latin_tableaux_count_mono_cell_addition_top_row
Let $d_1$ and $d_2$ be wide Young diagrams with the same number of rows $d_1.\text{colLen } 0 = d_2.\text{colLen } 0$. If $d_1.\text{rowLen } 0 = d_2.\text{rowLen } 0 + 1$ and $d_1.\text{rowLen } j = d_2.\text{rowLen } j$ for all $j \neq 0$ (a single cell is added to the top row), then $(latin\_tableaux\ d_2).card \le (latin\_tableaux\ d_1).card$.
Statement
theorem latin_tableaux_count_mono_cell_addition_top_row
(d₁ d₂ : YoungDiagram) (hwide₁ : is_wide_partition d₁) (hwide₂ : is_wide_partition d₂)
(hlen : d₁.colLen 0 = d₂.colLen 0)
(hrow_zero : d₁.rowLen 0 = d₂.rowLen 0 + 1)
(hrow_other : ∀ j, j ≠ 0 → d₁.rowLen j = d₂.rowLen j) :
(latin_tableaux d₂).card ≤ (latin_tableaux d₁).card- Formalizes
- arXiv:2610.29291 Proposition 5.5
- Verified
- 24 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 2
- 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…
- is_wide_partitionA Young diagram $d$ is wide if for every Young diagram $\mu$ whose list of row lengths $\mu.\text{rowLens}$ is a sublist of…
Read back from the Lean
For any two Young diagrams $d_1$ and $d_2$, assume:
1. $d_1$ is a "wide partition" (is_wide_partition d₁),
2. $d_2$ is a "wide partition" (is_wide_partition d₂),
3. the length of column 0 of $d_1$ equals that of $d_2$ (d₁.colLen 0 = d₂.colLen 0),
4. the length of row 0 of $d_1$ is one more than that of $d_2$ (d₁.rowLen 0 = d₂.rowLen 0 + 1), and
5. for all row indices $j \neq 0$, the length of row $j$ of $d_1$ equals that of $d_2$ (∀ j, j ≠ 0 → d₁.rowLen j = d₂.rowLen j).
Then the number of Latin tableaux of shape $d_2$ is less than or equal to the number of Latin tableaux of shape $d_1$ ((latin_tableaux d₂).card ≤ (latin_tableaux d₁).card).
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 theorem is proved by constructing an injective map from Latin tableaux of shape $d_2$ to Latin tableaux of shape $d_1$. First, a helper lemma establishes that every cell of $d_2$ is contained in $d_1$. Since $d_1$ is obtained from $d_2$ by adding a single cell $(0, d_2.\text{rowLen } 0)$ to the top row, any tableau $T$ on $d_2$ is extended by assigning the new cell the value $d_1.\text{rowLen } 0$. Because all entries of $T$ are bounded by $d_2.\text{rowLen } 0 < d_1.\text{rowLen } 0$, this maximum value cannot clash with any existing entry in row 0 or in the added column, ensuring the extended tableau is Latin. Restricting the extended tableau back to $d_2$ recovers $T$, proving injectivity and hence the cardinality inequality.
Every cell of $d_2$ is also a cell of $d_1$.
A cell $c = (i, j)$ belongs to a Young diagram if and only if $j < \text{rowLen } i$. If $c$ is in the top row ($i = 0$), then $j < d_2.\text{rowLen } 0 < d_2.\text{rowLen } 0 + 1 = d_1.\text{rowLen } 0$. If $i \neq 0$, the row lengths of $d_1$ and $d_2$ coincide, so $j < d_2.\text{rowLen } i = d_1.\text{rowLen } i$. Thus every cell of $d_2$ is a cell of $d_1$.
uses
YoungDiagram.mem_cells,YoungDiagram.mem_iff_lt_rowLenrw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc ⊢ by_cases h : c.1 = 0 · rw [h, hrow_zero] rw [h] at hc omega · rw [hrow_other c.1 h] exact hc
The inequality $|latin\_tableaux\ d_2| \le |latin\_tableaux\ d_1|$ reduces to showing that the tableau extension map $f$ maps $latin\_tableaux\ d_2$ into $latin\_tableaux\ d_1$ and is injective on it.
