Conjecture on log-concavity and unimodality of involution statistics

From papers

For n4n\geq 4, let Inmaj(q)\mathcal{I}_n^{\mathrm{maj}}(q) and Ininv(q)\mathcal{I}_n^{\mathrm{inv}}(q) denote the major-index and inversion generating polynomials over involutions, and let InvnB(q)Inv_n^B(q), InvnD(q)Inv_n^D(q), and InvnO(q)Inv_n^O(q) be the corresponding inversion polynomials for the types named in the source. For a polynomial P(q)P(q), write [qi]P(q)[q^i]P(q) for the coefficient of qiq^i.

The conjecture. For all n4n\geq 4:

  1. The sequence {[qi]Inmaj(q)}i=0(n2)\{[q^i]\mathcal{I}_n^{\mathrm{maj}}(q)\}_{i=0}^{\binom{n}{2}} is log-concave.
  2. For 2i(n2)22\leq i\leq \binom{n}{2}-2,
([qi]Ininv(q))2([qi2]Ininv(q))([qi+2]Ininv(q)).\bigl([q^i]\mathcal{I}_n^{\mathrm{inv}}(q)\bigr)^2\geq\bigl([q^{i-2}]\mathcal{I}_n^{\mathrm{inv}}(q)\bigr)\bigl([q^{i+2}]\mathcal{I}_n^{\mathrm{inv}}(q)\bigr).
  1. The sequences {[q2i]InvnB(q)}i0\{[q^{2i}]Inv_n^B(q)\}_{i\geq0} and {[q2i+1]InvnB(q)}i0\{[q^{2i+1}]Inv_n^B(q)\}_{i\geq0} are unimodal.
  2. The sequences {[q2i]InvnD(q)}i0\{[q^{2i}]Inv_n^D(q)\}_{i\geq0} and {[q2i+1]InvnD(q)}i0\{[q^{2i+1}]Inv_n^D(q)\}_{i\geq0} are unimodal.
  3. The sequences {[q2i]InvnO(q)}i0\{[q^{2i}]Inv_n^O(q)\}_{i\geq0} and {[q2i+1]InvnO(q)}i0\{[q^{2i+1}]Inv_n^O(q)\}_{i\geq0} are unimodal.

The conjecture extends the original log-concavity conjecture. The supplied passage reports numerical log-concavity through n14n\leq14 and notes that the relevant roots are not real and do not lie in the stated triangular region, but gives no resolution of these assertions.

Progress summary

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Sources & referencesView supporting material

Primary source

W. M. B. Dukes, “Permutation statistics on involutions”, arXiv:math/0412222 (2004).

Solutions 0

No solutions have been posted yet.