1. PF Contest - Win "Conquering the Physics GRE" book! Click Here to Enter
    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!

Analysis series proof

  1. Oct 11, 2009 #1
    1. The problem statement, all variables and given/known data
    Let (a_n) and (b_n) be sequences with bn > 0 for all n in N. Assume that lim (an/bn) = L with L > 0.
    Prove that the series:
    [tex]\sum{a_n}[/tex] converges if and only if [tex]\sum{b_n}[/tex] converges

    2. Relevant equations

    3. The attempt at a solution

    Since this is an if and only if proof, i know that I need to first assume if the an series converges, then so does the bn series. Next, I need to show if I assume the second series converges, then so does the first.

    Alright, this is what I tried so far:

    Assume [tex]\sum{a_n}[/tex] converges.
    Meaning that the sequence (an) converges to zero.

    If (an) converges to zero that means that for epsilon greater than zero there exists an N in N so that for n>=N:


    Since we are assuming lim (an/bn) converges to L>0 that means for epsilon greater than zero, there exists an N in N so that for n>=N:

    I think that to show this, I need to show that bn converges to zero as well, but I'm not too sure how to do that. Or also, there is a theorem stating if an and bn are sequences satisfying 0<=an<=bn for all n in N then
    if [tex]\sum{b_n}[/tex] converges then [tex]\sum{a_n}[/tex] converges.

    So I guess if I can show that an is larger than bn for all n in N then I can prove that the series converges as well.

    Any help/hints would be great!!!
  2. jcsd
  3. Oct 11, 2009 #2
    Just because bn converges to zero, that does not mean the series of bn will converge. It looks like the theorem you stated might be a good approach, but I'm not for sure. Note that if you can show bn≤can, for some constant c, that might help. Because if an converges, then can converges as well. I haven't worked this proof out, but these are just some thoughts.
  4. Oct 11, 2009 #3
    this was what I was thinking.... since the limit of an/bn>0 that means that an has to be >=bn for all n in N.
    We're also assuming bn>0 for all n in N which would mean
    0<=bn<=an for all n in N
    Which would mean if we assume that the series of an converges, that implies that the series of bn would converge.

    Is my assumption about the limits correct or am I way off?

    Also, then how would I go about proving that an converges if bn does if an is greater than or equal to bn for all n in N. Would I have to use the fact that if bn converges to zero if the series of bn converges?
  5. Oct 11, 2009 #4
    sorry i meant would i have to use the fact that if the series converges, then the sequence converges to zero
  6. Oct 11, 2009 #5

    I'll give you some hints to prove that if [itex]\sum{a_n}[/itex] converges, then so does [itex]\sum{b_n}[/itex]:

    1. When proving that an infinite sum converges, you can disregard any finite number of terms (the value of the sum might be affected, but not the convergence).
    2. If [itex]\sum{a_n}[/itex] converges and c is any constant, then [itex]\sum{ca_n}[/itex] also converges.
    3. If [itex]ca_n \geq 0[/itex] for some constant c and if [itex]|b_n| < ca_n[/itex] for all but finitely many n, then [itex]\sum{b_n}[/itex] converges (Comparison Test).
    4. We know that [itex]\lim_{x\rightarrow \infty}\frac{a_n}{b_n} = L > 0[/itex]. Use the [itex]\epsilon[/itex] definition of convergence with [itex]\epsilon = \frac{L}{2}[/itex]:

    There exists an N > 0 such that if n > N, then

    [tex]\left| \frac{a_n}{b_n} - L \right| < \frac{L}{2}[/tex]

    Simplify this inequality to conclude that

    [tex]0 < b_n < \frac{a_n}{\frac{L}{2}}[/tex]

    Now use observation #3 above to conclude that [itex]\sum{b_n}[/itex] converges. Note that the simplified inequality implies that [itex]a_n \geq 0[/itex] for all but finitely many [itex]a_n[/itex], justifying the use of #3.

    Let me know if this isn't clear, or if you would like more hints.

  7. Oct 12, 2009 #6
    That's what confusing to me...
    How do you get from
    |an/bn-L|<L/2 to that inequality? I know that 0<bn, but how do yo justify that an/(L/2)>b?

    Also, to show that an converges if bn converges would I have to find a constant c so that
    0<c*seriesan< series of bn?
  8. Oct 12, 2009 #7
    or actually i meant 0<can<bn for all n in N
  9. Oct 12, 2009 #8
    oh wait nevermind, I understand, i was confused cause I was looking at the part showing that an<=(3/2)L*bn
    But that is what I will use to prove that if bn converges then so does an. Thanks so much for your help!
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook