3

By the ratio test the series $$ \sum_{n\ge0}\frac{n^x}{n!} $$ is convergent, but I know no method to evaluate it.

Since it's a convergent series then my question here is:

Is there a closed form for $\displaystyle\sum_{n\geq 0}\frac{n^x}{n!}$ with $x$ a positive real number?

Jam
  • 10,632
  • 3
  • 29
  • 45

1 Answers1

0

I don’t think, that there is a closed form. The question should be changed to: "What interesting features does this series have (e.g. what integral representations are known) ?"

A note:

Be $\enspace\displaystyle f_{N,x}(z):= \sum\limits_{n=0}^N {\binom{N}{n}}n^x z^n\enspace $ with $\,x\geq 0\,$ , $\,z\in\mathbb{C}\,$ and $\,N\in\mathbb{N}_0\,$ .

It follows $\enspace\displaystyle \sum\limits_{n=0}^\infty \frac{n^x}{n!} = \lim\limits_{N\to\infty} f_{N,x}(\frac{1}{N}) \,$ .

For $\enspace x\in\mathbb{N}\enspace$ we have $\enspace\displaystyle f_{N,x}(z) = Nz(z+1)^{N-x}\sum\limits_{k=0}^{x-1}z^k \sum\limits_{v=0}^k {\binom{x-N}{k-v}}{\binom{N-1}{v}}(v+1)^{x-1}$

and we see that we get troubles generalizing (e.g) this to real $\,x> 0\,$ .

A hint $\,$ for $\,x\in\mathbb{N}\,$ :

$\displaystyle \sum\limits_{n=0}^\infty \frac{z^n n^x}{n!} = \lim\limits_{N\to\infty} f_{N,x}(\frac{z}{N}) =$

$\enspace\displaystyle = \lim\limits_{N\to\infty} (1+\frac{z}{N})^{N-x} z \lim\limits_{N\to\infty} \sum\limits_{k=0}^{x-1}(\frac{z}{N})^k \sum\limits_{v=0}^k {\binom{x-N}{k-v}}{\binom{N-1}{v}}(v+1)^{x-1}$

$\enspace\displaystyle =e^z \sum\limits_{k=0}^{x} S_{x,k} z^k\enspace$ with the Stirling numbers of the $2^{nd}$ kind $\,\displaystyle S_{x,k}:= \sum\limits_{v=0}^k \frac{(-1)^{k-v} v^x}{(k-v)!v!}\,$ .

The (first) series and the last sum of the equation chain are equal also for $\,x=0\,$ and $\,0^0:=1\,$.

user90369
  • 11,696
  • Thanks for this answer because i don't got any help from above comments , i think this series will make a sense when x is a complex variable – zeraoulia rafik Jan 08 '18 at 16:48
  • @zeraouliarafik : It doesn't change anything, if you use $x$ as a complex variable. But the formula for $f_{N,x}(z)$ leads to the (non-trivial) formula for $\displaystyle\sum\limits_{n=0}^{\infty} \frac{n^x z^n}{n!}$ if $x\in\mathbb{N}$ . Because of this it's possible to see, which problem we have, if $x$ isn't a positive integer. – user90369 Jan 08 '18 at 22:01
  • you are right ! my question In Mo has been reached ,then if you want to share this question in MO i'm agree probably we will find who give us enough informuation for this non-trivial formula – zeraoulia rafik Jan 08 '18 at 22:05
  • @zeraouliarafik : As Jack D'Aurizio has mentioned, you can discuss the Touchard polynomials with real index, maybe you can construct an approximation. And: With my answer I just wanted to put a more elementary view on the series which helps to understand, why the existence of a closed form is very unlikely. – user90369 Jan 09 '18 at 15:26