞ xk , thus,

∃k0 ∈ N s.d. ∀k ≥ k0 , xk ∈ A0 .

For each i ∈ 1, ..., k0 − 1, denote by

Ai a set which contains the element xi .

2
Therefore A ⊆ A0 ∪ A1 ∪ ... ∪ Ak0 , so we have found a finete subcover (Ai )i∈{0,...,k0 −1}
which is finite.
Therefore, A is compact.

Exercise 2: Given A, B ⊆ Rn

A + B = {x ∈ Rn : ∃a ∈ A, ∃b ∈ B s.t. x = a + b}.

a) Prove that if A is closed and B is compact then

A+B is closed.

b) Give examples of two closed sets C and D, for which C + D


is not closed.

Solution:
a) We will prove that A + B is closed by showing that

∀(ak ) ⊆ A + B with lim an = a =⇒ a ∈ A + B, (1)


k→∞

namely, each convergent sequence of elements from A + B, has the limit in A + B as well.
Let now (xk ) ⊆ A + B be a randomly chosen convergent sequence, thus ∃x0 ∈ Rn . We
will prove that
x0 ∈ A + B.
For a random k ∈ N, we have xk ∈ A + B which implies that

∃ak ∈ A, ∃bk ∈ B s.t. x k = ak + b k . (2)

This further implies the existence of two sequences (ak ) ⊆ A and (bk ) ⊆ B such that

lim (ak + bk ) = x0 . (3)


k→∞

Due to the fact that the set B is compact, there exists a subsequence

(bkj )j∈N

and ∃b0 ∈ B such that


lim bkj = b0 . (4)
j→∞

From (3) and (4) we get


lim (xkj − bkj ) = x0 − b0
j→∞

Using now (2) we obtain


∃ lim akj = x0 − b0 .
j→∞

3
Since A is a closed set and (akj ) is a subsequence of A which is convergent, it implies that
the limit belongs to A, namely
x0 − b0 ∈ A,
which, by simply adding b0 ∈ B, we get the desired conclusion that

x0 ∈ A + b0 ⊆ A + B.

Recall that the sequence (xk ) was chosen randomly, so the proof is complete.

b) It is quite clear, according to what we proved at a), that, in order to be able to


provide a good example, the two sets C and D cannot be one closed and the other one
compact. This is why, we look for two closed sets which are both not compact (and we
know that closed +bounded implies compact), so neither of them can be bounded.
Consider C := Z and
   
1 1 1 1
D := n + : n ∈ N = 1 + , 2 + , ..., i + ...
n 1 2 i

For each n ∈ N
1 1
= −n + n + ∈ C + D.
n n
1

Thus n ⊂ C + D. Take into account that

1
lim = 0 6∈ C + D,
n→∞ n

it follows that C + D is not closed.

Exercise 3: Given A, B ⊆ Rn , the distance between the two sets


is the nonnegative real number

d(A, B) = inf{d(a, b) : a ∈ A and b ∈ B}.

a) If A = {a} and B is closed, then

∃b ∈ B s.t. d(A, B) = d(a, b).

More precisely, the infimum becomes a minimum.


b) If A is compact and B is closed, then

∃a ∈ A and ∃b ∈ B s.t. d(A, B) = d(a, b).

More precisely, the infimum becomes a minimum.


c) Give an example of two sets for which the minimum is not
attained.

4
Solution:

a) Since the distance is an infimum, this means that

∃(bk ) ⊆ B s.t. lim d(a, bk ) = d(a, B.)


k→∞

By using the characterization of the limit, for ε = 1

∃k0 ∈ N s.t. ∀k ≥ k0 kd(a, bk ) − d(a, B)k < 1 ⇐⇒ (5)

⇐⇒ ∀k0 ∈ N − 1 ≤ d(a, bk ) − d(a, B) ≤ 1 =⇒


∀k ≥ k0 d(a, bk ) ≤ d(A, B) + a.
Denote r := d(a, B) + 1 > 0. Then we obtain

∀k ≥ k0 , bk ∈ B(a, r).

Recall that (bk ) ⊆ B, so we have that

∀k ≥ k0 bk ∈ B ∩ B(a, r).

Both B and B(a, r) are close sets, which implies that

B ∩ B(a, r) is closed.

