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!

P-series ln(n)/n

  1. May 26, 2014 #1
    In determining whether ∑ln(n)/n converges or diverges, where n=[0,∞], ln(n)/n must be evaluated at the upper and lower bound. My notes breezed through this part without much explanation on finding the bounded values for ln(n)/n. I suspect it might have something to do with squeeze theorem.
    Or rather, why is n [itex]\geq 3[/itex] a good value?
    (I'm not sure if this should be homework because technically it isn't but some clarification on the question I have would be great!)
     
    Last edited: May 26, 2014
  2. jcsd
  3. May 26, 2014 #2

    pasmith

    User Avatar
    Homework Helper

    Let [itex]N \in \mathbb{N}[/itex]. If [itex]f : [N, \infty) \to \mathbb{R}[/itex] is a strictly decreasing strictly positive function, then [itex]\sum_{n=N}^M f(n)[/itex] can be bounded by [tex]
    \int_{N}^M f(x)\,dx \leq \sum_{n=N}^M f(n) \leq f(N) + \int_{N+1}^M f(x-1)\,dx
    = f(N) + \int_N^{M-1} f(x)\,dx.[/tex] To see this, draw the graphs of [itex]f(x)[/itex], [itex]f([x])[/itex] and [itex]f(x-1)[/itex], where [itex][x][/itex] is the greatest integer less than or equal to [itex]x[/itex] and compare the areas under each. Thus by the squeeze theorem, [tex]
    \lim_{M \to \infty} \int_{N}^M f(x)\,dx \leq \lim_{M \to \infty}\sum_{n=N}^M f(n) \leq
    f(N) + \lim_{M \to \infty} \int_N^{M-1} f(x)\,dx \leq f(N) + \lim_{M \to \infty} \int_N^M f(x)\,dx[/tex] so that the sum converges if and only if the integral converges.

    [itex]\ln(x)/x[/itex] is not strictly decreasing or strictly positive on [itex][1,\infty)[/itex] and isn't even defined for [itex]x = 0[/itex]. It is strictly increasing on [itex](0, e][/itex] and is then strictly decreasing and strictly positive on [itex][e, \infty)[/itex]. Thus we must take [itex]N = 3[/itex] and split the sum as [tex]
    \lim_{M \to \infty} \sum_{n=1}^M \frac{\ln(n)}n = \frac{\ln(2)}2 + \lim_{M \to \infty} \sum_{n=3}^M\frac{\ln(n)}n.[/tex]
     
    Last edited: May 26, 2014
  4. May 26, 2014 #3

    mathman

    User Avatar
    Science Advisor
    Gold Member

    ln(n)/n > 1/n (n>2). ∑1/n diverges, therefore ∑ln(n)/n diverges.
     
  5. May 26, 2014 #4
    Hi

    Why isn't ln(x)/x strictly increasing on the domain [1,∞)?
    ln(x)/x - ln(x-1)/(x-1) gives a positive increment and continues on as our domain tends towards infinity. Or is it categourically "not strictly decreasing" because it grows too slow?
    To begin with, what is the definitition of strictly positive in this context?
     
  6. May 27, 2014 #5

    mathman

    User Avatar
    Science Advisor
    Gold Member

    ln(x)/x is decreasing -> 0.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook




Similar Discussions: P-series ln(n)/n
  1. Partition Function p(n) (Replies: 20)

  2. N /n! = ? (Replies: 21)

Loading...