AFTD Auto-Formalizing Theoretical Domains

latin_tableaux_count_mono_three_two_one

theoremverified

The number of Latin tableaux of shape (3, 2, 1) is at most the number of Latin tableaux of shape (3, 3, 1), verifying Conjecture 6.1 for the addition of a box to the middle row of (3, 2, 1).

Statement

theorem latin_tableaux_count_mono_three_two_one :
    (latin_tableaux (YoungDiagram.ofRowLens [3, 2, 1] (by decide))).card ≤
    (latin_tableaux (YoungDiagram.ofRowLens [3, 3, 1] (by decide))).card
Source
arXiv:2610.29291 Conjecture 6.1
Verified
24 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 1
  • latin_tableauxFor a Young diagram μ, the set of Latin tableaux of shape μ and type μ' is the finset of fillings T from the cells of μ to natural numbers…

Read back from the Lean

The number of latin tableaux of shape [3, 2, 1] is less than or equal to the number of latin tableaux of shape [3, 3, 1].

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

The proof, explained

The proof establishes monotonicity by sandwiching both counts around 1: the number of Latin tableaux of shape (3, 2, 1) is at most 1, and the number of Latin tableaux of shape (3, 3, 1) is at least 1. For shape (3, 2, 1), the row bounds and column/row injectivity conditions force the entry in every cell uniquely, proving that any Latin tableau equals a single explicit tableau. For shape (3, 3, 1), an explicit candidate tableau is verified by exhaustive case analysis to satisfy the bound, row-injectivity, and column-injectivity conditions. Chaining the resulting inequalities yields the desired comparison.

  1. The Young diagram of partition (3, 2, 1) consists of six cells (0,0), (0,1), (0,2), (1,0), (1,1), and (2,0); its row lengths are 3, 2, and 1; and any Latin tableau T of this shape satisfies row bounds and injectivity along rows and columns.

    We define each of the six cells of the Young diagram of shape (3, 2, 1) and verify cell membership from row and column indices. From the definition of `latin_tableaux`, we unpack the hypothesis that T is a Latin tableau into value bounds (each cell (i, j) receives a value in [1, rowLen(i)]) and injectivity along each row and column. Finally, we compute the lengths of rows 0, 1, and 2 as 3, 2, and 1, respectively.

    lines 16–31

    uses YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens, Finset.mem_filter.mp, Fintype.mem_piFinset, YoungDiagram.rowLen_ofRowLens

      let c00 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(0,0), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩
      let c01 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(0,1), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩
      let c02 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(0,2), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩
      let c10 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(1,0), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩
      let c11 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(1,1), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩
      let c20 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(2,0), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩
      have hbounds := (Finset.mem_filter.mp hT).1
      rw [Fintype.mem_piFinset] at hbounds
      have hrow := (Finset.mem_filter.mp hT).2.1
      have hcol := (Finset.mem_filter.mp hT).2.2
      have hrl0 : (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 = 3 := by
        have := YoungDiagram.rowLen_ofRowLens (w := [3,2,1]) (hw := hw) ⟨0, by decide⟩; simpa using this
      have hrl1 : (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 1 = 2 := by
        have := YoungDiagram.rowLen_ofRowLens (w := [3,2,1]) (hw := hw) ⟨1, by decide⟩; simpa using this
      have hrl2 : (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 2 = 1 := by
        have := YoungDiagram.rowLen_ofRowLens (w := [3,2,1]) (hw := hw) ⟨2, by decide⟩; simpa using this
  2. In any Latin tableau T of shape (3, 2, 1), the first column is uniquely determined: T(2, 0) = 1, T(1, 0) = 2, and T(0, 0) = 3.

    For cell (2, 0) in row 2 of length 1, the bound 1 ≤ T(2, 0) ≤ 1 forces T(2, 0) = 1. In row 1 of length 2, the bound restricts T(1, 0) to {1, 2}, but column injectivity with (2, 0) rules out 1, leaving T(1, 0) = 2. In row 0 of length 3, the bound restricts T(0, 0) to {1, 2, 3}, but column injectivity with (2, 0) and (1, 0) rules out 1 and 2, forcing T(0, 0) = 3.

    lines 32–60

    uses Finset.mem_Icc.mp

      have h20 : T c20 = 1 := by
        have hle : T c20 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 2 := (Finset.mem_Icc.mp (hbounds c20)).2
        have hge : 1 ≤ T c20 := (Finset.mem_Icc.mp (hbounds c20)).1
        rw [hrl2] at hle; omega
      have h10 : T c10 = 2 := by
        have hle : T c10 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 1 := (Finset.mem_Icc.mp (hbounds c10)).2
        have hge : 1 ≤ T c10 := (Finset.mem_Icc.mp (hbounds c10)).1
        rw [hrl1] at hle
        have hne : T c10 ≠ 1 := by
          intro h
          have hh := hcol c10 c20 (by rfl) (by rw [h, h20])
          simp only [c10, c20] at hh
          omega
        omega
      have h00v : T c00 = 3 := by
        have hle : T c00 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 := (Finset.mem_Icc.mp (hbounds c00)).2
        have hge : 1 ≤ T c00 := (Finset.mem_Icc.mp (hbounds c00)).1
        rw [hrl0] at hle
        have hne1 : T c00 ≠ 1 := by
          intro h
          have hh := hcol c00 c20 (by rfl) (by rw [h, h20])
          simp only [c00, c20] at hh
          omega
        have hne2 : T c00 ≠ 2 := by
          intro h
          have hh := hcol c00 c10 (by rfl) (by rw [h, h10])
          simp only [c00, c10] at hh
          omega
        omega
  3. The remaining entries of T are uniquely determined: T(1, 1) = 1, T(0, 1) = 2, and T(0, 2) = 1.

    In row 1 of length 2, T(1, 1) ∈ {1, 2}; row injectivity with (1, 0) (value 2) forces T(1, 1) = 1. In row 0 of length 3, T(0, 1) ∈ {1, 2, 3}; row injectivity with (0, 0) (value 3) and column injectivity with (1, 1) (value 1) force T(0, 1) = 2. Finally, T(0, 2) ∈ {1, 2, 3} shares row 0 with (0, 0) and (0, 1), forcing T(0, 2) = 1.

    lines 61–100

    uses Finset.mem_Icc.mp

      have h11v : T c11 = 1 := by
        have hle : T c11 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 1 := (Finset.mem_Icc.mp (hbounds c11)).2
        have hge : 1 ≤ T c11 := (Finset.mem_Icc.mp (hbounds c11)).1
        rw [hrl1] at hle
        have hne : T c11 ≠ 2 := by
          intro h
          have hh := hrow c11 c10 (by rfl) (by rw [h, h10])
          simp only [c11, c10] at hh
          omega
        omega
      have h01v : T c01 = 2 := by
        have hle : T c01 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 := (Finset.mem_Icc.mp (hbounds c01)).2
        have hge : 1 ≤ T c01 := (Finset.mem_Icc.mp (hbounds c01)).1
        rw [hrl0] at hle
        have hne3 : T c01 ≠ 3 := by
          intro h
          have hh := hrow c01 c00 (by rfl) (by rw [h, h00v])
          simp only [c01, c00] at hh
          omega
        have hne1 : T c01 ≠ 1 := by
          intro h
          have hh := hcol c01 c11 (by rfl) (by rw [h, h11v])
          simp only [c01, c11] at hh
          omega
        omega
      have h02v : T c02 = 1 := by
        have hle : T c02 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 := (Finset.mem_Icc.mp (hbounds c02)).2
        have hge : 1 ≤ T c02 := (Finset.mem_Icc.mp (hbounds c02)).1
        rw [hrl0] at hle
        have hne3 : T c02 ≠ 3 := by
          intro h
          have hh := hrow c02 c00 (by rfl) (by rw [h, h00v])
          simp only [c02, c00] at hh
          omega
        have hne2 : T c02 ≠ 2 := by
          intro h
          have hh := hrow c02 c01 (by rfl) (by rw [h, h01v])
          simp only [c02, c01] at hh
          omega
        omega
  4. Any Latin tableau T of shape (3, 2, 1) equals the candidate tableau `latin_tableaux_count_mono_three_two_one_tab321`.

