Consider the expression $\frac{1}{2}+\frac{2}{5}+\frac{3}{11}+\frac{4}{23}+...$

Denote the numerator and the denominator of the $j^\text{th}$ term by $N_{j}$ and $D_{j}$, respectively. Then, $N_1=1$, $D_1=2$, and, for every $j>1$, $$ N_j= N_{j-1}+1\qquad D_j= 2D_{j-1}+1$$

What is the $50^\text{th}$ term?

Must we evaluate that term-by-term until we reach the $50^\text{th}$ term?

What is the sum of the first $25$ term?

Must we add them one-by-one?

What is the exact value of the sum to $\infty$?

  • 2
    $\begingroup$ Answers to your question: no and no. $\endgroup$ – Yves Daoust Dec 9 '18 at 9:19
  • 1
    $\begingroup$ Did you mean to write $D_j = 2D_{j-1}+1$ for $j > 1$? $\endgroup$ – JimmyK4542 Dec 9 '18 at 9:20
  • $\begingroup$ @JimmyK4542 Yes, I did. $\endgroup$ – Hussain-Alqatari Dec 9 '18 at 9:23
  • $\begingroup$ @AweKumarJha , $\frac{50}{523}$ is not true, the $50^\text{th}$ term is much less than that. $\endgroup$ – Hussain-Alqatari Dec 9 '18 at 9:26
  • $\begingroup$ Then it just looks like a very similar series involving primes. Try using the method of generating functions to convert the recursion into explicit formula. $\endgroup$ – Awe Kumar Jha Dec 9 '18 at 9:28

The $n^{th}$ term is $\frac{n}{3\cdot 2^{n-1}-1}$, as can easily be proven by induction. [You can guess this by simply looking at the first two terms as you know it has to be of the form $c\cdot2^n+d$]

$\therefore 50^{th}$ term$=\frac{50}{3\cdot 2^{49}-1}$.

I tried all the methods I know(which includes generating functions, bruteforce calculation, CAS). The sum does not have a closed form formula.(As is the case with most nontrivial rapidly converging sums). You may however calculate the sum with arbitrary precision pretty easily.

  • $\begingroup$ This is helpful. Thank you. What about my second and the third questions? $\endgroup$ – Hussain-Alqatari Dec 9 '18 at 9:30
  • $\begingroup$ I am trying those parts. Will edit as soon as I get something useful. As of now, I think there is no closed form for the summation. $\endgroup$ – Anubhab Ghosal Dec 9 '18 at 9:32
  • 1
    $\begingroup$ Please use \cdot instead of .. The notation is really confusing. $\endgroup$ – Kemono Chen Dec 9 '18 at 9:35
  • $\begingroup$ @KemonoChen, fixed it. $\endgroup$ – Anubhab Ghosal Dec 9 '18 at 9:36
  • 1
    $\begingroup$ Appreciate the work, THANKS! $\endgroup$ – Hussain-Alqatari Dec 9 '18 at 10:36

If $d_{j+1}=2d_j+1$ then $d_{j+1}+1=2(d_j+1)\Rightarrow d_n+1=(d_1+1)2^{n-1}=3\cdot2^{n-1}$.

So 50th term is $\frac{50}{3\cdot2^{49}-1}$


$D_j=2D_{j-1}+1\\\ \ \ \ =2(2D_{j-2}+1)+1\\\ \ \ \ =4D_{j-2}+1+2\\\ \ \ \ =4(2D_{j-3}+1)+1+2\\\ \ \ \ \ \ \ \ \vdots\\\ \ \ \ =2^kD_{j-k}+2^k-1\\\ \ \ \ =2^{j-1}D_1+2^{j-1}-1\\\ \ \ \ =3\cdot2^{j-1}-1$

$s_n=N_n/D_n=\displaystyle\frac n{3\cdot2^{n-1}-1}, n\ge1$

$s_n<\displaystyle\frac n{3\cdot2^{n-1}-2^{n-1}}=\frac n{2^n}$

$\displaystyle\sum_1^\infty s_n<\sum_1^\infty \frac n{2^n}$ which is an AP-GP series

$\sum_1^\infty \frac n{2^n}=\frac12+\frac24+\frac38...$

$\frac12\sum_1^\infty \frac n{2^n}=0+\frac14+\frac28+\frac3{16}...$

$\sum_1^\infty \frac n{2^n}-\frac12\sum_1^\infty \frac n{2^n}=\frac12\sum_1^\infty \frac n{2^n}=\frac12+\frac14+\frac18...=1$

$\displaystyle\implies0<\sum_1^\infty s_n<2$

  • $\begingroup$ The sum is not less than $4/3$, it is $1.5997809\dots$. $\endgroup$ – Hussain-Alqatari Dec 9 '18 at 9:58
  • $\begingroup$ My bad, I'll recheck what I wrote. $\endgroup$ – Shubham Johri Dec 9 '18 at 10:00
  • $\begingroup$ Actually, s_n > n/(3⋅2^n−1) $\endgroup$ – Ankit Kumar Dec 9 '18 at 10:01
  • $\begingroup$ Yeah, thanks for pointing it out. $\endgroup$ – Shubham Johri Dec 9 '18 at 10:03

Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.