OctoInform Beta

A Cyclic Polytabloid Proof of Saxl's Conjecture

For a partition α⊢n\alpha\vdash n, let SαS^\alpha be the Specht module of α\alpha, the irreducible complex representation of the symmetric group SnS_n indexed by α\alpha. For partitions α,β,λ⊢n\alpha,\beta,\lambda\vdash n, the Kronecker coefficient g(α,β,λ)=dim⁡Hom⁡Sn(Sλ,Sα⊗Sβ)g(\alpha,\beta,\lambda)=\dim\operatorname{Hom}_{S_n}(S^\lambda,S^\alpha\otimes S^\beta) is the multiplicity of SλS^\lambda in Sα⊗SβS^\alpha\otimes S^\beta, on which SnS_n acts diagonally. For m≥1m\ge1, let Nm=m(m+1)/2N_m=m(m+1)/2. The staircase ρm=(m,m−1,…,1)\rho_m=(m,m-1,\ldots,1) is a partition of NmN_m. Saxl’s conjecture asserts that g(ρm,ρm,λ)>0g(\rho_m,\rho_m,\lambda)>0 for every m≥1m\ge1 and every λ⊢Nm\lambda\vdash N_m, that is, that the tensor square Sρm⊗SρmS^{\rho_m}\otimes S^{\rho_m} contains every irreducible complex representation of SNmS_{N_m}.

Saxl proposed the conjecture in a UCLA Combinatorics Seminar on 20 March 2012, as recorded by Pak, Panova and Vallejo. They proved it for hooks λ\lambda and for two-row partitions λ\lambda, when mm is sufficiently large. Ikenmeyer proved it whenever λ\lambda and ρm\rho_m are comparable in the dominance order. The Durfee square of a partition is the largest square contained in its diagram. A double hook is a partition whose Durfee square has side 22. A triple hook is a partition whose Durfee square has side 33. Bessenrodt settled the case where λ\lambda is a double hook. Li settled the case where λ\lambda is a triple hook, with computer-assisted finite reductions. Bessenrodt, Bowman and Sutton showed that every SλS^\lambda of odd dimension occurs in Sρm⊗SρmS^{\rho_m}\otimes S^{\rho_m}. Luo and Sellke proved that a random partition of NmN_m, under the uniform or the Plancherel measure, indexes a constituent of Sρm⊗SρmS^{\rho_m}\otimes S^{\rho_m} with probability tending to one as m→∞m\to\infty. Harman and Ryba proved that the tensor cube Sρm⊗Sρm⊗SρmS^{\rho_m}\otimes S^{\rho_m}\otimes S^{\rho_m} contains every irreducible representation of SNmS_{N_m}.

Let Bm={(i,j):i,j≥1,i+j≤m+1}B_m=\{(i,j):i,j\ge1,\ i+j\le m+1\} be the diagram of ρm\rho_m. Let e1,…,eme_1,\ldots,e_m be the standard basis of ℂm\mathbb{C}^m. For 1≤r≤m1\le r\le m, let ar=∑π∈Srsgn⁡(π)eπ(1)⊗⋯⊗eπ(r)a_r=\sum_{\pi\in S_r}\operatorname{sgn}(\pi)\,e_{\pi(1)}\otimes\cdots\otimes e_{\pi(r)} be the alternating sum in (ℂm)⊗r(\mathbb{C}^m)^{\otimes r}. Let vRm∈(ℂm)⊗Bmv_{R_m}\in(\mathbb{C}^m)^{\otimes B_m} be the tensor product over the rows of BmB_m of the tensors ara_r, each placed along a row of length rr. Let vCmv_{C_m} be the same tensor product over the columns. Let SBmS_{B_m} be the symmetric group of the NmN_m cells of BmB_m. A numbering of these cells identifies SBmS_{B_m} with SNmS_{N_m}. The group SBmS_{B_m} acts on (ℂm)⊗Bm⊗(ℂm)⊗Bm(\mathbb{C}^m)^{\otimes B_m}\otimes(\mathbb{C}^m)^{\otimes B_m} by permuting the positions of both factors simultaneously. Let Wm=ℂ[SBm](vRm⊗vCm)W_m=\mathbb{C}[S_{B_m}](v_{R_m}\otimes v_{C_m}) be the cyclic module generated by vRm⊗vCmv_{R_m}\otimes v_{C_m} in (ℂm)⊗Bm⊗(ℂm)⊗Bm(\mathbb{C}^m)^{\otimes B_m}\otimes(\mathbb{C}^m)^{\otimes B_m}. The module WmW_m embeds into Sρm⊗SρmS^{\rho_m}\otimes S^{\rho_m} as a module over SBm≅SNmS_{B_m}\cong S_{N_m}. The preprint proves Saxl’s conjecture in the stronger form that SλS^\lambda occurs in WmW_m for every m≥1m\ge1 and every λ⊢Nm\lambda\vdash N_m. The companion preprint Universal Tensor Squares for Symmetric Groups uses this stronger form, together with further arguments, to construct for every positive integer n∉{2,4,9}n\notin\{2,4,9\} an irreducible complex representation of SnS_n whose tensor square contains every irreducible representation of SnS_n, a construction the Lean library does not contain.

