Challenging limit

Calculus Level 5

lim n k = 0 n 2 ( n k k ) 1 2 n k = ? \large \lim_{n\to\infty} \sum_{k=0}^{ \big\lfloor \frac n2 \big \rfloor } \binom{n-k}k \frac1{2^{n-k}} = \, ?

3 5 \frac{3}{5} 5 8 \frac{5}{8} 2 3 \frac{2}{3} 5 7 \frac{5}{7}

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Michael Mendrin
Jun 28, 2018

k = 0 n ( n k k ) 1 2 n k \displaystyle \sum _{ k=0 }^{ n }{ \left( \begin{matrix} n-k \\ k \end{matrix} \right) \dfrac { 1 }{ { 2 }^{ n-k } } }

generates the series

1 2 , 3 4 , 5 8 , 11 16 , 21 32 , 43 64 , 85 128 , . . . \dfrac { 1 }{ 2 } ,\dfrac { 3 }{ 4 } ,\dfrac { 5 }{ 8 } ,\dfrac { 11 }{ 16 } ,\dfrac { 21 }{ 32 } ,\dfrac { 43 }{ 64 } ,\dfrac { 85 }{ 128 } ,...

where the numerators are Lucas numbers, generated by 2 L n 2 + L n 1 = L n 2L_{n-2}+L_{n-1}=L_n . Thus, the sum can be expressed as

2 n + 1 ( 1 ) n + 1 3 2 n \dfrac{2^{n+1}-(-1)^{n+1}}{3\cdot2^n}

which has the limiting value of 2 3 \dfrac{2}{3} as n n \rightarrow \infty

This solution assumes that the sum will keep generating the Lucas numbers ad infinitum. Otherswise, prove by induction.

Thank you for your surprisingly short and very good solution. Mine is much longer . When a spinner is split into two equal regions and numbers 1 and 2 are assigned to regions one and two respectively, the probability to get a running total of n, is calculated by the sum in the limit. I figured out a direct formula for calculation the same probability. It's 2 3 + 1 3 ( 1 2 ) n \frac{2}{3} + \frac{1}{3}(\frac{-1}{2})^n . It's limit is obviously 2 3 \frac{2}{3} .

A Former Brilliant Member - 2 years, 11 months ago

Log in to reply

Before you edited your problem, you suggested this means of figuring out the answer, but then you took it off, and I couldn't remember the details. It's interesting to see how this probability is related to the Lucas number series, starting with 1 , 3 , 5 1, 3, 5 .

Michael Mendrin - 2 years, 11 months ago

Log in to reply

Hi Michael. I you provide an email, I am willing to share all my findings related to this problem. It is several pages and probably can not be posted in here. Thank you for your interest.

A Former Brilliant Member - 2 years, 11 months ago

Log in to reply

@A Former Brilliant Member What you can do is to post a Note on Brilliant and then upload your findings, so that others may see it besides me.

Michael Mendrin - 2 years, 11 months ago

Log in to reply

@Michael Mendrin My findings are written as Microsoft file and the formulas are unrecognizable in the new format in a note. I tried to create a Website file. The formulas look better. Try this link C:\Users\Emil\Documents\Spinner's magic for brilliant.mht
I hope it will work.

A Former Brilliant Member - 2 years, 11 months ago

Log in to reply

@A Former Brilliant Member This is not an URL file, but apparently a file in your computer. Another thing you can try to do is to take a screenshot of your paper perhaps several, and save them as image files, and then upload them in a solution box here. The solution box has a taskbar on top that allows you to select image files from your computer to upload.

Another thing you can try is to copy your MS file as it appears in edit form, and paste it to some image software such as IrfanView, which wil convert it into an image file. Sometimes that works but that depends on your image software.

I prefer working with JPG image files, but Brilliant.or will display several other image file types.

Michael Mendrin - 2 years, 11 months ago

Log in to reply

@Michael Mendrin It's done. Thank you for your patience.

A Former Brilliant Member - 2 years, 11 months ago

Log in to reply

@A Former Brilliant Member How were you able to do math symbolism using Microsoft?

Also, it's rare for such a full mathematics paper would be presented in Brilliant.org. Even a lot of Brilliant wikis aren't this extensive.

Michael Mendrin - 2 years, 11 months ago
Patrick Corn
Jul 13, 2018

Let f ( n ) = k = 0 ( n k k ) 1 2 n k . f(n) = \sum_{k=0}^\infty \binom{n-k}{k} \frac1{2^{n-k}}. Consider the generating function n = 0 f ( n ) x n . \sum_{n=0}^\infty f(n) x^n. We get n = 0 f ( n ) x n = n = 0 k = 0 ( n k k ) 1 2 n k x n , \sum_{n=0}^\infty f(n) x^n = \sum_{n=0}^\infty \sum_{k=0}^\infty \binom{n-k}{k} \frac1{2^{n-k}} x^n, and now we make the substitution m = n k m=n-k to get n = 0 f ( n ) x n = m = 0 k = 0 ( m k ) 1 2 m x m + k = m = 0 ( x / 2 ) m k = 0 ( m k ) x k = m = 0 ( x / 2 ) m ( 1 + x ) m = 1 1 x 2 ( 1 + x ) = 2 2 x x 2 = 2 3 ( 1 1 x + 1 2 + x ) = 2 3 ( 1 1 x + 1 / 2 1 + x / 2 ) = 2 3 ( n = 0 x n + 1 2 n = 0 ( 1 / 2 ) n x n ) , \begin{aligned} \sum_{n=0}^\infty f(n) x^n &= \sum_{m=0}^\infty \sum_{k=0}^\infty \binom{m}{k} \frac1{2^m} x^{m+k} \\ &= \sum_{m=0}^\infty (x/2)^m \sum_{k=0}^\infty \binom{m}{k} x^k \\ &= \sum_{m=0}^\infty (x/2)^m (1+x)^m \\ &= \frac1{1-\frac{x}2(1+x)} \\ &= \frac2{2-x-x^2} \\ &= \frac23 \left( \frac1{1-x} + \frac1{2+x} \right) \\ &= \frac23 \left( \frac1{1-x} + \frac{1/2}{1+x/2} \right) \\ &= \frac23 \left( \sum_{n=0}^\infty x^n + \frac12 \sum_{n=0}^\infty (-1/2)^n x^n \right), \end{aligned} so equating coefficients of x n x^n gives f ( n ) = 2 3 + 1 3 ( 1 2 ) n , f(n) = \frac23 + \frac13 \left( -\frac12 \right)^n, which approaches 2 3 \frac23 as n . n\to\infty.

The best solution so far. Thank you!!!

A Former Brilliant Member - 2 years, 11 months ago

Did the same

Aditya Kumar - 2 years, 8 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...