100% found this document useful (1 vote)
342 views

Nested Interval Theorem and Its Applications

The document summarizes several theorems related to analysis including the nested interval theorem, Bolzano-Weierstrass theorem, and intermediate value theorem. The nested interval theorem states that if intervals Jn are nested such that Jn+1 is contained in Jn for all n, then the intersection of all the intervals is non-empty. If the lengths of the intervals also approach 0, then the intersection contains a single point. The Bolzano-Weierstrass theorem proves that every bounded sequence contains a convergent subsequence. The intermediate value theorem proves that if a continuous function f has values of opposite signs at the endpoints of an interval, there exists a point in the interval where the function value is 0
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
342 views

Nested Interval Theorem and Its Applications

The document summarizes several theorems related to analysis including the nested interval theorem, Bolzano-Weierstrass theorem, and intermediate value theorem. The nested interval theorem states that if intervals Jn are nested such that Jn+1 is contained in Jn for all n, then the intersection of all the intervals is non-empty. If the lengths of the intervals also approach 0, then the intersection contains a single point. The Bolzano-Weierstrass theorem proves that every bounded sequence contains a convergent subsequence. The intermediate value theorem proves that if a continuous function f has values of opposite signs at the endpoints of an interval, there exists a point in the interval where the function value is 0
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

Nested Interval Theorem and Its Applications

S. Kumaresan
School of Math. and Stat.
University of Hyderabad
Hyderabad 500046
kumaresa@gmail.com

If J = [a, b] is an interval, we let `(J) := b − a, the length of the interval. We shall


repeatedly use the following two trivial observations.

Lemma 1. If x, y ∈ [a, b], then |x − y| ≤ b − a.

Proof. Without loss of generality assume that x ≤ y and hence |x − y| = y−x. Since x ∈ [a, b],
we have a ≤ x and hence −x ≤ −a. Also, y ∈ [a, b] and therefore y ≤ b. It follows that

|x − y| = y − x ≤ b − a.

Lemma 2. Let [a, b] ⊂ [c, d]. Then c ≤ a ≤ b ≤ d.

Theorem 3. [Nested Interval Theorem] Let Jn := [an , bn ] be intervals in R such that


Jn+1 ⊆ Jn for all n ∈ N. Then ∩Jn 6= ∅.
If, furthermore, we assume that lim `(Jn ) → 0, then ∩n Jn contains precisely one point.
n→∞

Proof. Note that the hypothesis means that [an+1 , bn+1 ] ⊂ [an , bn ] for all n. In particular,
an ≤ an+1 and bn+1 ≤ bn for all n ∈ N. Observe also that if s ≥ r, then Js ⊂ Jr :

Js ⊆ Js−1 ⊆ · · · ⊆ Jr+1 ⊆ Jr .

Hence, in view of Lemma 2, we have

ar ≤ as ≤ bs ≤ br , in particular as ≤ br . (1)

Let E be the set of left endpoints of Jn . Thus, E := {a ∈ R : a = an for some n}. E is


nonempty.

an a1 a2 b2 b1 bn

Figure 1:

1
We claim that bk is an upper bound for E for each k ∈ N, i.e., an ≤ bk for all n and k.
If k ≤ n then [an , bn ] ⊆ [ak , bk ] and hence an ≤ bn ≤ bk . (Draw pictures!) If k > n then
an ≤ ak ≤ bk . (Use Eq. 1.) Thus the claim is proved. By the LUB axiom there exists c ∈ R
such that c = sup E. We claim that c ∈ Jn for all n. Since c is an upper bound for E we have
an ≤ c for all n. Since each bn is an upper bound for E and c is the least upper bound for E
we see that c ≤ bn . Thus we conclude that an ≤ c ≤ bn or c ∈ Jn for all n. Hence c ∈ ∩Jn .
Let us now assume further that `(Jn ) → 0 as n → ∞. By the first part we know that
∩Jn 6= ∅. Let x, y ∈ ∩n Jn . We claim that x = y. For, since x, y ∈ Jn for all n, we have

|x − y| ≤ `(Jn ) → 0, as n → ∞.