We define $f$ to extend any tableau $T$ on $d_2$ to a tableau on $d_1$ by preserving existing entries and assigning the top row's new length $d_1.\text{rowLen } 0$ to the added cell. By `Finset.card_le_card_of_injOn`, finding such an injection between two finite sets suffices to show that the cardinality of $latin\_tableaux\ d_2$ is bounded by that of $latin\_tableaux\ d_1$.
uses
Finset.card_le_card_of_injOnlet f := latin_tableaux_count_mono_cell_addition_top_row_extend d₁ d₂ apply Finset.card_le_card_of_injOn f
Showing that $f(T)$ is a Latin tableau of $d_1$ reduces to verifying that its entries satisfy the row-length bounds, that entries along each row are distinct, and that entries along each column are distinct.
Let $T$ be a Latin tableau of $d_2$. Unfolding the definition of `latin_tableaux`, a valid tableau on $d_1$ is a function on $d_1.\text{cells}$ such that every cell $(r, k)$ has value in $[1, d_1.\text{rowLen } r]$, no row contains duplicate entries, and no column contains duplicate entries. We split the goal into these three independent requirements.
uses
Finset.mem_coe,Finset.mem_filter· intro T hT rw [Finset.mem_coe] at hT ⊢ rw [latin_tableaux, Finset.mem_filter] refine ⟨?_, ?_, ?_⟩
Every cell $c = (r, k)$ of $d_1$ has its assigned entry $f(T)(c)$ in $[1, d_1.\text{rowLen } r]$.
We unpack the product-set condition and consider an arbitrary cell $c = (r, k) \in d_1$, splitting on whether $c \in d_2$. If $c \in d_2$, then $f(T)(c) = T(c)$, which lies in $[1, d_2.\text{rowLen } r]$ since $T$ is a Latin tableau; because $d_1.\text{rowLen } 0 = d_2.\text{rowLen } 0 + 1$ and other row lengths are equal, $d_2.\text{rowLen } r \le d_1.\text{rowLen } r$, so the bound holds. If $c \notin d_2$, then comparing the cell bounds in $d_1$ and $d_2$ reveals that $c$ must be the single new cell $(0, d_2.\text{rowLen } 0)$; its assigned value is $d_1.\text{rowLen } 0 = d_2.\text{rowLen } 0 + 1$, which is in $[1, d_1.\text{rowLen } 0]$ by linear arithmetic.
uses
Fintype.mem_piFinset,mem_Icc,YoungDiagram.mem_cells,YoungDiagram.mem_iff_lt_rowLen,Prod.ext,False.elim,congr_arg,Prod.fst· rw [Fintype.mem_piFinset] intro c rcases c with ⟨⟨r, k⟩, hc⟩ dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] split_ifs with h · simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT have hrange := hT.1 ⟨(r, k), h⟩ rw [mem_Icc] refine ⟨hrange.1, ?_⟩ have hle : d₂.rowLen r ≤ d₁.rowLen r := by by_cases hc0 : r = 0 · rw [hc0, hrow_zero] omega · rw [hrow_other r hc0] exact hrange.2.trans hle · rw [mem_Icc] have hc_not : (r, k) = (0, d₂.rowLen 0) := by have hc2 : (r, k) ∉ d₂.cells := h rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc2 have hc1 := hc rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc1 by_cases h0 : r = 0 · rw [h0, hrow_zero] at hc1 have hnot : ¬ (k < d₂.rowLen 0) := by rwa [h0] at hc2 have heq2 : k = d₂.rowLen 0 := by omega exact Prod.ext h0 heq2 · rw [hrow_other r h0] at hc1 exact False.elim (hc2 hc1) have hc0 : r = 0 := congr_arg Prod.fst hc_not have : d₁.rowLen r = d₁.rowLen 0 := by rw [hc0] rw [this, hrow_zero] omega
The extended tableau $f(T)$ has strictly distinct entries in each row.
Given two cells $c_1, c_2 \in d_1$ in the same row with $f(T)(c_1) = f(T)(c_2)$, we show their column coordinates coincide. First, because row lengths are non-increasing, every cell in $d_2$ has entry at most $d_2.\text{rowLen } 0 < d_1.\text{rowLen } 0$. Second, any cell of $d_1$ not in $d_2$ must be the newly added cell $(0, d_2.\text{rowLen } 0)$. We then case-split on whether $c_1, c_2 \in d_2$. If both lie in $d_2$, row uniqueness of $T$ applies. If one lies in $d_2$ and the other does not, their values cannot match because values in $d_2$ are strictly less than $d_1.\text{rowLen } 0$, a contradiction. If neither lies in $d_2$, both are the unique new cell, so their coordinates coincide.
uses
Nat.zero_le,YoungDiagram.mem_cells,YoungDiagram.mem_iff_lt_rowLen,Prod.ext,False.elim,dif_pos,dif_neg· intro c₁ c₂ hr heq have hT_val : ∀ c : d₂.cells, T c < d₁.rowLen 0 := by intro c simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT have hrange := hT.1 c have hanti : d₂.rowLen c.1.1 ≤ d₂.rowLen 0 := d₂.rowLen_anti 0 c.1.1 (Nat.zero_le _) rw [hrow_zero] omega have hc_not : ∀ c : d₁.cells, c.1 ∉ d₂.cells → c.1 = (0, d₂.rowLen 0) := by intro c hc2 rcases c with ⟨⟨r, k⟩, hc⟩ rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc2 have hc1 := hc rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc1 by_cases h0 : r = 0 · rw [h0, hrow_zero] at hc1 have hnot : ¬ (k < d₂.rowLen 0) := by rwa [h0] at hc2 have heq2 : k = d₂.rowLen 0 := by omega exact Prod.ext h0 heq2 · rw [hrow_other r h0] at hc1 exact False.elim (hc2 hc1) dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] at heq by_cases h1 : c₁.1 ∈ d₂.cells <;> by_cases h2 : c₂.1 ∈ d₂.cells · rw [dif_pos h1, dif_pos h2] at heq simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT exact hT.2.1 ⟨c₁.1, h1⟩ ⟨c₂.1, h2⟩ hr heq · rw [dif_pos h1, dif_neg h2] at heq have hlt := hT_val ⟨c₁.1, h1⟩ omega · rw [dif_neg h1, dif_pos h2] at heq have hlt := hT_val ⟨c₂.1, h2⟩ omega · have e1 := hc_not c₁ h1 have e2 := hc_not c₂ h2 rw [e1, e2]
The extended tableau $f(T)$ has strictly distinct entries in each column.
Given two cells $c_1, c_2 \in d_1$ in the same column with $f(T)(c_1) = f(T)(c_2)$, we show their row coordinates coincide. As in the row condition, all entries in $T$ are strictly less than $d_1.\text{rowLen } 0$, and any cell in $d_1 \setminus d_2$ is $(0, d_2.\text{rowLen } 0)$. Case-splitting on whether $c_1, c_2 \in d_2$: if both lie in $d_2$, column uniqueness of $T$ applies; mixed cases are impossible by value bounds; and if neither lies in $d_2$, both are the unique new cell, so their row coordinates coincide.
uses
Nat.zero_le,YoungDiagram.mem_cells,YoungDiagram.mem_iff_lt_rowLen,Prod.ext,False.elim,dif_pos,dif_neg· intro c₁ c₂ hc_col heq have hT_val : ∀ c : d₂.cells, T c < d₁.rowLen 0 := by intro c simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT have hrange := hT.1 c have hanti : d₂.rowLen c.1.1 ≤ d₂.rowLen 0 := d₂.rowLen_anti 0 c.1.1 (Nat.zero_le _) rw [hrow_zero] omega have hc_not : ∀ c : d₁.cells, c.1 ∉ d₂.cells → c.1 = (0, d₂.rowLen 0) := by intro c hc2 rcases c with ⟨⟨r, k⟩, hc⟩ rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc2 have hc1 := hc rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc1 by_cases h0 : r = 0 · rw [h0, hrow_zero] at hc1 have hnot : ¬ (k < d₂.rowLen 0) := by rwa [h0] at hc2 have heq2 : k = d₂.rowLen 0 := by omega exact Prod.ext h0 heq2 · rw [hrow_other r h0] at hc1 exact False.elim (hc2 hc1) dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] at heq by_cases h1 : c₁.1 ∈ d₂.cells <;> by_cases h2 : c₂.1 ∈ d₂.cells · rw [dif_pos h1, dif_pos h2] at heq simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT exact hT.2.2 ⟨c₁.1, h1⟩ ⟨c₂.1, h2⟩ hc_col heq · rw [dif_pos h1, dif_neg h2] at heq have hlt := hT_val ⟨c₁.1, h1⟩ omega · rw [dif_neg h1, dif_pos h2] at heq have hlt := hT_val ⟨c₂.1, h2⟩ omega · have e1 := hc_not c₁ h1 have e2 := hc_not c₂ h2 rw [e1, e2]
The extension map $f$ is injective on $latin\_tableaux\ d_2$.