Moreover, since B(a, r) is bounded, it is clear that

B ∩ B(a, r) is bounded.

In conclusion, the set being both closed and bounded, is compact. Hence

B ∩ B(a, r) is compact, and ∀k ≥ k0 bk ∈ B ∩ B(a, r).

Thus, there exists a convergent subsequence (bkj )j∈N ∈ B ∩ B(a, r), and there exists
b0 ∈ B ∩ B(a, r) such that
lim bkj = b0 .
j→∞

Hence
lim d(a, bkj ) = d(a, b0 ).
j→∞

Recall that, being a subsequence, we have (from the beginning)

lim d(a, bkj ) = d(a, B)


j→∞

Thus
d(a, b0 ) = d(a, B).

b) With an argument similar to the one in the proof of a), from the definition of the
infimum, we know that
1
∀k ∈ N, ∃ak ∈ A, ∃bk ∈ B s.t. d(A, B) < d(ak , bk ) < d(A, B) + .
k
5
(ak ) is a sequence of the compact set A, this means that there exists a subsequence
(akj )j∈N and a ∈ A such that
lim akj = a,
j→∞

and, due to this, there exists R > 0 such that d(akj , a) ≤ R for all j ∈ N. By using the
transitivity of the distance, we have for all j ∈ N
1
d(bkj , a) ≤ d(akj , bkj ) − d(akj , a) ≤ d(A, B) + + R ≤ d(A, B) + 1 + R.
k
Denote now r := d(A, B) + 1 + R > 0. Then

(bkj ) ⊆ B ∩ B(a, r).

Exactly like in the proof of a), B ∩ B(a, r) is a compact set (being both closed and
bounded). Hence, we can find (exists )a convergent sub sequence (bkjl )l∈N ⊂, and exists
b ∈ B ∩ B(a, r) such that
lim bkjl = b.
l→∞

Therefore
lim d(akjl , bkjl =d(a,b)=d(A,B) .
l→∞

Hence, the infimum is attained.

c) Consider
A = {(x, ex ) : x ∈ R}
and
B = {(x, 0) : x ∈ R}.
Both sets are closed, however

6 ∃a ∈ A, B ∈ B s.t. d(a, b) = 0.

Please draw the graph and notice that d(A, B) = 0, even though A ∩ B = ∅.

Exercise 4: Let f : Rn → Rm be a continuous function, and let


M ⊆ Rm be a n open set. Prove that the set

f −1 (M ) = {x ∈ Rn : f (x) ∈ M }
is an open set (in Rn ).
Solution:
We prove that f −1 (M ) is open, by using the definition, thus we prove that

∀x ∈ f −1 (M ) ∃r > 0, s.t. B(x, r) ⊆ f −1 (M ). (6)

Let x0 ∈ f −1 (M ) be a randomly chosen point. Be will determine a ball around it, which
is in f −1 (M ).

6
x0 ∈ f −1 (M ) ⇐⇒ f (x0 ) ∈ M =⇒ M ∈ V(x0 ) from M an open set ⇐⇒

⇐⇒ ∃ε > 0 s.t. B(x0 , ε) ⊆ M. (7)


Now we have an ε > 0 whose existence is assured. For this given (fixed) number, we
apply the characterization of the continuity of f at x0 . Thus,

f continuous at xO =⇒ for ε > 0 ∃δ > 0 s.t. ∀x ∈ Rn with kx − x0 k < δ,

to hold kf (x) − f (x0 )k < ε.


This meas that ∀x ∈ B(x0 , δ), it holds that f (x) ∈ B(f (x0 ), ε). By considering (7),
this furhterm imples that f (x) ∈ M. Hence

∀x ∈ B(x0 , δ), f (x) ∈ M ⇐⇒ B(x0 , δ) ⊆ f −1 (M ).

Recall that this last inclusion satisfies (6) for the randomly considered x0 . Thus the
conclusion is proved.

Exercise 5: Prove that the assertions of Exercise 4 remain true,


when we replace the word open, by closed.
Solution: We actually have to prove that for each H ⊆ Rm closed, the set

f −1 (H) = {x ∈ Rn : f (x) ∈ H}

is closed.
According to the definitions:

f −1 (H) closed ⇐⇒ Rn \f −1 (H) open ⇐⇒

∀a ∈ Rn \f −1 (H), ∃δ > 0 s.t. B(a, δ) ⊆ Rn \f −1 (H).


