1
$\begingroup$

Can you prove (preferably combinatorially) the following identity for the total number of perfect matchings of the complete graph $K_{2n}$, where the edges in the matching are ordered, i.e., $\binom{2n}{2,2,\ldots ,2} = \frac{(2n)!}{2^{n}} = n!(2n-1)(2n-3)\cdots 1$:

$$ \binom{2n}{2,2,\ldots ,2} = \prod_{k=1}^{n} \left[\binom{n+1}{2}-\binom{k}{2}\right] = n\prod_{j=1}^{n-1} [n+(n-1)+ \ldots + (n-j)]. $$

It is also immediate to show that the total number of matchings is $\prod_{i=2}^{n} [i(i+(i-1))]$.

Thus it suffices to show that for every $n\geq 1$, \begin{equation} n[n+(n-1)]\cdots [n+(n-1)+\cdots + 1] = [n(n+(n-1))][(n-1)((n-1)+(n-2))]\cdots [2(2+1)]. \end{equation}

Algebraic proof would be also of some help. I tried by induction unsuccessfully.

$\endgroup$
9
  • $\begingroup$ Have you checked oeis.org/A000680? Also look at (2n)!/(2^n n!), i.e., oeis.org/A001147, more often associated with perfect matchings of the complete graphs. $\endgroup$ Commented Oct 11, 2022 at 21:07
  • $\begingroup$ @TomCopeland I have checked and this formula is indeed there, but I cannot find a way to contact the person who entered it there (Peter Bala). $\endgroup$
    – sdd
    Commented Oct 11, 2022 at 21:41
  • $\begingroup$ Peter Bala has made many contributions to the OEIS that overlap with my interests, but I have found no contact info for him. You might try emailing Sloane and requesting CI from him. $\endgroup$ Commented Oct 11, 2022 at 22:30
  • $\begingroup$ Thanks. I was just thinking that this should be easy to prove, but I cannot do it. $\endgroup$
    – sdd
    Commented Oct 12, 2022 at 0:43
  • 2
    $\begingroup$ Using factorization ${n+1\choose 2}-{k\choose 2}=(n-k+1)(n+k)/2$ you easily verify your formula. Combinatorial proof seems to be more tricky. $\endgroup$ Commented Oct 12, 2022 at 20:56

1 Answer 1

1
$\begingroup$

It is somewhat unclear from the question which $=$ indicate known identities and which indicate conjectured identities. I assume from the "it suffices" that what you want is a proof of $$n[n+(n-1)]\cdots [n+(n-1)+\cdots + 1] \stackrel?= [n(n+(n-1))][(n-1)((n-1)+(n-2))]\cdots [2(2+1)]$$ or, more succinctly, $$\prod_{k=1}^n \left(kn - \frac{k(k-1)}{2}\right) \stackrel?= n! (2n-1)!!$$

If we multiply both sides by $2^n$ we get $$\prod_{k=1}^n \left(2kn - k(k-1)\right) \stackrel?= (2n)!$$

But then $$\prod_{k=1}^n \left(2kn - k(k-1)\right) = \left(\prod_{k=1}^n k \right) \prod_{k=1}^n \left(2n - (k-1)\right) = \left(\prod_{k=1}^n k \right) \prod_{j=n+1}^{2n} j = (2n)!$$ with the substitution $j = 2n - (k-1)$.

Note as a corollary which I, at least, find interesting, that $$\prod_{k=1}^n \left(n - \frac{k-1}{2}\right) = (2n-1)!!$$


This suggests that a combinatorial proof will have to treat the terms differently according to the parity of $k$:

$$\prod_{k=1}^n \left(kn - \frac{k(k-1)}{2}\right) = \underbrace{\left(\prod_{j=1}^{n/2} \color{red}{j} \color{blue}{(2n - 2j + 1)}\right)}_{k \textrm{ even}}\, \underbrace{\prod_{j=0}^{(n-1)/2} \color{blue}{(2j+1)} \color{\red}{(n - j)}}_{k \textrm{ odd}} = \color{red}{n!} \color{blue}{(2n-1)!!}$$

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .