1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Homework Help: Metric Spaces

  1. Dec 15, 2011 #1
    1. The problem statement, all variables and given/known data


    2. Relevant equations

    3. The attempt at a solution

    I've done the first 3 parts. I've come to the bit on Cauchy sequences at the end. How do I show [itex]x_n = n[/itex] is/isn't a Cauchy sequence in the 2 metrics?

    [itex](x_n)[/itex] is a Cauchy sequence in a metric space [itex](X,d)[/itex] if for any [itex]\varepsilon >0[/itex] there exists [itex]N\in \mathbb{N}[/itex] such that if [itex]m,n > N[/itex] then [itex]d(x_m , x_n ) < \varepsilon[/itex].

    The metric space [itex](X,d)[/itex] is complete if every Cauchy sequence in [itex](X,d)[/itex] converges to a limit in [itex]X[/itex].
  2. jcsd
  3. Dec 15, 2011 #2
    So let's prove that it isn't a Cauchy sequence in the standard metric.

    So, we need to find an epsilon (let's take [itex]\epsilon=1[/itex] for a moment), such that for all N, there exists n,m>N such that

    [tex]d(x_n,x_m)\geq 1[/tex]

    So we must find n,m>N such that

    [tex]|n-m|\geq 1[/tex]

    Can you find such a n and m??
  4. Dec 15, 2011 #3


    User Avatar
    Homework Helper

    Well, let ε be some number in <0, 1>. Does there exist some positive integer N such that for all m, n >= N you have |xm - xn| < ε?

    Edit: sorry, micromass seems to have answered first.
  5. Dec 15, 2011 #4
    Setting [itex]m=N+1[/itex] and [itex]n=N+2[/itex] we have that [itex]m,n > N[/itex] and [tex]|m-n|=|N+1-(N+2)|=|1-2|=1 \geqslant 1\;,\;\text{for all}\;N\in \mathbb{N}[/tex]
  6. Dec 15, 2011 #5
    Good!! So that proves that it isn't a Cauchy sequence.

    Now, to prove that it is a Cauchy sequence in the other metric, you must make


    smaller than [itex]\varepsilon[/itex]
  7. Dec 15, 2011 #6
    How do I make [itex]|\tan^{-1}(m) - \tan^{-1}(n)|<\varepsilon[/itex] ? I'm guessing that the fact given in the question that [itex]d(x,y)< \pi[/itex] might help?

    However, for the last part on whether [itex](\mathbb{R} ,d)[/itex] is complete, can I say: if it is complete then every Cauchy sequence in [itex](\mathbb{R} ,d)[/itex] must converge to a limit in [itex]\mathbb{R}[/itex].

    Suppose the Cauchy sequence [itex]x_n=n[/itex] converges in [itex](\mathbb{R} ,d)[/itex]. Then by part (c), [itex]x_n[/itex] must converge in the standard metric. Every convergent sequence is Cauchy, but [itex]x_n[/itex] is not Cauchy in the standard metric - contradiction. So [itex]x_n[/itex] does not converge in [itex](\mathbb{R} , d)[/itex]; therefore not every Cacuhy sequence in [itex](\mathbb{R} , d)[/itex] converges so it is not complete.
    Last edited: Dec 15, 2011
  8. Dec 15, 2011 #7
    Well, this showing that [itex]x_n=n[/itex] is Cauchy in [itex](\mathbb{R},d)[/itex] is equivalent to showing that [itex]y_n=\tan^{-1}(n)[/itex] is Cauchy in the standard metric.

    That is good.
  9. Dec 16, 2011 #8
    In examples I've seen on showing a sequence is Cauchy they often involve getting [itex]d(x_m , x_n) \leqslant X \to 0[/itex] where [itex]X[/itex] is some upper bound. Why does this show the sequence is Cauchy?

    For my sequence [itex]x_n = n[/itex] I could use the triangle inequality [itex]d(x_m , x_n) \leqslant d(x_m , \frac{\pi}{2} ) + d(\frac{\pi}{2} , x_n)[/itex]: [tex]d(x_m ,x_n) = | \tan^{-1} (x_m) - \tan^{-1}(x_n) |[/tex]
    [itex]\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\; \leqslant | \tan^{-1}(m) - \frac{\pi}{2} | + | \tan^{-1}(n) - \frac{\pi}{2} | \to 0 + 0=0[/itex] as [itex]m,n \to \infty[/itex]

    since [itex]\displaystyle \lim_{r\to\infty} \tan^{-1}(r) = \frac{\pi}{2}[/itex]
    Last edited: Dec 16, 2011
  10. Dec 16, 2011 #9
    Well, write out what it means that [itex]d(x_m,x_n)\rightarrow 0[/itex]...

    That's good.
  11. Dec 16, 2011 #10


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    Be careful when you're writing this up! You're using the triangle inequality for absolute value here, but you're not using the triangle inequality for the metric d here - if you were actually using d(xn,pi/2) you would end up with |arctan(n)-arctan(pi/2)|+|arctan(m)-arctan(pi/2)| which doesn't go to zero as n and m go to infinity
  12. Dec 18, 2011 #11
    I'm not getting any credit for this; it's all past exam paper stuff.

    But, yes - it's the scalar triangle inequality I should be using!
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook