0
$\begingroup$

I'm aware of the replies to this question, but I was not able to apply the suggested methods in my case.

I would like to compute the asymptotics of the resulting sum $S$ when computing the number of possibilities to distribute $k$ indistinguishable objects in $m$ distinguishable boxes with size limitation $R$ (see derivation here), which is given by \begin{align} S_{m,R}(k)=\sum_{t}(-1)^t \binom{m}{t}\binom{m+k-t(R+1)-1}{m-1}\,, \end{align} where I would like to take the limit $m\to\infty$, while keeping $R$ and the ratio $k/m$ fixed. From numerics, I can see explicitly that $S_{m,R}(k)$ rapidly approaches a Gaussian distribution centered around $m R/2$, but I would like to find an analytical form of this asymptotic Gaussian distribution.

If I define the quasi-continuous variable $x=k/m$, I expect the asymptotics \begin{align} S_{m,R}(x)\sim \alpha p(m) e^{\beta m-\gamma(x-x_0)^2 m}\quad\text{as}\quad m\to\infty\,, \end{align} where $p(m)$ is some polynomial in $m$ and $x_0=R/2$, while $\alpha$, $\beta$ and $\gamma$ are some real numbers.

$\endgroup$

1 Answer 1

1
$\begingroup$

The result is effectively derived in the reference for above. I overlooked this at first.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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