Thus we conclude that |x − y| = 0 and hence x = y.

Theorem 4. The set R of real numbers is uncountable.

Proof. We shall show that the set [0, 1] is uncountable. The theorem follows from this.
Assume that [0, 1] is countable. That is, there exists a map f : N → [0, 1] which is onto.
Let xn := f (n) ∈ [0, 1]. To arrive at a contradiction, we shall employ the Nested interval
theorem. Let us subdivide the interval [0, 1] into three closed subintervals of equal length.
Then there exists at least one subinterval which does not contain f (1) = x1 . (Draw a picture.
The worst possible case is when f (1) = x1 happens to be either 1/3 or 2/3.)

x1

0 1/3 2/3 1

x1

0 1/3 2/3 1

x1

0 1/3 2/3 1

Figure 2:

Select a subinterval which does not contain x1 and call it J1 . Let us subdivide J1 into
three equal closed subintervals. Either x2 = f (2) ∈ / J1 or it lies in at most two of the
subintervals of J1 . In any case there exists a subinterval of J1 which does not contain f (2).
Choose one such and call it J2 . We proceed along this line to construct a subinterval Jn of
Jn−1 which does not contain f (n). Thus we would have obtained a nested sequence (Jn ) of
closed and bounded intervals. By the nested interval theorem their intersection is nonempty:
∩n Jn 6= ∅. Let x ∈ ∩Jn . Then x ∈ [0, 1]. So there exists an k ∈ N such that x = f (k). Thus
f (k) = x ∈ ∩n Jn . In particular, f (k) ∈ Jk . This contradicts our choice of Jn ’s. Hence our
assumption that [0, 1] is countable is wrong.

Theorem 5 (Bolzano-Weierstrass). Let (xn ) be a bounded sequence of real numbers. Then


there exists a convergent subsequence (xnk ) of (xn ).

2
Proof. Recall that a real sequence is a function x : N → R and that we let xn := x(n). Thus
the image of the sequence is the subset x(N) ≡ {xn : n ∈ N} ⊂ R.
If the image x(N) is finite, then there exists a real number α such that xn = α for infinitely
many n ∈ N, say, for all k ∈ S ⊂ N, an infinite subset. Using the well-ordering principle, we
can exhibit the elements of S as an increasing sequence of natural numbers:

n1 < n2 < · · · < nk < · · ·

Then the subsequence (xnk ) is the constant sequence α and hence is convergent.
So, we now assume that x(N) is infinite. Since (xn ) is bounded, there exists a positive
real number M such that

−M ≤ xn ≤ M, for all n ∈ N.

We bisect the interval J0 := [−M, M ] into two subintervals of equal length, [−M, 0] and
[0, M ]. Since x(N) ⊂ [−M, M ] = [−M, 0] ∪ [0, M ], at least one of the subintervals will contain
infinitely many elements of x(N), that is, infinitely many terms of the given sequences. Call
one such subinterval as J1 . Now we again bisect J1 into two subintervals of equal length. Since
x(N)∩J1 is infinite by our choice of J1 , one of the subintervals of J1 must have infinite number
of elements from x(N) ∩ J1 . Call it J2 . Thus, J2 ⊂ J1 , `(J2 ) = 2−1 `(J1 ) = 2−2 `(J0 ) = 2−2 2M .
Also, J2 has infinitely many terms from (xn ).
Proceeding inductively, we construct a nested sequence of intervals Jn with the following
properties:
(i) Jn+1 ⊂ Jn for all n ∈ N.
(ii) `(Jn ) = 2−n+1 M for all n ∈ N.
(iii) For each n ∈ N, the interval Jn contains infinitely many terms of the sequence (xn ).
Using the nested interval theorem, we get a real number α such that α ∈ ∩n Jn . Now
we inductively define a subsequence which will converge to α. Since J1 contains infinitely
many terms of the sequence (xn ), there exists n1 ∈ N such that x1 ∈ J1 . Assume that for all
1 ≤ i ≤ k, we have found xni such that xni ∈ Ji . Now, Jk+1 contains infinitely many terms
of the given sequence and hence there exists nk such that nk+1 > ni for 1 ≤ i ≤ k with the
property that xnk+1 ∈ Jk+1 . Thus we get a subsequence (xnk ).
We claim that xnk → α as k → ∞. For since α, xnk ∈ Jk , we have

|xnk − α| ≤ `(Jk ) = 2−k+1 M → 0, as k → ∞.

