Recall that the $q$-analogue $[n]_q\in\mathbb Z[q]$ of a natural number $n\in\mathbb N$ is defined as $$ [n]_q := \frac{q^n -1}{q-1}$$ the idea being that formulas involving $q$ will specialize along $\mathbb Z[q]/(q-p^n)$ to counting formulas about finite fields, while specializing along $\mathbb Z[q]/(q-1)$ will yield counting formulas about finite sets.

For example, defining the $q$-factorial by $$[n]_q! := [1]_q [2]_q \dotsm [n]_q,$$ we have that the number of $k$-dimensional subspaces of $\mathbb F_q^n$ is $$\#\mathrm{Gr}_{n,k}(\mathbb F_q) = \binom nk_q :=\frac{[n]_q!}{[n-k]_q![k]_q!}$$ and $\binom nk_q$ reduces to $\binom nk$ when $q\to 1$.


Now let $p$ be a fixed prime. In my work, I've come across the expressions $$ [p]_q^{k_1} [p^2]_q^{k_2} \dotsm [p^n]_q^{k_n} $$ with $k_i\ge0$, $i=1,\dotsc, n$, as well as funny things like $$ [p^r]_{q^{p^s}} = \frac{q^{p^{r+s}}-1}{q^{p^s}-1}$$

Do these have any known combinatorial interpretation?

I will usually want to consider the above product when the values of $n$ and $\sum k_i$ are fixed, so we can consider it ranging over all partitions of $\sum k_i$ into $n$ non-negative integers.


Ultimately I'm interested in the specialization $\mathbb Z[q] \to \mathbb Z_p[\![q-1]\!] \to \mathbf A_{\mathrm{inf}}(R)$, where $R$ is a perfectoid ring containing a compatible choice of roots of unity $\zeta_{p^\infty}$. Letting $$\epsilon = (1,\zeta_p,\dots) \in R^\flat,$$ the structure map $\mathbb Z[\![q-1]\!] \to \mathbf A_{\mathrm{inf}}$ is given by $q\mapsto[\epsilon]$. Then the Fontaine map $$\tilde\theta_r\colon \mathbf A_{\mathrm{inf}}(R) \to W_r(R)$$ has kernel generated by $[p^r]_q$ (Example 3.16), while the map $\theta_r = \tilde\theta_r \varphi^r$ has kernel generated by $[p^r]_{q^{1/p^r}}$. You can get the product of these ideals, above, out of some $RO(S^1)$-graded THH calculations.

  • $\begingroup$ Are the $k_i$ restricted to be less than $p$ or just have the sum fixed and no other restriction (besides non-negativity). Is $q$ an indeterminate or do you want it to be real in $(0,1)$ or complex with $|q| < 1$ or $p$-adic with $|q-1|_p < 1$? $\endgroup$ – KConrad Jun 5 at 0:23
  • $\begingroup$ If you haven't come across them combinatorially, would you say how you came across them? $\endgroup$ – AHusain Jun 5 at 0:34
  • $\begingroup$ @KConrad there's no restriction on the $k_i$. For the combinatorial interpretation I'd want to know what happens when $q=1$ and when $q=p^m$, but for my applications $q-1$ is going to specialize to $\mu\in \mathbf A_{\mathrm{inf}}$; see the motivation added to the question. $\endgroup$ – Yuri Sulyma Jun 5 at 1:54

Recall Legendre's formula

$$ v_p(n!) = \sum_{s=1}^\infty\left\lfloor\frac n{p^s}\right\rfloor = \sum_{r=0}^\infty a_r[r]_p $$ where $n = \sum a_r p^r$ is the base-$p$ expansion of $n$.

A $q$-analogue of this formula is provided by Lemma 4.8 of The $p$-completed cyclotomic trace in degree 2, which states that

\begin{equation}\label{AClB}\tag{$\ast$} \large [n]_q! = u\prod_{r=1}^\infty \varphi^{r-1}([p]_q)^{\lfloor n/p^r\rfloor} \end{equation}

for a unit $u\in\mathbb Z_p[\![q-1]\!]^\times$. While my computations yielded products $$ [p]_q^{k_1} [p^2]_q^{k_2} \dotsm [p^n]_q^{k_n}, $$ for arbitrary $k_i\ge0$, the ones most relevant to my eventual application (the regular slice filtration on THH) were the principal ideal generated by the RHS of \eqref{AClB} (when $p$ divides $n$).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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