    By functional extensionality, we compare T and `tab321` on an arbitrary cell c = (i, j). An exhaustive case analysis over row indices i < 3 and column indices j < [3, 2, 1][i] shows that T(c) agrees with `tab321(c)` on all six cells.

    lines 101–114

    uses YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens, latin_tableaux_count_mono_three_two_one_tab321

      funext c
      rcases c with ⟨⟨i,j⟩, hc⟩
      have hi : i < 3 := by
        have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1
      have hj : j < [3,2,1][i] := by
        have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2
      interval_cases i
      · have hj' : j < 3 := by simpa using hj
        interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab321, c00, c01, c02, h00v, h01v, h02v]
      · have hj' : j < 2 := by simpa using hj
        interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab321, c10, c11, h10, h11v]
      · have hj' : j < 1 := by simpa using hj
        interval_cases j
        simp [latin_tableaux_count_mono_three_two_one_tab321, c20, h20]
  5. The row lengths of shape (3, 3, 1) are 3, 3, and 1, and the candidate tableau `tab331` satisfies the required entry bounds on every cell.

    We compute the row lengths of shape (3, 3, 1) using `YoungDiagram.rowLen_ofRowLens`. To show that `tab331` is a Latin tableau, we split the definition into bound, row-injectivity, and column-injectivity conditions. For the bounds, an exhaustive case analysis over all cells (i, j) confirms that tab331(i, j) lies in [1, rowLen(i)].

    lines 118–139

    uses YoungDiagram.rowLen_ofRowLens, Finset.mem_filter.mpr, Fintype.mem_piFinset, YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens, Finset.mem_Icc, latin_tableaux_count_mono_three_two_one_tab331

      have hrl0 : (YoungDiagram.ofRowLens [3,3,1] hw).rowLen 0 = 3 := by
        have := YoungDiagram.rowLen_ofRowLens (w := [3,3,1]) (hw := hw) ⟨0, by decide⟩; simpa using this
      have hrl1 : (YoungDiagram.ofRowLens [3,3,1] hw).rowLen 1 = 3 := by
        have := YoungDiagram.rowLen_ofRowLens (w := [3,3,1]) (hw := hw) ⟨1, by decide⟩; simpa using this
      have hrl2 : (YoungDiagram.ofRowLens [3,3,1] hw).rowLen 2 = 1 := by
        have := YoungDiagram.rowLen_ofRowLens (w := [3,3,1]) (hw := hw) ⟨2, by decide⟩; simpa using this
      refine Finset.mem_filter.mpr ⟨?_, ?_, ?_⟩
      · rw [Fintype.mem_piFinset]
        intro c
        rcases c with ⟨⟨i,j⟩, hc⟩
        have hi : i < 3 := by
          have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1
        have hj : j < [3,3,1][i] := by
          have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2
        rw [Finset.mem_Icc]
        interval_cases i
        · have hj' : j < 3 := by simpa using hj
          interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab331, hrl0]
        · have hj' : j < 3 := by simpa using hj
          interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab331, hrl1]
        · have hj' : j < 1 := by simpa using hj
          interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab331, hrl2]
  6. The candidate tableau `tab331` has strictly distinct entries along each row. automation

    For any two cells in the same row with identical values under `tab331`, exhaustive case analysis over all pairs of column indices in row 0, row 1, and row 2 confirms they must be the same cell.

    lines 140–161

    uses YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens, latin_tableaux_count_mono_three_two_one_tab331

      · intro c₁ c₂ hrow hval
        rcases c₁ with ⟨⟨i₁,j₁⟩, hc₁⟩
        rcases c₂ with ⟨⟨i₂,j₂⟩, hc₂⟩
        have hi₁ : i₁ < 3 := by
          have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1
        have hj₁ : j₁ < [3,3,1][i₁] := by
          have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2
        have hi₂ : i₂ < 3 := by
          have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1
        have hj₂ : j₂ < [3,3,1][i₂] := by
          have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2
        subst hrow
        interval_cases i₁
        · have hj₁' : j₁ < 3 := by simpa using hj₁
          have hj₂' : j₂ < 3 := by simpa using hj₂
          interval_cases j₁ <;> interval_cases j₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]
        · have hj₁' : j₁ < 3 := by simpa using hj₁
          have hj₂' : j₂ < 3 := by simpa using hj₂
          interval_cases j₁ <;> interval_cases j₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]
        · have hj₁' : j₁ < 1 := by simpa using hj₁
          have hj₂' : j₂ < 1 := by simpa using hj₂
          interval_cases j₁ <;> interval_cases j₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]
  7. The candidate tableau `tab331` has strictly distinct entries along each column, so `tab331` is a valid Latin tableau of shape (3, 3, 1). automation

    For any two cells in the same column with identical values under `tab331`, exhaustive case analysis over all pairs of row indices confirms they must be the same cell. This completes the verification that `tab331` belongs to `latin_tableaux`.

    lines 162–178

    uses YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens, latin_tableaux_count_mono_three_two_one_tab331

      · intro c₁ c₂ hcol hval
        rcases c₁ with ⟨⟨i₁,j₁⟩, hc₁⟩
        rcases c₂ with ⟨⟨i₂,j₂⟩, hc₂⟩
        have hi₁ : i₁ < 3 := by
          have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1
        have hj₁ : j₁ < [3,3,1][i₁] := by
          have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2
        have hi₂ : i₂ < 3 := by
          have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1
        have hj₂ : j₂ < [3,3,1][i₂] := by
          have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2
        subst hcol
        have hb1 : j₁ < 3 := by
          have hle : [3,3,1][i₁] ≤ 3 := by
            interval_cases i₁ <;> simp
          omega
        interval_cases j₁ <;> interval_cases i₁ <;> interval_cases i₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]
  8. The partition lists [3, 2, 1] and [3, 3, 1] are sorted, and the number of Latin tableaux of shape (3, 2, 1) is at most 1.

    The sort conditions for the Young diagram constructions in the theorem statement (lines 182–183) and in the hypotheses (lines 184–185) are verified automatically by `decide`. Since every Latin tableau of shape (3, 2, 1) is equal to `tab321` by `latin_tableaux_count_mono_three_two_one_tab321_eq`, any two Latin tableaux T and S are equal. By `Finset.card_le_one`, the cardinality of this set is at most 1.

    lines 182–190

    uses Finset.card_le_one, latin_tableaux_count_mono_three_two_one_tab321_eq

        (latin_tableaux (YoungDiagram.ofRowLens [3, 2, 1] (by decide))).card ≤
        (latin_tableaux (YoungDiagram.ofRowLens [3, 3, 1] (by decide))).card := by
      have hw321 : [3,2,1].SortedGE := by decide
      have hw331 : [3,3,1].SortedGE := by decide
      have hle : (latin_tableaux (YoungDiagram.ofRowLens [3,2,1] hw321)).card ≤ 1 := by
        rw [Finset.card_le_one]
        intro T hT S hS
        rw [latin_tableaux_count_mono_three_two_one_tab321_eq hw321 T hT,
            latin_tableaux_count_mono_three_two_one_tab321_eq hw321 S hS]
  9. The number of Latin tableaux of shape (3, 3, 1) is at least 1.

    Because `tab331` is a verified member of `latin_tableaux` for shape (3, 3, 1) by `latin_tableaux_count_mono_three_two_one_tab331_mem`, the set of tableaux is nonempty, and by `Finset.card_pos.mpr` its cardinality is at least 1.

    lines 191–193

    uses Finset.card_pos.mpr, latin_tableaux_count_mono_three_two_one_tab331, latin_tableaux_count_mono_three_two_one_tab331_mem

      have hpos : 1 ≤ (latin_tableaux (YoungDiagram.ofRowLens [3,3,1] hw331)).card :=
        Finset.card_pos.mpr ⟨latin_tableaux_count_mono_three_two_one_tab331 hw331,
          latin_tableaux_count_mono_three_two_one_tab331_mem hw331⟩
  10. The number of Latin tableaux of shape (3, 2, 1) is at most the number of Latin tableaux of shape (3, 3, 1).

    Combining the two bounds through a transitivity calculation chain yields |latin_tableaux(3, 2, 1)| ≤ 1 ≤ |latin_tableaux(3, 3, 1)|.

    lines 194–198

      calc (latin_tableaux (YoungDiagram.ofRowLens [3, 2, 1] (by decide))).card
          = (latin_tableaux (YoungDiagram.ofRowLens [3,2,1] hw321)).card := rfl
        _ ≤ 1 := hle
        _ ≤ (latin_tableaux (YoungDiagram.ofRowLens [3,3,1] hw331)).card := hpos
        _ = (latin_tableaux (YoungDiagram.ofRowLens [3, 3, 1] (by decide))).card := rfl
YoungDiagram.mem_cells
A cell (represented as a row-column pair) belongs to `μ.cells` if and only if it belongs to the Young diagram `μ`.
YoungDiagram.mem_ofRowLens
A cell (i, j) belongs to the Young diagram defined by row lengths `w` if and only if row index i is within bounds and column index j is strictly less than the length of row i.
Finset.mem_filter.mp
If an element belongs to a filtered subfinset, it belongs to the underlying finset and satisfies the filtering predicate.
Finset.mem_filter.mpr
If an element belongs to a finset and satisfies a predicate, it belongs to the filtered subfinset.
Fintype.mem_piFinset
A function belongs to the Cartesian product finset over a finite domain if and only if its value at every point belongs to the specified target finset.
YoungDiagram.rowLen_ofRowLens
In a Young diagram constructed from a list of row lengths `w`, the length of row i equals the i-th element of `w`.
Finset.mem_Icc
An integer belongs to the closed interval [a, b] if and only if it is at least a and at most b.
Finset.mem_Icc.mp
If an integer belongs to the closed interval [a, b], then it is at least a and at most b.
Finset.card_le_one
A finite set has cardinality at most 1 if and only if any two of its elements are identical.
Finset.card_pos.mpr
A finite set has strictly positive cardinality if and only if it is nonempty.
latin_tableaux_count_mono_three_two_one_tab321
The explicit candidate tableau on shape (3, 2, 1) mapping cells (0,0)↦3, (0,1)↦2, (0,2)↦1, (1,0)↦2, (1,1)↦1, (2,0)↦1.
latin_tableaux_count_mono_three_two_one_tab321_eq
Every Latin tableau of shape (3, 2, 1) is identical to the explicit candidate tableau `tab321`.
latin_tableaux_count_mono_three_two_one_tab331
The explicit candidate tableau on shape (3, 3, 1) mapping cells (0,0)↦2, (0,1)↦1, (0,2)↦3, (1,0)↦3, (1,1)↦2, (1,2)↦1, (2,0)↦1.
latin_tableaux_count_mono_three_two_one_tab331_mem
The candidate tableau `tab331` is a valid Latin tableau of shape (3, 3, 1).

Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.

Lean source view module on GitHub

def latin_tableaux_count_mono_three_two_one_tab321 (hw : [3,2,1].SortedGE)    (c : (YoungDiagram.ofRowLens [3,2,1] hw).cells) : ℕ :=  if c.1.1 = 0 then (if c.1.2 = 0 then 3 else if c.1.2 = 1 then 2 else 1)  else if c.1.1 = 1 then (if c.1.2 = 0 then 2 else 1) else 1def latin_tableaux_count_mono_three_two_one_tab331 (hw : [3,3,1].SortedGE)    (c : (YoungDiagram.ofRowLens [3,3,1] hw).cells) : ℕ :=  if c.1.1 = 0 then (if c.1.2 = 0 then 2 else if c.1.2 = 1 then 1 else 3)  else if c.1.1 = 1 then (if c.1.2 = 0 then 3 else if c.1.2 = 1 then 2 else 1)  else 1theorem latin_tableaux_count_mono_three_two_one_tab321_eq (hw : [3,2,1].SortedGE)    (T : (YoungDiagram.ofRowLens [3,2,1] hw).cells → ℕ)    (hT : T ∈ latin_tableaux (YoungDiagram.ofRowLens [3,2,1] hw)) :    T = latin_tableaux_count_mono_three_two_one_tab321 hw := by  let c00 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(0,0), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩  let c01 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(0,1), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩  let c02 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(0,2), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩  let c10 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(1,0), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩  let c11 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(1,1), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩  let c20 : (YoungDiagram.ofRowLens [3,2,1] hw).cells := ⟨(2,0), by rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens]; exact ⟨by decide, by decide⟩⟩  have hbounds := (Finset.mem_filter.mp hT).1  rw [Fintype.mem_piFinset] at hbounds  have hrow := (Finset.mem_filter.mp hT).2.1  have hcol := (Finset.mem_filter.mp hT).2.2  have hrl0 : (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 = 3 := by    have := YoungDiagram.rowLen_ofRowLens (w := [3,2,1]) (hw := hw) ⟨0, by decide⟩; simpa using this  have hrl1 : (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 1 = 2 := by    have := YoungDiagram.rowLen_ofRowLens (w := [3,2,1]) (hw := hw) ⟨1, by decide⟩; simpa using this  have hrl2 : (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 2 = 1 := by    have := YoungDiagram.rowLen_ofRowLens (w := [3,2,1]) (hw := hw) ⟨2, by decide⟩; simpa using this  have h20 : T c20 = 1 := by    have hle : T c20 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 2 := (Finset.mem_Icc.mp (hbounds c20)).2    have hge : 1 ≤ T c20 := (Finset.mem_Icc.mp (hbounds c20)).1    rw [hrl2] at hle; omega  have h10 : T c10 = 2 := by    have hle : T c10 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 1 := (Finset.mem_Icc.mp (hbounds c10)).2    have hge : 1 ≤ T c10 := (Finset.mem_Icc.mp (hbounds c10)).1    rw [hrl1] at hle    have hne : T c10 ≠ 1 := by      intro h      have hh := hcol c10 c20 (by rfl) (by rw [h, h20])      simp only [c10, c20] at hh      omega    omega  have h00v : T c00 = 3 := by    have hle : T c00 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 := (Finset.mem_Icc.mp (hbounds c00)).2    have hge : 1 ≤ T c00 := (Finset.mem_Icc.mp (hbounds c00)).1    rw [hrl0] at hle    have hne1 : T c00 ≠ 1 := by      intro h      have hh := hcol c00 c20 (by rfl) (by rw [h, h20])      simp only [c00, c20] at hh      omega    have hne2 : T c00 ≠ 2 := by      intro h      have hh := hcol c00 c10 (by rfl) (by rw [h, h10])      simp only [c00, c10] at hh      omega    omega  have h11v : T c11 = 1 := by    have hle : T c11 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 1 := (Finset.mem_Icc.mp (hbounds c11)).2    have hge : 1 ≤ T c11 := (Finset.mem_Icc.mp (hbounds c11)).1    rw [hrl1] at hle    have hne : T c11 ≠ 2 := by      intro h      have hh := hrow c11 c10 (by rfl) (by rw [h, h10])      simp only [c11, c10] at hh      omega    omega  have h01v : T c01 = 2 := by    have hle : T c01 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 := (Finset.mem_Icc.mp (hbounds c01)).2    have hge : 1 ≤ T c01 := (Finset.mem_Icc.mp (hbounds c01)).1    rw [hrl0] at hle    have hne3 : T c01 ≠ 3 := by      intro h      have hh := hrow c01 c00 (by rfl) (by rw [h, h00v])      simp only [c01, c00] at hh      omega    have hne1 : T c01 ≠ 1 := by      intro h      have hh := hcol c01 c11 (by rfl) (by rw [h, h11v])      simp only [c01, c11] at hh      omega    omega  have h02v : T c02 = 1 := by    have hle : T c02 ≤ (YoungDiagram.ofRowLens [3,2,1] hw).rowLen 0 := (Finset.mem_Icc.mp (hbounds c02)).2    have hge : 1 ≤ T c02 := (Finset.mem_Icc.mp (hbounds c02)).1    rw [hrl0] at hle    have hne3 : T c02 ≠ 3 := by      intro h      have hh := hrow c02 c00 (by rfl) (by rw [h, h00v])      simp only [c02, c00] at hh      omega    have hne2 : T c02 ≠ 2 := by      intro h      have hh := hrow c02 c01 (by rfl) (by rw [h, h01v])      simp only [c02, c01] at hh      omega    omega  funext c  rcases c with ⟨⟨i,j⟩, hc⟩  have hi : i < 3 := by    have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1  have hj : j < [3,2,1][i] := by    have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2  interval_cases i  · have hj' : j < 3 := by simpa using hj    interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab321, c00, c01, c02, h00v, h01v, h02v]  · have hj' : j < 2 := by simpa using hj    interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab321, c10, c11, h10, h11v]  · have hj' : j < 1 := by simpa using hj    interval_cases j    simp [latin_tableaux_count_mono_three_two_one_tab321, c20, h20]theorem latin_tableaux_count_mono_three_two_one_tab331_mem (hw : [3,3,1].SortedGE) :    latin_tableaux_count_mono_three_two_one_tab331 hw ∈ latin_tableaux (YoungDiagram.ofRowLens [3,3,1] hw) := by  have hrl0 : (YoungDiagram.ofRowLens [3,3,1] hw).rowLen 0 = 3 := by    have := YoungDiagram.rowLen_ofRowLens (w := [3,3,1]) (hw := hw) ⟨0, by decide⟩; simpa using this  have hrl1 : (YoungDiagram.ofRowLens [3,3,1] hw).rowLen 1 = 3 := by    have := YoungDiagram.rowLen_ofRowLens (w := [3,3,1]) (hw := hw) ⟨1, by decide⟩; simpa using this  have hrl2 : (YoungDiagram.ofRowLens [3,3,1] hw).rowLen 2 = 1 := by    have := YoungDiagram.rowLen_ofRowLens (w := [3,3,1]) (hw := hw) ⟨2, by decide⟩; simpa using this  refine Finset.mem_filter.mpr ⟨?_, ?_, ?