The Lean library Saxl proves Saxl’s conjecture and its stronger form. Of the companion preprint Universal Tensor Squares for Symmetric Groups, the Lean library contains only Theorem 2.4, which is Theorem 3.1 of the preprint, and Lemmas 2.1, 2.2, 4.2 and 4.5. Lean has checked every result of the Lean library, and the Lean library is therefore the reference. Each result on these pages is stated as the Lean library states it and is linked to the Lean declaration behind it. Where a preprint states a result differently, its statement is given beside the Lean library’s. The preprints are guides only, giving the results their order and their names. The Lean library takes no result as a hypothesis. The Lean library proves the facts on Specht modules that the preprint cites from James. The SλS^\lambda with λ⊢n\lambda\vdash n are pairwise nonisomorphic. They exhaust the irreducible complex representations of SnS_n. Each SλS^\lambda is self-dual. Let λt\lambda^t be the conjugate partition of λ\lambda. Then Sλ⊗sgn⁡≅SλtS^\lambda\otimes\operatorname{sgn}\cong S^{\lambda^t}.

The Lean library: Saxl

Saxl is a Lean library: 28 items, in seven groups. Each has a page written by OctoInform, listed below.

§1. Introduction 2 items
  1. §1 (setting): Specht modules, Kronecker coefficients and the staircase Specht modules in tensor space and Kronecker coefficients. §1
  2. Theorem 1.1 [Saxl's conjecture] The staircase Specht module’s tensor square contains every irreducible. §1
§2. Specht modules and horizontal strips 6 items
  1. §2 (setting): dominance, position tensors and block tensors Dominance, word spaces and block tensors for symmetric groups. §2
  2. §2 (displayed): conjugation reverses dominance (2.1) Conjugation reverses dominance on diagrams of equal size. §2
  3. Lemma 2.1 [Alternating blocks as polytabloids] Polytabloids in a tensor power span irreducible Specht modules. §2
  4. §2 (displayed): classification, self-duality and the sign twist (2.3) Every finite-dimensional complex irreducible of SnS_n is a Specht module. §2
  5. Lemma 2.2 [Pieri rule] A horizontal strip λ/ν\lambda/\nu maps SλS^\lambda onto SνS^\nu. §2
  6. Lemma 2.3 [Young's rule: support and diagonal multiplicity] Dominance gives invariant functionals on Specht modules. §2
§3. The triangular cyclic module and dominance 3 items
  1. §3 (setting): the triangle Bₘ, its row and column tensors, and Wₘ (3.1) to (3.3) The staircase cyclic module embeds in a Specht tensor square. §3
  2. Theorem 3.1 [Cyclic staircase support] The staircase cyclic module contains every Specht module. §3
  3. Proposition 3.2 Partitions dominated by the staircase occur in its cyclic module. §3
§4. A quotient obtained by cutting off a band 3 items
  1. §4 (setting): the cut, the band generator and its cyclic module (4.1), (4.2) Cutting BhB_h off the staircase BmB_m and the band generator. §4
  2. Lemma 4.1 [Induction from coordinate sectors] When averaging a sector-supported map gives a nonzero intertwiner. §4
  3. Proposition 4.2 [Band quotient] A projection splits the staircase generator along a band. §4
§5. Cyclic support of the bands 5 items
  1. §5 (displayed): the width-one band (5.1) One-row Specht modules occur in the width-one band module. §5
  2. §5 (setting): the path numbering (5.2), the path generator u and (5.3) The width-two staircase band tensor along a zigzag path. §5
  3. Lemma 5.1 [Path contraction] Contracting the path generator gives a matrix entry. §5
  4. Lemma 5.2 [Segment alternation] Segment matrices are nonzero multiples of basis matrices. §5
  5. Proposition 5.3 [Four-row support of the width-two band] Four-row Specht modules occur in the width-two band. §5
§6. Completion of the induction 3 items
  1. Lemma 6.1 [Horizontal-strip reduction] Horizontal strips reduce partitions not dominated by the staircase. §6
  2. §6 (proof of Theorem 3.1): the induction Every Specht module occurs in the staircase cyclic module. §6
  3. §6 (proof of Theorem 1.1) Saxl’s conjecture follows from its cyclic form. §6
The companion preprint: Universal Tensor Squares for Symmetric Groups 6 items
  1. §2.3 (setting): the cyclic staircase input (2.3) The cyclic staircase module and its map to the tensor square. Universal Tensor Squares for Symmetric Groups, §2.3
  2. Lemma 2.1 [Dual-polytabloid test] A nonzero pairing gives a nonzero map into a cyclic module. Universal Tensor Squares for Symmetric Groups, §2
  3. Lemma 2.2 [Young support] Specht modules of dominating shapes occur in permutation modules. Universal Tensor Squares for Symmetric Groups, §2
  4. Theorem 2.4 [Cyclic staircase support, Theorem 3.1 of the preprint] Every Specht module of SNmS_{N_m} occurs in the staircase cyclic module. Universal Tensor Squares for Symmetric Groups, §2
  5. Lemma 4.2 [Induction from coordinate sectors] Nonzero sector-supported equivariant maps give nonzero intertwiners. Universal Tensor Squares for Symmetric Groups, §4
  6. Lemma 4.5 [Odd-path support] The odd-path contraction is a twisted matrix product. Universal Tensor Squares for Symmetric Groups, §4

No item matches.