3

Right now I am working through archived papers of a math aptitude quiz. For some reason I seem to be haveing a hard time with these series problems. I have managed to write the above series in a compact form but thats as far as I got.

$$\frac{1}{5}-\frac{1 \cdot 4}{5 \cdot 10}+\frac{1 \cdot 4 \cdot 7}{5 \cdot 10 \cdot 15}-\ldots = \sum\limits_{i=0}^\infty (-1)^i\frac{\prod\limits_{j=0}^i (3j+1)}{5^{i+1}(i+1)!}$$

Help please!

2 Answers2

6

Lab Bhattacharjee is right. This is just an instance of the usual binomial series. More precisely, $$ (1+x)^{-1/3}=1-\frac13x+\frac{1\cdot4}{3^2\cdot 2!}x^2-\frac{1\cdot4\cdot7}{3^3\cdot 3!}x^3+\cdots, $$ that converges and has the prescribed sum, whenever $-1<x< 1$. Plugging in $x=3/5$ gives $$\frac12\root 3\of5=1-\frac15+\frac{1\cdot4}{5\cdot10}-\frac{1\cdot4\cdot7}{5\cdot10\cdot15}+\cdots.$$

This yields $1-\dfrac12\root3\of5$ as the value of the sum.

Jyrki Lahtonen
  • 133,153
5

Since: $$ \prod_{j=0}^{i}\frac{3j+1}{5j+5} = \frac{1}{\left(\frac{5}{3}\right)^{i+1}\Gamma\left(\frac{1}{3}\right)\Gamma\left(\frac{2}{3}\right)}\cdot B\left(\frac{4}{3}+i,\frac{2}{3}\right) \tag{1}$$ we have:

$$ S=\sum_{i\geq 0}(-1)^i \prod_{j=0}^{i}\frac{3j+1}{5j+5} = \frac{\sqrt{3}}{2\pi}\int_{0}^{1}\sum_{i\geq 0}(-1)^i \left(\frac{3}{5}\right)^{i+1}x^{1/3+i}(1-x)^{-1/3}\,dx \tag{2}$$ hence: $$ S = \frac{\sqrt{3}}{2\pi}\int_{0}^{1}\frac{3x^{1/3}}{(1-x)^{1/3}(5+3x)}\,dx =\frac{\sqrt{3}}{2\pi}\int_{0}^{+\infty}\frac{3z^{1/3}}{(1+z)(5+8z)}\,dz\tag{3}$$ where in the last step we set $x=\frac{z}{1+z}$. That leads to: $$ S = \frac{\sqrt{3}}{2\pi}\int_{0}^{+\infty}\frac{9t^3}{(1+t^3)(5+8t^3)}\,dt=\frac{\sqrt{3}}{2\pi}\left(\int_{0}^{+\infty}\frac{3\,dt}{1+t^3}-\int_{0}^{+\infty}\frac{15\,dt}{5+8t^3}\right) \tag{4}$$ and the last integrals can be easily computed through the residue theorem or other techniques, leading to:

$$ S = \color{red}{1-\frac{1}{2}\sqrt[3]{5}}.\tag{5} $$

Jack D'Aurizio
  • 353,855
  • 1
    Easily? I respectfully beg to differ. The integral in (3) looks somewhat easier than the one in (4) if you are to take a residue approach. But perhaps you might want to summarize how you would apply the residue theorem to the integral in (4) to all the poor unwashed masses who are not as facile with this method of integration than yourself. – Ron Gordon Jan 19 '16 at 12:35
  • @RonGordon: with an extra step of partial fraction decomposition, the problem just boils down to computing $\int_{0}^{+\infty}\frac{dt}{1+t^3}$ that can be tackled in many ways. My updated answer clearly shows that. – Jack D'Aurizio Jan 19 '16 at 12:45