Convergence of a particular infinite sum

Join the discussion
Registration is free. Ask a follow-up in this thread, or start your own.
3 replies · 3K views
krissycokl
Messages
8
Reaction score
0

Homework Statement


Let [itex]b_n[/itex] be a bounded sequence of nonnegative numbers. Let r be a number such that [itex]0 \leq r < 1[/itex].
Define [itex]s_n = b_1*r + b_2*r^2 + ... + b_n*r^n[/itex], for all natural numbers n.
Prove that [itex]{s_n}[/itex] converges.


Homework Equations


Sum of first n terms of geometric series = [itex]sum_n = (a_1)(1-r^{n+1})/(1-r)[/itex]


The Attempt at a Solution


Clearly, [itex]{s_n}[/itex] is monotonically increasing.
Since [itex]{b_n}[/itex] is bounded, [itex]|b_n| \leq M[/itex], for all natural numbers n.

I want to use the fact that if [itex]{s_n}[/itex] is both monotonically increasing and is bounded, then it must converge. The part of the problem that has stumped me for the past 45 minutes is how to show that [itex]{s_n}[/itex] is bounded.

The only material we've covered regarding infinite series thus far is for purely geometric series, which doesn't fit this problem precisely--but I included the formula anyway.

Help would be greatly appreciated! I have an exam on Monday and getting so completely stymied by a simple problem is not doing wonders for my confidence.
 
Physics news on Phys.org
Aghhhh, so simple! Something about the way I do proofs is just <i>wrong</i>, I always get caught in roadblocks of thinking and miss simple detours like that.

[itex]s_n \leq M*r + M*r^2 + ... + M*r^n[/itex]
Then [itex]s_n \leq M(r-r^{n+1})/(1-r)[/itex]
Then [itex]s_n \leq Mr/(1-r) = M'[/itex]
Then [itex]|s_n| \leq M'[/itex] for all natural numbers n, so [itex]s_n[/itex] is bounded.
Thus, since [itex]s_n[/itex] is bounded and monotonically increasing, we have [itex]s_n[/itex] converges.

Or, I think that's right, anyway.

Thanks a bunch!
 
krissycokl said:
Aghhhh, so simple! Something about the way I do proofs is just <i>wrong</i>, I always get caught in roadblocks of thinking and miss simple detours like that.

[itex]s_n \leq M*r + M*r^2 + ... + M*r^n[/itex]
Then [itex]s_n \leq M(r-r^{n+1})/(1-r)[/itex]
Then [itex]s_n \leq Mr/(1-r) = M'[/itex]
Then [itex]|s_n| \leq M'[/itex] for all natural numbers n, so [itex]s_n[/itex] is bounded.
Thus, since [itex]s_n[/itex] is bounded and monotonically increasing, we have [itex]s_n[/itex] converges.

Or, I think that's right, anyway.

Thanks a bunch!

It's exactly right. Good take on the hint.