3
$\begingroup$

Recall the constructions $[n]_q=\frac{1-q^n}{1-q}, [n]!_q=[1]_q[2]_q\cdots[n]_q$ with $[0]!_q:=1$ and the $q$-binomials (Gaussian polynomials) $$\binom{n}k_q=\frac{[n]!_q}{[k]!_q[n-k]!_q}.$$ Given two polynomials $f(q)$ and $g(q)$, we write $f(q)\geq g(q)$ provided that $f(q)-g(q)$ is a polynomial having non-negative coefficients.

I would like to ask:

QUESTION. Suppose $0\leq k\leq a<b$ are integers. Is it true that $\binom{b+a}{b-k}_q\geq\binom{a+b}{a-k}_q$?

$\endgroup$
1
  • $\begingroup$ There is a closely related question on unimodality discussed here: arxiv.org/pdf/1410.7087.pdf The cool thing is that a similar difference gets an interpretation from representation theory, in terms of Kronecker coefficients. $\endgroup$ Oct 11, 2022 at 6:09

2 Answers 2

6
$\begingroup$

We know that there is a $q$-unimodality of the $q$-binomial coefficients. That is, $\binom{n}{k}_q - \binom{n}{k-1}_q$ has nonnegative coefficients for $k \leq n/2$. This was shown by Lynne M. Butler in A unimodality result in the enumeration of subgroups of a finite abelian group (in more generality than just $q$-binomials).

Now we just observe that $b-k$ is closer to the center peak at $(a+b)/2$ then $a-k$ is. So, $\binom{b+a}{b-k}_q - \binom{a+b}{a-k}$ has nonnegative coefficients as desired.

$\endgroup$
1
  • $\begingroup$ A paper by G.E.Andrews, On the difference of successive Gaussian polynomials, J. Statistical Planning and Inference, 1993 gives a combinatorial proof regrading $\binom{n}k_q-\binom{n}{k-1}_q\geq0$. The last line in your argument seems to require more justification (although it is plausible, in principle). $\endgroup$ Oct 10, 2022 at 19:29
0
$\begingroup$

Just to add an alternating approach to that of Butler and Andrews, let's show that $\binom{n}k_q-\binom{n}{k-1}_q\geq0$, provided $2k\leq n$.

Let $n=\alpha k+d$ where $0\leq d<k$. Rewrite \begin{align*} \binom{n}k_q-\binom{n}{k-1}_q &=q^k\binom{n}{k-1}_q\frac{1-q^{(\alpha-2)k}}{1-q^k} +q^{(\alpha-1)k}\binom{n}{k-1}_q\frac{1-q^{d+1}}{1-q^k}. \end{align*} Observe $\frac{1-q^{(\alpha-2)k}}{1-q^k}$ is already a polynomial with non-negative coefficients. Furthermore, since $U(q):=\binom{n}{k-1}_q$ is unimodal, the coefficient of $q^j$ in $U(q)\cdot(1-q^{d+1})$ is non-negative as long as $2j\leq\deg(U)$. The same is true for $U(q)\frac{1-q^{d+1}}{1-q^k}$ as a formal power series. Since the polynomial $U(q)\frac{1-q^{d+1}}{1-q^k}$ is symmetric, having degree no greater than $U(q)$, all remaining coefficients of $U(q)\frac{1-q^{d+1}}{1-q^k}$ are non-negative.

$\endgroup$

Your Answer

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

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