This completes the proof of the theorem.

An immediate corollary is the following:

Corollary 6. If (xn ) is a Cauchy sequence of real numbers, then (xn ) is convergent.

Proof. Recall the following facts.


(i) Any Cauchy sequence is bounded.
(ii) If (xn ) is a Cauchy sequence and (xnk ) is a subsequence convergent to a real number
α, then (xn ) converges to α.

3
The proof is an easy consequence of these two facts along with the last theorem. The
reader should complete the proof on his own.
Let (xn ) be a Cauchy sequence of real numbers. By the first fact, (xn ) is bounded. By
Bolzano-Weierstrass theorem, there exists a convergent subsequence (xnk ). It follows from
the second fact that the given Cauchy sequence is convergent.

Ex. 7. Let A be subset of R. We say that a real number x is an accumulation point (or a
cluster point) of A if for every ε > 0, the intersection (x − ε, x + ε) ∩ A contains a point other
than x. (The intersection may or may not contain x!) Adapt the proof of Thm. 5 to prove
the following version of Bolzano-Weierstrass theorem: If A is a bounded infinite subset of R,
there is an accumulation point of A in R.

Theorem 8 (Intermediate Value Theorem). Let f : [a, b] → R be continuous. Assume that


f (a) and f (b) are of different signs, say, f (a) < 0 and f (b) > 0. Then there exists c ∈ (a, b)
such that f (c) = 0.

Proof. Let J0 := [a, b]. Let c1 be the mid point of [a, b]. Now there are three possibilities for
f (c1 ). It is zero, negative or positive. If f (c1 ) = 0, then the proof is over. If not, we choose
one of the intervals [a, c1 ] or [c1 , b] so that f assumes values with opposite signs at the end
points. To spell it out, if f (c1 ) < 0, then we take the subinterval [c1 , b]. If f (c1 ) > 0, then
we take the subinterval [a, c1 ]. The chosen subinterval will be called J1 and we write it as
[a1 , b1 ].

a a
c

b c b

Figure 3: Figure 4:

a c

Figure 5:

We now bisect the interval J1 and choose one of the two subintervals as J2 := [a2 , b2 ] so
that f takes values with opposite signs at the end points. We continue this process recursively.
We thus obtain a sequence (Jn ) of intervals with the following properties:
(i) If Jn = [an , bn ], then f (an ) ≤ 0 and f (bn ) ≥ 0.
(ii) Jn+1 ⊂ Jn .
(iii) `(Jn ) = 2−n `(J0 ) = 2−n (b − a).

4
By nested interval theorem there exists a unique c ∈ ∩Jn . Since an , bn , c ∈ Jn , we have

|c − an | ≤ `(Jn ) = 2−n (b − a) and |c − bn | ≤ `(Jn ) = 2−n (b − a).

Hence it follows that lim an = c = lim bn . Since c ∈ J and f is continuous on J, we have

lim f (an ) = f (c) and lim f (bn ) = f (c).


n→∞ n→∞

Since f (an ) ≤ 0 for all n, it follows that limn f (an ) ≤ 0, that is, f (c) ≤ 0. In an analogous way,
f (c) = lim f (bn ) ≥ 0. We are forced to conclude that f (c) = 0. The proof is complete.

Ex. 9. Deduce from Thm. 8 the standard version of the Intermediate value theorem: Let
f : [a, b] → R be continuous. Let α be a real number in between f (a) and f (b), that is,
f (a) < α < f (b) or f (b) < α < f (a) whichever makes sense. Then there exists c ∈ (a, b)
such that f (c) = α. Hint: Assume f (a) < α < f (b). Consider g(x) = f (x) − α for x ∈ [a, b].
Apply Thm. 8.