_⟩  · rw [Fintype.mem_piFinset]    intro c    rcases c with ⟨⟨i,j⟩, hc⟩    have hi : i < 3 := by      have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1    have hj : j < [3,3,1][i] := by      have h := hc; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2    rw [Finset.mem_Icc]    interval_cases i    · have hj' : j < 3 := by simpa using hj      interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab331, hrl0]    · have hj' : j < 3 := by simpa using hj      interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab331, hrl1]    · have hj' : j < 1 := by simpa using hj      interval_cases j <;> simp [latin_tableaux_count_mono_three_two_one_tab331, hrl2]  · intro c₁ c₂ hrow hval    rcases c₁ with ⟨⟨i₁,j₁⟩, hc₁⟩    rcases c₂ with ⟨⟨i₂,j₂⟩, hc₂⟩    have hi₁ : i₁ < 3 := by      have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1    have hj₁ : j₁ < [3,3,1][i₁] := by      have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2    have hi₂ : i₂ < 3 := by      have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1    have hj₂ : j₂ < [3,3,1][i₂] := by      have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2    subst hrow    interval_cases i₁    · have hj₁' : j₁ < 3 := by simpa using hj₁      have hj₂' : j₂ < 3 := by simpa using hj₂      interval_cases j₁ <;> interval_cases j₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]    · have hj₁' : j₁ < 3 := by simpa using hj₁      have hj₂' : j₂ < 3 := by simpa using hj₂      interval_cases j₁ <;> interval_cases j₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]    · have hj₁' : j₁ < 1 := by simpa using hj₁      have hj₂' : j₂ < 1 := by simpa using hj₂      interval_cases j₁ <;> interval_cases j₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]  · intro c₁ c₂ hcol hval    rcases c₁ with ⟨⟨i₁,j₁⟩, hc₁⟩    rcases c₂ with ⟨⟨i₂,j₂⟩, hc₂⟩    have hi₁ : i₁ < 3 := by      have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1    have hj₁ : j₁ < [3,3,1][i₁] := by      have h := hc₁; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2    have hi₂ : i₂ < 3 := by      have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.1    have hj₂ : j₂ < [3,3,1][i₂] := by      have h := hc₂; rw [YoungDiagram.mem_cells, YoungDiagram.mem_ofRowLens] at h; exact h.2    subst hcol    have hb1 : j₁ < 3 := by      have hle : [3,3,1][i₁] ≤ 3 := by        interval_cases i₁ <;> simp      omega    interval_cases j₁ <;> interval_cases i₁ <;> interval_cases i₂ <;> simp_all [latin_tableaux_count_mono_three_two_one_tab331]/-- Monotonicity of Latin tableaux count for the transition (3, 2, 1) to (3, 3, 1). -/theorem latin_tableaux_count_mono_three_two_one :    (latin_tableaux (YoungDiagram.ofRowLens [3, 2, 1] (by decide))).card ≤    (latin_tableaux (YoungDiagram.ofRowLens [3, 3, 1] (by decide))).card := by  have hw321 : [3,2,1].SortedGE := by decide  have hw331 : [3,3,1].SortedGE := by decide  have hle : (latin_tableaux (YoungDiagram.ofRowLens [3,2,1] hw321)).card ≤ 1 := by    rw [Finset.card_le_one]    intro T hT S hS    rw [latin_tableaux_count_mono_three_two_one_tab321_eq hw321 T hT,        latin_tableaux_count_mono_three_two_one_tab321_eq hw321 S hS]  have hpos : 1 ≤ (latin_tableaux (YoungDiagram.ofRowLens [3,3,1] hw331)).card :=    Finset.card_pos.mpr ⟨latin_tableaux_count_mono_three_two_one_tab331 hw331,      latin_tableaux_count_mono_three_two_one_tab331_mem hw331⟩  calc (latin_tableaux (YoungDiagram.ofRowLens [3, 2, 1] (by decide))).card      = (latin_tableaux (YoungDiagram.ofRowLens [3,2,1] hw321)).card := rfl    _ ≤ 1 := hle    _ ≤ (latin_tableaux (YoungDiagram.ofRowLens [3,3,1] hw331)).card := hpos    _ = (latin_tableaux (YoungDiagram.ofRowLens [3, 3, 1] (by decide))).card := rfl
See it in the graphDiscuss this resultChallenge it