3
$\begingroup$

In the paper [A new $q$-Analog of Stirling Numbers],(https://hal.archives-ouvertes.fr/hal-01372920/document)[PDF] J. Cigler defined $q$-Stirling numbers of the second kind as the following:

He considered the weight $w(\pi)$ for each partition $\pi$ of the set $A=\{0,\cdots,n-1\}$, and distinguished the part containing the zero element and called it $B_0$. Then,

$$w(\pi)=q^{\sum_{i\in B_0} i}$$ For each set $A$ of partitions, let $w(A)=\sum _{\pi\in A}w(\pi)$. Also, he considered $A_{n,k}$ to be the set of all partitions of $\{0,\cdots,n-1\}$ into $k$ non-empty parts. So $${n\brace k}=w(A_{n,k})$$

My question is:

What Mathematica code would generate these Stirling numbers of the second kind?

$\endgroup$
6
  • 2
    $\begingroup$ At least for the usual Stirling numbers of the second kind, Mathematica has these implemented as StirlingS2, see here $\endgroup$
    – Hausdorff
    Jan 16, 2022 at 11:08
  • $\begingroup$ Yes, I know it. Also Mathematica has a code fo q-binomial $\endgroup$
    – d.y
    Jan 16, 2022 at 13:37
  • 1
    $\begingroup$ What have you tried? $\endgroup$
    – MarcoB
    Jan 16, 2022 at 13:49
  • $\begingroup$ Welcome to Mathematica.SE! I hope you will become a regular contributor. To get started, 1) take the introductory tour now, 2) when you see good questions and answers, vote them up by clicking the gray triangles, because the credibility of the system is based on the reputation gained by users sharing their knowledge, 3) remember to accept the answer, if any, that solves your problem, by clicking the checkmark sign, and 4) give help too, by answering questions in your areas of expertise. $\endgroup$
    – bbgodfrey
    Jan 16, 2022 at 15:27
  • $\begingroup$ @Hausdorff Thank you so much. But , you write this code based on definition of Matthieu Josuat-Verges paper. our definition for my new problems is according to the Cigler definition. It is possible write the Mathematica code of q-stirling numbers according Ciqler definitions? Thanks in advance $\endgroup$
    – d.y
    Jan 18, 2022 at 12:40

1 Answer 1

7
$\begingroup$

On second thought, it is much more convenient to use Theorem 4.5 of the paper as a definition, so

A[n_, k_, i_, j_] := Binomial[n, k + i] Binomial[n, k - j] - 
  Binomial[n, k + i + 1] Binomial[n, k - j - 1];

QStirlingS2Fast[n_, k_, q_] := 
  1/(1 - q)^(n - k) Sum[(-1)^i*
  A[n, k, i, j] q^Binomial[j + 1, 2] QBinomial[i, j, q], 
  {j, 0,k}, {i, j, n - k}];

This is of course much faster than the implementation via partitions

QStirlingS2[10, 5, 3] // RepeatedTiming
(* {2.00612, 3421737} *)

QStirlingS2Fast[10, 5, 3] // RepeatedTiming
(* {0.000111509, 3421737} *)

and works for large inputs as well

QStirlingS2Fast[1000, 20, 3] // N // RepeatedTiming  
(* {0.8409685, 1.465117*^8966} *)

Original post

Here is one possible implementation, which is a bit brute force. I am using Combinatorica` to obtain the set partitions, as well as the definition of the q-Stirling number of the second kind via the number of partition crossings $cr(\pi)$ given in the linked paper.

<< Combinatorica`

cr[pi_] := Cases[
  Flatten[
    Tuples /@ Map[
      Partition[#, 2, 1] &, 
      Subsets[Replace[pi, {_} :> Nothing, 1], {2}], 
    {2}], 
  1],
  {{i_, j_}, {k_, l_}} /; i < k < j < l || k < i < l < j
];


QStirlingS2[n_, k_, q_] := Sum[q^Length@cr[pi], {pi, KSetPartitions[Range@n, k]}]

This has the behavior for $q\to0$ given in eq.$(13)$,

Narayana[n_, k_] := 1/n*Binomial[n, k - 1] Binomial[n, k];
{ QStirlingS2[10, 4, q] /. q -> 0, Narayana[10, 4] }
(* {2520, 2520} *)

and also satisfies the identity of Theorem 4.5 in your reference

A[n_, k_, i_, j_] := Binomial[n, k + i] Binomial[n, k - j] - 
  Binomial[n, k + i + 1] Binomial[n, k - j - 1];

Theorem45[n_, k_, q_] := 
{
  QStirlingS2[n, k, q] (1 - q)^(n - k),
  Sum[(-1)^i*A[n, k, i, j] q^Binomial[j + 1, 2] QBinomial[i, j, q], 
  {j, 0, k}, {i, j, n - k}] 
}

Theorem45[6, 2, 3]
(* {1872, 1872} *)
$\endgroup$

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.