Additional enumerations for 321-avoiding Fishburn permutations

From papers

Let Fn(σ1,,σk)F_n(\sigma_1,\ldots,\sigma_k) denote Fishburn permutations of length nn avoiding each listed pattern, and let (Fn)(F_n) be the Fibonacci sequence under the paper's indexing convention.

Additional enumeration conjectures. The following identities are conjectured:

Fn(321,1243)=Fn(321,2134)=n23n+4(n2),|F_n(321,1243)|=|F_n(321,2134)|=n^2-3n+4\quad(n\geq 2), Fn(321,1324)=32n2132n+10(n3),|F_n(321,1324)|=\frac{3}{2}n^2-\frac{13}{2}n+10\quad(n\geq 3), Fn(321,3142,2143)=Fn(321,1423,2143)=Fn(321,2143,3124)=Fn(321,2143,4123)=(n2)+1(n0),|F_n(321,3142,2143)|=|F_n(321,1423,2143)|=|F_n(321,2143,3124)|=|F_n(321,2143,4123)|=\binom{n}{2}+1\quad(n\geq 0), Fn(321,1423,3124)=Fn+2(n4),|F_n(321,1423,3124)|=F_n+2\quad(n\geq 4), Fn(321,1423,4123)=Fn(321,3124,4123)=Fn+11(n1),|F_n(321,1423,4123)|=|F_n(321,3124,4123)|=F_{n+1}-1\quad(n\geq 1), Fn(321,14253)=Fn(321,21354)=2n(n2)1(n1).|F_n(321,14253)|=|F_n(321,21354)|=2^n-\binom{n}{2}-1\quad(n\geq 1).

The source says all these identities were verified for n20n\leq 20; no general proofs are given.

Progress summary

Solved

A 2023 paper gives general proofs of all the enumerations, so the conjectures are now settled.

Eric S. Egge posed these identities as part of his conjectures on pattern-avoiding Fishburn permutations; the original work reported verification only through n20n\leq 20.

Known results

Earlier work proved related formulas such as Fn(321,1423)=Fn(321,3124)=Fn+2n1|F_n(321,1423)|=|F_n(321,3124)|=F_{n+2}-n-1 and Fn(321,2143)=2n1|F_n(321,2143)|=2^{n-1}, while leaving the additional identities conjectural.

March 2023 general proof

Yujie Du and Philip B. Zhang’s paper proves Egge’s Conjectures 10.14 and 10.17. Its theorem sequence establishes the quadratic, (n2)+1\binom{n}{2}+1, Fibonacci, and 2n(n2)12^n-\binom{n}{2}-1 formulas listed here for all stated ranges, replacing finite verification with general proofs.

Current status (as of August 2026): All identities in the problem are proved by Du and Zhang; no part of this conjectural list remains open.

Sources
Sources & referencesView supporting material

Primary source

Eric S. Egge, “Pattern-Avoiding Fishburn Permutations and Ascent Sequences”, arXiv:2208.01484 (2022).

Solutions 1

Proof

This conjecture has been proved in later literature.

The identities in this MathDB entry are exactly Conjecture 10.17 of Eric S. Egge:

https://arxiv.org/abs/2208.01484

Yujie Du and Philip B. Zhang prove that conjecture in arXiv:2302.13767v3:

https://arxiv.org/abs/2302.13767v3

The journal version is Yujie Du and Philip B. Zhang, “On enumeration of pattern-avoiding Fishburn permutations,” Discrete Mathematics 347 (2024), Article 113952:

https://doi.org/10.1016/j.disc.2024.113952

The exact correspondence is:

  • Theorem 2.1 proves the formula for Fn(321,1243)F_n(321,1243).
  • Theorem 2.4 proves the formula for Fn(321,2134)F_n(321,2134).
  • Theorem 2.6 proves the formula for Fn(321,1324)F_n(321,1324).
  • Theorems 3.1, 3.4, 3.5, and 3.7 prove the four formulas equal to (n2)+1\binom{n}{2}+1.
  • Theorem 3.9 proves the formula Fn(321,1423,3124)=Fn+2|F_n(321,1423,3124)|=F_n+2.
  • Theorems 3.12 and 3.15 prove the two formulas equal to Fn+11F_{n+1}-1.
  • Theorems 4.1 and 4.3 prove the two formulas equal to 2n(n2)12^n-\binom{n}{2}-1.

The common structural starting point is Lemma 1.1 of that paper: if πFn(321)\pi\in F_n(321), then the entry 11 is in the first or second position. Indeed, suppose it occurs later and write x=π1x=\pi_1, y=π2y=\pi_2. If x>yx>y, then x,y,1x,y,1 form a 321-pattern. If x<yx<y, then x1x-1 occurs after yy, so x,y,x1x,y,x-1 form the forbidden Fishburn bivincular pattern. Both alternatives are impossible.

The paper then splits every avoidance class according to whether π1=1\pi_1=1 or π2=1\pi_2=1, classifies the resulting normal forms, and sums their cardinalities. The theorem numbers above give the resulting all-nn enumerations.

Thus all twelve enumerative statements, equivalently all identities (38)–(43) of Egge’s Conjecture 10.17, are proved. This MathDB entry should therefore be marked solved.

0 endorsements
Samuel Schlesinger ·