Theorem 10. Let f : [a, b] → R be continuous. Then f is bounded on [a, b]. That is, there
exist L, M ∈ R such that
L ≤ f (x) ≤ M, for all x ∈ [a, b].

Proof. We say that a function f : [a, b] → R is bounded above if there exists M ∈ R such that
f (x) ≤ M for all x ∈ [a, b]. If f is not bounded above, then given any M ∈ R, there exists an
x ∈ [a, b] such that f (x) > M .
Let f : [a, b] → R be continuous. We prove that it is bounded above in [a, b]. By way of
contradiction, assume that it is not bounded above. We bisect J0 := [a, b] into subintervals
[a, c] and [c, b] of equal length as usual. Then f is not bounded above in at least one of the
subintervals. For, if not, then there exist M1 and M2 such that f (x) ≤ M1 for all x ∈ [a, c]
and f (x) ≤ M2 for all x ∈ [c, b]. It follows that if M = max{M1 , M2 }, then f (x) ≤ M for
all x ∈ [a, b], contrary to our assumption. Hence we conclude that f is unbounded in at least
one of the subintervals. Select one such subinterval and call it J1 .
We again bisect J1 and repeat the argument to conclude there exists a subinterval J2 on
which f is not bounded. Proceeding inductively, we get a sequence of intervals Jn with the
following properties:
(i) Jn+1 ⊂ Jn for all n.
(ii) `(Jn ) = 2−n `(J0 ) = 2−n (b − a).
(iii) f is unbounded on each Jn .
By nested interval theorem, there exists an x ∈ ∩Jn . Given n ∈ N, by the observation
made at the beginning of the proof, there exists xn ∈ Jn such that f (xn ) > n. Clearly,
as seen earlier, xn → x. By continuity of f at x ∈ [a, b], we have limn f (xn ) = f (x). In
particular, the sequence (f (xn )) is convergent. But this is impossible, since it is divergent to
+∞. This contradiction proves that our assumption that f is not bounded above in [a, b] is
wrong. Hence f is bounded above in [a, b].
(Another way of arriving at a contradiction runs as follows: By continuity of f at x, for
ε = 1, there exists a δ > 0 such that for all y ∈ (x − δ, x + δ), we have |f (y) − f (x)| < 1. It

5
follows that for any y ∈ (x − δ, x + δ), we have |f (y)| < 1 + |f (x)|. Since xn → x, for δ as
above, we can find n0 such that xn ∈ (x − δ, x + δ) for all n ≥ n0 . But for such n, we have

n < f (xn ) < 1 + |f (x)| .

As a consequence, the set N of natural numbers is bounded above by max{n0 , 1 + |f (x)|}.


This contradicts the Archimedean property.)
In a similar way, one can prove that f is bounded below, that is, there exists L ∈ R such
that f (x) ≥ L for all x ∈ R. Or, we apply the above result to the function g = −f to find a C
such that g(x) ≤ C for all x ∈ [a, b]. This implies that f (x) ≥ −C for all x ∈ [a, b]. That is,
f is bounded below in [a, b]. Thus we have shown that f is bounded above as well as below.
This completes the proof of the theorem.

Ex. 11. Let f : [a, b] → R be a continuous functions. Show that m := inf{f (x) : x ∈ [a.b]}
and M := sup{f (x) : x ∈ [a, b]} exist as real numbers and that there exist c, d ∈ [a, b] such
that f (c) = m and f (d) = M . Hint: That M is finite follows from Thm. 10. If f (x) 6= M for
all x ∈ [a, b], then consider g(x) := 1/(M − f (x)). Apply Thm. 10

Ex. 12. Use Thm. 5 to prove Thm. 10 as well as solve Exer. 11. Hint: Apply Thm. 5 to a
sequence (xn ) in [a, b] such that f (xn ) > n.

Theorem 13 (Heine-Borel). Let I = [a, b] be a closed and bounded interval. Let {Jα : α ∈ Λ}
be a family of open intervals such that J ⊂ ∪α Jα . Then there exists a finite subset F ⊂ Λ such
that J ⊂ ∪α∈F Jα , that is, J is contained in the union of a finite number of open intervals of
the given family.

Proof. We prove this by contradiction. Assume that J is not contained in the union of
any finite number of intervals from the given family. Bisect I as usual. Then one of the
subintervals is not contained in the union of any finite numbers of Jα ’s. (Why?) Call one
with this property as a rogue interval. Let I1 be a rogue interval of I. If we bisect I1 again,
one of the subintervals of I1 must be a rogue. Select one and call it I2 .
Continuing this process, we get a nested sequence (In ) of subintervals of I such that each
of them is a rogue and `(In ) ≡ 2−n (b − a) → 0 as n → ∞. By nested interval theorem, there
exists a unique x ∈ ∩In . Now x ∈ I ⊂ ∪Jα so that there exists λ ∈ Λ such that x ∈ Jλ . Since
Jλ is an open interval, there exists ε > 0 such that (x − ε, x + ε) ⊂ Jλ . (Why? If Jλ = (aλ , bλ ),
take ε < min{x − aλ , bλ − x}.)
Choose any N such that 2−N (b − a) < ε. We claim that IN ⊂ (x − ε, x + ε). Let y ∈ IN .
Since x, y ∈ IN , we have
|x − y| ≤ `(IN ) = 2−N (b − a) < ε.
Since y ∈ IN is arbitrary, it follows that IN ⊂ (x − ε, x + ε). Since (x − ε, x + ε) ⊂ Jλ , we
deduce that IN ⊂ Jλ . This is a contradiction to our choice of In ’s all of which are rogues!
This contradiction proves that our assumption that I is rogue is false and hence the theorem
is proved.

Ex. 14. Deduce from Thm. 8 the standard version of the Intermediate value theorem: Let
f : [a, b] → R be continuous. Let α be a real number in between f (a) and f (b), that is,
f (a) < α < f (b) or f (b) < α < f (a) whichever makes sense. Then there exists c ∈ (a, b)

6
such that f (c) = α. Hint: Assume f (a) < α < f (b). Consider g(x) = f (x) − α for x ∈ [a, b].
Apply Thm. 8.
Ex. 15. Let A be subset of R. We say that a real number x is an accumulation point (or a
cluster point) of A if for every ε > 0, the intersection (x − ε, x + ε) ∩ A contains a point other
than x. (The intersection may or may not contain x!) Adapt the proof of Thm. 5 to prove
the following version of Bolzano-Weierstrass theorem: If A is a bounded infinite subset of R,
there is an accumulation point of A in R.
Ex. 16. Let f : [a, b] → R be a continuous functions. Show that m := inf{f (x) : x ∈ [a.b]}
and M := sup{f (x) : x ∈ [a, b]} exist as real numbers and that there exist c, d ∈ [a, b] such
that f (c) = m and f (d) = M . Hint: That M is finite follows from Thm. 8. If f (x) 6= M for
all x ∈ [a, b], the consider g(x) := 1/(M − f (x)). Apply once again Thm. 8
Ex. 17. Use Thm. 5 to prove Thm. 8. Hint: Apply Thm. 5 to a sequence (xn ) in [a, b] such
that f (xn ) > n.
Theorem 18 (Density of Rational Numbers in R). Let x < y be real numbers. Then there
exists a rational number r ∈ Q such that x < y < r.

Proof. This proof is due to Ms. Udita, an MTTS participant.


If x (respectively y) is a rational number, then, by Archimedean property, we can find a
natural number N such that y − x > 1/N . Then the rational number x + N1 (respectively
y − N1 ) is as required. So we assume that neither of the given numbers is rational.
Since Z is unbounded in R, we can find integers m and n such that m < x < y < n.
Consider the midpoint (m + n)/2. It is different from x and y. We write
m+n m+n
I0 = [m, n] = I01 ∪ I02 ≡ [m, ]∪[ , n].
2 2
Now, if x and y lie in different intervals, it follows that x < m+n 2 < y. If they lie in the same
interval, say, [m, m+n
2 ], call it I1 . Now we bisect I 1 = I 11 ∪ I 12 and ask whether x and y lie in
different subintervals or not. If they do, we achieved what we wanted. Otherwise, we repeat
the process. We claim that at some finite stage, we must have x and y lying in different
subintervals In1 and In2 . For, otherwise, the infinite process will give us a sequence (In ) of
nested intervals, with `(In ) = 2−n n − m and such that x, y ∈ ∩n In . Since the lengths of the
intervals converge to 0, there can be only one point in ∩In . But this forces us to conclude
x = y, a contradiction. Hence at some n-th stage, x ∈ In1 ad y ∈ In2 . The common end point
of these subintervals is a rational number.

Volterra’s Proof of Nonexistence of a Function

Consider the function defined on (0, 1) by


(
1/q, if x = pq with g.c.d(p, q) = 1,
f (x) =
0 if x is irrational.

It is easy to show that f is continuous at each irrational point and discontinuous at all rational
points of (0, 1). One may now want know whether there exists a function on (0, 1) which is

7
continuous at all rationals and discontinuous at all irrationals. The nonexistence of such a
function is usually proved using Baire Category theorem. Volterra proved this using a very
ingenious idea without using Baire’s theorem. His proof uses the nested interval theorem,
density of rationals and irrationals and the existence of a function which is discontinuous
only at rationals! We shall indicate his proof below.
Let us assume that there exists g : (0, 1) → R which is continuous at the rational points
and discontinuous at irrationals. Let f be the function defined above. Choose any irrational
point x0 ∈ (0, 1). By continuity of f at x0 , given ε = 1/2, there exists a δ > 0 such that

|f (x) − f (x0 )| < 1/2, whenever |x − x0 | < δ.

Select points a1 < b1 ∈ (x0 − δ, x0 + δ). Then for all x, y ∈ [a1 , b1 ], we have

|f (x) − f (y)| ≤ |f (x) − f (x0 )| + |f (x0 ) − f (y)| < 1/2 + 1/2 = 1.

We now select a rational point y0 in the open interval (a1 , b1 ). We repeat the above argument
using now the continuity of g at y0 to construct a closed interval [c1 , d1 ] ⊂ (a1 , b1 ) such that

|g(x) − g(y)| < 1, for all x, y ∈ [c1 , d1 ].

Note that we have

|f (x) − f (y)| < 1 and |g(x) − g(y)| < 1 for all x, y ∈ [c1 , d1 ].

We repeat this argument replacing the open interval (0, 1) by the open interval (c1 , d1 ) to find
a closed interval [c2 , d2 ] ⊂ (c1 , d1 ) such that

|f (x) − f (y)| < 1/2 and |g(x) − g(y)| < 1/2 for all x, y ∈ [c2 , d2 ].

By induction we construct a sequence of nested intervals [ck , dk ] ⊂ (ck−1 , dk−1 ) for k ∈ N with
the property that

|f (x) − f (y)| < 2−k+1 and |g(x) − g(y)| < 2−k+1 for all x, y ∈ [ck , dk ].

By the nested interval theorem, there exists a unique point a ∈ [ck , dk ] for k ∈ N. Note that by
the fact that [ck+1 , dk+1 ] ⊂ (ck , dk ), the point a ∈ (ck , dk ) for all k. We now show that f and
g are continuous at a. Given ε > 0, choose n such that 2−n < ε. Then, for x ∈ [cn+1 , dn+1 ],
we have
|f (x) − f (a)| < ε and |g(x) − g(a)| < ε. (2)
Since a ∈ (cn+1 , dn+1 ), we can find δ > 0 such that (a − δ, a + δ) ⊂ (cn+1 , dn+1 ). It is clear
that if |x − a| < δ, then (2) holds, that is, f and g are continuous at a. It follows then that
a must be rational as well as irrational, which is absurd! Thus we conclude no such f exists.

Remark 19. I learnt this proof from Dr. V. Sholapurkar, S.P. College, Pune.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy