Solving Very Difficult Limit Homework

  • Thread starter Thread starter dirk_mec1
  • Start date Start date
  • Tags Tags
    Limit
dirk_mec1
Messages
755
Reaction score
13

Homework Statement



http://www.sosmath.com/CBB/latexrender/pictures/f45d5e5b84cb5aecef3fd2f568739e32.gif


Homework Equations



hint:

http://www.sosmath.com/CBB/latexrender/pictures/fee1863dab606636f986e63c261b565f.gif
http://www.sosmath.com/CBB/latexrender/pictures/61d6e1cc712fcb6ac6f19b40851e85be.gif
http://www.sosmath.com/CBB/latexrender/pictures/ae169a7ff6b1d2c09a977e5aad223650.gif
http://www.sosmath.com/CBB/latexrender/pictures/cd14347d968bdb5be67f72e832558702.gif

The Attempt at a Solution


I don't understand how the hint helps in this limit. Could someone point me in the right direction?
 
Last edited by a moderator:
Physics news on Phys.org
I didn't finish it, but here's my possible hint. If L=your limit, consider log L.

Edited to add: f(x) will be log(1+x). I "feel" like this is right.
 
Yes, that's helps a little bit but what do you mean by f(x) will be 1+x? How is that helpful?
 
dirk_mec1 said:
what do you mean by f(x) will be 1+x?

It will turn out that you will want to use f(x)=log(1+x).

Anyway, I did finish it now, and it works out. Use the log of the limit idea as I first suggested and apply properties of logs. At some step you will get terms with log(n+k). Rewrite as log[n(1+k/n)]=log n + log(1+k/n). Eventually you'll have to decide what to use for x, and it's not too hard to guess. Write out what you can, and I bet you will get it. If not, post your work and I'll give you more hints. I predict from time to time you'll say things to yourself like "wow, I can't believe those terms canceled out exactly, this must be right."
 
\ln(L) = \sum_{k=0}^n \frac{1}{2^n} \left( \stackrel{n}{k} \right) \left( \ln(n) + \ln \left(1+\frac{k}{n} \right) \right) - \ln(n)
Now if f(x) = \ln(1+x) then B_n(1) is what I have, I only need to add an extra (1/2^n)* ln(2) and (subtract it later) to use the hint but for B_n(1)\ \sum_{k=0}^n \left( \stackrel{n}{k} \right) \ln \left(1+\frac{k}{n} \right) - \ln(2) \leq \frac{1}{2 \sqrt{n}} \leq \frac{1}{2}
\left( \frac{1}{2} \right) ^n \sum_{k=0}^n \left( \stackrel{n}{k} \right) \ln(n) = \ln(n)

so this cancels the last term leaving only that annoying thing from above plus that term which I needed to substract:
\left( \frac{1}{2} \right) ^n \left[ \frac{1}{2} + \ln(2) \right]So the limit is 1? Am I going in the right direction?
 
Last edited:
Looking good, but B_n(1) would be 0. I used a different value for x.

(Also, leave \frac{1}{2\sqrt{n}} as is. Don't replace it by 1/2.)
 
x=0? then

<br /> \ \sum_{k=0}^n \left( \stackrel{n}{k} \right) \ln \left(1+\frac{k}{n} \right) - \ln(1) \leq \frac{1}{2 \sqrt{n}} <br />

This results in:

<br /> \left( \frac{1}{2} \right) ^n \left[ \frac{1}{2\sqrt{n}}\right] = \frac{ \frac{1}{2}^{n+1} }{\sqrt{n}} \rightarrow 0 <br />

as n \rightarrow \infty

Conclusion: the limit is 1.
 
Last edited:
dirk_mec1 said:
x=0?



B_n(x)=\sum_{k=0}^n \binom{n}{k} x^k (1-x)^{n-k} f(\frac{k}{n})

so B_n(0)=\sum_{k=0}^n \binom{n}{k} 0^k 1^{n-k} f(\frac{k}{n})

so B_n(0)=\binom{n}{0} 0^0 f(0), which can't be helpful.

Also, B_n(1)=\sum_{k=0}^n \binom{n}{k} 1^k 0^{n-k} f(\frac{k}{n})

so B_n(1)=\sum_{k=0}^n \binom{n}{k} 1^k 0^{n-k} f(\frac{k}{n})

so B_n(1)=\binom{n}{n} 0^0 f(1) which may not be 0 as I claimed earlier, but it still can't be helpful.

Oh well, average the two...
 
Oops :biggrin:

I see, if you take x =1/2 it matches my term exactly:
<br /> B_n \left( \frac{1}{2} \right) =\sum_{k=0}^n \binom{n}{k} \left( \frac{1}{2} \right) ^n f \left(\frac{k}{n} \right)<br />Now

| B_n \left( \frac{1}{2} \right) - f \left(\frac{1}{2} \right)| \leq \frac{1}{2 \sqrt{n} } <br /> \frac{1}{2 \sqrt{n}} + \ln \left( \frac{3}{2} \right) \longrightarrow \ln \left( \frac{3}{2}\ \right) \mbox{as}\ n \rightarrow \infty<br />So the limit is 3/2!
 
Last edited:
  • #10
dirk_mec1 said:
So the limit is 3/2!

Hooray! Isn't that nice!

I wrote \ln \left( \frac{3}{2} \right)-\frac{1}{2 \sqrt{n} }\le B_n \left( \frac{1}{2} \right) \le \ln \left( \frac{3}{2} \right)+ \frac{1}{2 \sqrt{n} } to make the final steps crystal clear.

Don't forget to verify |f(x)-f(y)|\le|x-y| so that the hint applies.
 
  • #11
Billy Bob said:
Hooray! Isn't that nice!
Yes it is :)

I wrote \ln \left( \frac{3}{2} \right)-\frac{1}{2 \sqrt{n} }\le B_n \left( \frac{1}{2} \right) \le \ln \left( \frac{3}{2} \right)+ \frac{1}{2 \sqrt{n} } to make the final steps crystal clear.
Right, so the squeeze theorem applies.

Don't forget to verify |f(x)-f(y)|\le|x-y| so that the hint applies.

I'm currently thinking on the proof for the hint but CS (after triangle inequality) gives me some nasty squares...
 
Back
Top