Nested intervals uniqueness proof in Sohrab theorem 2.1.45

  • Level: Undergrad 
  • Thread starter Math Amateur
  • Start date
  • Tags Tags
    Property Supremum
3 replies · 3K views · last reply
Math Amateur
Gold Member
MHB
Messages
3,920
Reaction score
48
I am reading Houshang H. Sohrab's book: "Basic Real Analysis" (Second Edition).

I am focused on Chapter 2: Sequences and Series of Real Numbers ... ...

I need help with yet a further issue/problem with Theorem 2.1.45 concerning the Supremum Property (AoC), the Archimedean Property, and the Nested Intervals Theorem ... ...

Theorem 2.1.45 reads as follows:
Printed textbook page with the statement of Theorem 2.1.45, the Supremum Property of R, followed by the beginning of its proof with set notation and inequalities.

Printed proof text from a real analysis textbook about nested intervals: showing the intersection of the intervals I_n is a single point u, and that u is the supremum of S.
In the above proof by Sohrab, we read the following:

" ... ...The Nested Intervals Theorem now implies that ##\bigcap_{ n = 1}^{ \infty } I_n = \{ u \}## for a unique ##u \in \mathbb{R}##. Indeed, if ##u \lt v## and ##u,v \in \bigcap_{ n = 1}^{ \infty } I_n##, then ##v - u \gt \frac{1}{2n}## for some ##n \in \mathbb{N}##, which contradicts ##u, v \in I_n##, since ##I_n## has length ##2^{ -n }##. ... ... "I am unsure of Sohrab's process and assumptions as he is moving through the proof in the above quote ... could someone confirm (or otherwise) my interpretations as follows ... there are essentially 4 questions ( Q1, Q2, Q3 and Q4 respectively ...) ... ...First issue ... ... I assume that when Sohrab writes: "Indeed, if ##u \lt v## ... etc etc ... " ... he is verifying his statement that ##\bigcap_{ n = 1}^{ \infty } I_n = \{ u \}## for a unique ##u \in \mathbb{R}##? Is that right? (Q1) Second issue ... ... when Sohrab writes: "Indeed, if ##u \lt v## ... etc etc ... " ... ... he could have said ##u \gt v## ... but he is just taking ##u \lt v## as an example ... and we are left to infer that ##u \gt v## works similarly ... in other words there is no reason that ##u## is taken as less than ##v## as against taking ##v \lt u## ... ... Is that right? (Q2)


Third issue ... ... Sohrab then asserts that ##v - u \gt \frac{1}{ 2^n }## ... ... and I am assuming this follows because ...

##u \lt v##

##\Longrightarrow v - u \gt 0##

##\Longrightarrow v - u \gt \frac{1}{n}## for some ##n \in \mathbb{N}## ... (Corollary 2.1.32 (b) Archimedean Property ... see scanned text insert below)

##\Longrightarrow v - u \gt \frac{1}{ 2^n }## ... ... ... ( Is this valid? (Q3) ... looks OK ... but justification ?

So indeed ... given we are doing analysis ... how do we justify ##\frac{1}{n} \gt \frac{1}{ 2^n }## or ##2^n \gt n##?

and further ... is my interpretation above for the third issue correct (Q4)


Help will be appreciated ...

Peter==========================================================================================The above theorem concerns the Supremum Property, the Archimedean Property and the Nested Intervals Theorem ... so to give readers the context and notation regarding the above post I am posting the basic information on these properties/theorems ...
Printed textbook page stating the Completeness Axiom (Supremum Property) for real numbers, followed by Exercise 2.1.29 asking to show equivalence with the Infimum Property, with a hint to use inf(S) = -sup(-S).
Textbook page showing Theorem 2.1.31 (Archimedean Property of R) with its proof, followed by Corollary 2.1.32 with parts (a), (b) and (c) on existence of natural numbers n relative to positive numbers x and y.
Textbook page showing Theorem 2.1.43 (Nested Intervals Theorem) with its proof, Exercise 2.1.44 about the intersection of nested intervals equaling [alpha, beta], and a closing paragraph on the Supremum and Archimedean Properties.
 
Attachments
  • Screenshot of a textbook page showing Theorem 2.1.45, The Supremum Property of R, followed by the start of its proof involving nested intervals and the Archimedean Property.
    Sohrab - 1 - Theorem 2.1.45 ... - PART 1 ... ....png
    37.9 KB · Views: 598
  • Printed proof text from a real analysis book, a proof using the Nested Intervals Theorem to show the intersection of intervals I_n is a single point u, and that u equals sup(S).
    Sohrab - 2 - Theorem 2.1.45 ... - PART 2 ... ....png
    29.9 KB · Views: 573
  • Textbook page excerpt stating the Completeness Axiom (Supremum Property) for real numbers and Exercise 2.1.29 asking to show it is equivalent to the Infimum Property, with a hint about inf(S) = -sup(-S).
    Sohrab - Axiom of Completeness ... Supremum Property ....png
    31.8 KB · Views: 564
  • Textbook page showing Theorem 2.1.31 (Archimedean Property of R) with its proof, followed by Corollary 2.1.32 with parts (a), (b) and (c) on inequalities involving positive numbers x and y and a natural number n.
    Sohrab - Theorem 2.1.31 - Archimedean Property ... ....png
    28.3 KB · Views: 859
  • Textbook page showing Theorem 2.1.43 (Nested Intervals Theorem) with its proof, followed by Exercise 2.1.44 asking to show that the intersection of nested closed intervals equals [alpha, beta].
    Sohrab - Theorem 2.1.43 ... Nested Intervals Theorem ....png
    48 KB · Views: 563
I'm only going to address Q1 and Q2 in this post. Really, since we know the diameter of the nested sequences converges to 0, we also know such a unique ##u## exists. Probably for pedagogical purposes, he shows that there cannot be two elements in the infinite intersection. Because these two elements ##u, v## are arbitrary anyway, there is no need to show both cases ##u<v## AND ##v<u##.
 
  • Like
Likes   Reactions: Math Amateur
With regard to the fact that ##v-u>2^{-n}##, your reasoning is sound. It is easy to show that ##2^n>n## for all natural numbers. First, ##2^0=1>0##. Second, ##2^n>n \implies 2^{n+1}>n+1##. Adding 1 to the antecedent, ##2^n+1>n+1##. Since ##2 \cdot 2^n > 2^n+1## is true whenever ##n>0## (verify this by subtracting ##2^n## and taking ##\log_2##), we know that ##2^{n+1}>n+1## as desired.

This reasoning is a bit tedious, though. It's pretty obvious that ##(1/2^n)_{n \in \mathbb{N}} \to 0##,and any such sequence must get smaller than any given positive number—this is the spirit of the Archimedean property: there are no arbitrarily small numbers. Therefore, I imagine he took for granted that there existed a ##v-u>1/2^n## for some n.

Let me know if you have follow up questions.
 
  • Like
Likes   Reactions: Math Amateur
Thanks for your guidance and support Someone2841 ... really helpful ...

Peter