I need to change the summation order in the sum $$ \sum_{m,n=0}^\infty \left( \sum_{l=0}^{n+m} \left( \sum_{t=0}^l \binom{m}{t} \binom{n}{l-t} a_{n-l+2t,m+l-2t} \right) \right) \frac{ b_{m,n}}{m! n!}. $$ By experiments I get the following conjecture $$ \sum_{m,n=0}^\infty \left( \sum_{l=0}^{n+m} \left( \sum_{t=0}^l \binom{m}{t} \binom{n}{l-t} a_{n-l+2t,m+l-2t} \right) \right) \frac{ b_{m,n}}{m! n!}=\sum_{t,l=0}^\infty \left( \sum_{m=0}^t \sum_{n=0}^l \binom{t}{m} \binom{l}{n} a_{t-m+n,l-n+m}\, b_{t+l-m-n,m+n}\right) \frac{1}{t! l!}. $$ Any ideas how to prove it?
2026-03-25 17:44:53.1774460693
On
Change summation order
104 Views Asked by Bumbble Comm https://math.techqa.club/user/bumbble-comm/detail At
2
There are 2 best solutions below
0
On
Well, let's see.
$\begin{array}\\ \sum_{m,n=0}^\infty \sum_{l=0}^{n+m} \sum_{t=0}^l &=\sum_{m=0}^\infty\sum_{n=0}^\infty \sum_{l=0}^{n+m} \sum_{t=0}^l\\ &=\sum_{k=0}^\infty\sum_{n=0}^k \sum_{l=0}^{k} \sum_{t=0}^l \qquad k = n+m, n \le k, m = k-n\\ &=\sum_{k=0}^\infty\sum_{n=0}^k \sum_{t=0}^{k} \sum_{l=t}^k \qquad t \le l \implies l \ge t\\ &=\sum_{k=0}^\infty\sum_{t=0}^{k} \sum_{l=t}^k\sum_{n=0}^k \qquad\text{move sums around}\\ &=\sum_{t=0}^{\infty} \sum_{k=t}^\infty \sum_{l=t}^k\sum_{n=0}^k \qquad t \le k \implies k \ge t\\ &=\sum_{t=0}^{\infty}\sum_{l=t}^{\infty} \sum_{k=l}^\infty \sum_{n=0}^k \qquad t \le l \le k \end{array} $
Note: This is only a partial answer. The idea of the steps below is when doing rearrangements we also take care of $\binom{p}{q}=0$ whenever we have integral $0< p<q$.
Comment:
In (1) we split the sum using that $\binom{m}{t}=0$ if $t > m$, so that the upper index of the right-hand inner sum is set to $m$.
In (2) we exchange the inner left-hand sums respecting the index range $0\leq t\leq l\leq m$. We also simply exchange the right-hand inner sums and we shift the index to start with $l=1$.
In (3) we can now factor out $\sum_{t=0}^m\binom{m}{t}$ and we shift the index of the left inner sum by $t$ to start with $l=0$.
In (4) we do a reordering of the terms by $t\to m-t$. We also set the upper index of the right-most inner sum from $n$ to $n-t$, since other values do not contribute according to $\binom{n}{l+t}=0$.
In (5) we shift the index of the right inner sum by $t$ to start with $l=t+1$.
In (6) we can finally merge the two inner sums and obtain the wanted structure of the sums.
In (7) we replace $m\leftrightarrow t$ and $n\leftrightarrow l$.