Let a ∈ Rn \f −1 (H) be randomly chosen. This mean that

f (a) ∈ Rm \H.

Since Rm \H is open, it follows that

∃ε > 0 s.t. B(f (a), ε) ⊆ Rm \H. (8)

For that given ε > 0, we apply the continuity of a. This means that

∃δ > 0 s.t. ∀x ∈ B(a, δ), f (x) ∈ B(f (a), ε).

This actually means, by considering (8) that

f (B(a, δ)) ⊆ Rm \H ⇐⇒ B(a, δ) ⊆ Rn \f −1 (H).

7
Since a was randomly chosen, the proof is complete.

Exercise 6: Let f : [a, b] → R be a random function, and let


Gf = {(x, f (x)) : x ∈ [a, b]}
be its graph. Prove that f is continuous if and only if Gf si a
compact subset of R2 ,
Solution:
=⇒ The necessity. We know that f is continuous on [a, b] and we want to prove
that Gf is a compact set.
In order to do that, we use the characterization that a set is compact, if and only if
each sequence of points in Gf , posses a convergent subsequence, whose limit point belongs
to Gf .
Let (xk , f (xk ))k≥1 ⊆ Gf be a random sequence. Notice that (xk ) ⊆ [a, b], and take into
account that [a, b] is a compact set. This means that (xk ) posses a convergent subsequence
(xkj )j≥1 , for which
∃x0 ∈ [a, b] such that lim xkj = x0 .
j→∞

Recall the definition of continuity which states that if f is continuous at p, then for each
given sequence (pk ) ⊂ [a, b] with limk→∞ pk = p, it holds that limk→∞ f (pk ) = f (p).
Coming back to our problem we have that

x0 ∈ [a, b], lim xkj = x0 and f is continuous at x0 .


j→∞

This implies automatically that

lim f (xkj ) = f (x0 ).


j→∞

Hence 
lim xkj , f (xkj ) = (x0 , f (x0 )) ∈ Gf .
j→∞

Thus, we have determined a convergent subsequence of (xk , f (xk )), which is convergent,
and the limit is in Gf . Due to the fact that (xk , f (xk )) was randomly chosen, the proof
is complete.

⇐= The Sufficiency
Now we know that Gf is compact, and we want to prove that f is continuous on [a, b].
We prove that f is continuous at a random point in x0 ∈ [a, b] by showing that
(according to the definition)

∀(xk ) ⊆ [a, b] with lim xk = x0 , it holds lim f (xk ) = f (x0 ).


k→∞ k→∞

Thus, let us consider x0 ∈ [a, b] randomly chosen. Assume by contradiction that there
exists a sequence (xk ) ⊂ [a, b], with limk→∞ xk = x0 for which

lim f (xk ) 6= f (x0 ). (9)


k→∞

8
This implies that ∃ε > 0 s.t. ∀k ∈ N ∃k0 ≥ k s.t. kf (xk ) − f (x0 )k ≥ ε.
In particular, this implies that there exists a sequence of natural numbers

∃k1 < k2 < .... < kj < ...

with the property that


kf (xkj ) − f (x0 )k ≥ ε, ∀j ≥ 1. (10)

Since xkj , f (xkj ) j≥1 ⊆ Gf , and since Gf is a compact set, there exists a convergent

subsequence xkj l , f (xkj l ) l≥1 ⊆ Gf , so there exists (t, f (t)) ∈ Gf such that

lim (xkj l , f (xkj l ) = (t, f (t)) ∈ Gf (11)


l→∞

Since (xkjl ) is a subsequence of (xk ), whose limit is x0 it follows that

t = x0 and f (t) = f (x0 )

Hence liml→∞ f (xkj l ) = f (x0 ). For the ε > 0 considered at the beginning, by applying
the characterization of the limit, we get that

∃l0 ∈ N such that ∀l ≥ l0 kf (xkjl ) − f (x0 ) < ε. (12)


0

From (10) and (12) it follows that

ε ≤ kf (xkjl ) − f (x0 ) < ε,


0

which is a contradiction.
The proof is complete.

You might also like