Suppose $T_1, T_2 \in latin\_tableaux\ d_2$ satisfy $f(T_1) = f(T_2)$. By `latin_tableaux_count_mono_cell_addition_top_row_cell_subset`, every cell $c \in d_2$ is also a cell of $d_1$. Specializing the equality $f(T_1) = f(T_2)$ at $c$ and evaluating the definition of $f$ yields $T_1(c) = T_2(c)$ for every cell $c$ of $d_2$. By function extensionality, $T_1 = T_2$."
uses
latin_tableaux_count_mono_cell_addition_top_row_cell_subset,congr_fun,dif_pos· intro T₁ hT₁ T₂ hT₂ heq ext c have hc1 : c.1 ∈ d₁.cells := latin_tableaux_count_mono_cell_addition_top_row_cell_subset d₁ d₂ hrow_zero hrow_other c.1 c.2 have h_eq_c := congr_fun heq ⟨c.1, hc1⟩ dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] at h_eq_c rw [dif_pos c.2, dif_pos c.2] at h_eq_c exact h_eq_c
Finset.card_le_card_of_injOn- If an injective function on a finite set $s$ maps into another finite set $t$, then the cardinality of $s$ is at most the cardinality of $t$.
Finset.mem_coe- An element belongs to the set coercion of a finite set if and only if it belongs to the finite set.
Finset.mem_filter- An element belongs to the sub-collection `filter p s` if and only if it belongs to $s$ and satisfies the property $p$.
Fintype.mem_piFinset- A function belongs to the product of finite sets `piFinset t` if and only if its value at every argument $a$ belongs to $t(a)$.
mem_Icc- An element $x$ belongs to the closed interval $[a, b]$ if and only if $a \le x$ and $x \le b$.
YoungDiagram.mem_cells- A coordinate pair belongs to the cell set of a Young diagram if and only if it belongs to the diagram.
YoungDiagram.mem_iff_lt_rowLen- A coordinate pair $(i, j)$ belongs to a Young diagram if and only if $j < \text{rowLen } i$.
Prod.ext- Two ordered pairs are equal if their first coordinates are equal and their second coordinates are equal.
False.elim- From a logical contradiction (a proof of False), any conclusion can be deduced.
congr_arg- Applying a function to equal arguments produces equal results.
Prod.fst- The projection function returning the first component of an ordered pair.
Nat.zero_le- Zero is less than or equal to any natural number.
dif_pos- A dependent conditional expression evaluates to its positive branch when the condition is satisfied.
dif_neg- A dependent conditional expression evaluates to its negative branch when the condition is false.
latin_tableaux_count_mono_cell_addition_top_row_cell_subset- When diagram $d_1$ is obtained from $d_2$ by extending row 0 by 1 and leaving all other rows unchanged, every cell of $d_2$ is also a cell of $d_1$.
congr_fun- Equal functions evaluate to equal values at the same input.
- The helper lemma `latin_tableaux_count_mono_cell_addition_top_row_cell_subset` (lines 10–20) establishes the inclusion of cells $d_2.\text{cells} \subseteq d_1.\text{cells}$ needed for injectivity.
- The hypotheses `hwide₁`, `hwide₂` (that $d_1$ and $d_2$ are wide partitions) and `hlen` (that $d_1$ and $d_2$ have the same number of rows) are never used in the formal proof; monotonicity under adding a cell to the top row holds for any two Young diagrams.
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
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
open Finsetdef latin_tableaux_count_mono_cell_addition_top_row_extend (d₁ d₂ : YoungDiagram) (T : d₂.cells → ℕ) : d₁.cells → ℕ := fun c => if h : c.1 ∈ d₂.cells then T ⟨c.1, h⟩ else d₁.rowLen 0lemma latin_tableaux_count_mono_cell_addition_top_row_cell_subset (d₁ d₂ : YoungDiagram) (hrow_zero : d₁.rowLen 0 = d₂.rowLen 0 + 1) (hrow_other : ∀ j, j ≠ 0 → d₁.rowLen j = d₂.rowLen j) (c : ℕ × ℕ) (hc : c ∈ d₂.cells) : c ∈ d₁.cells := by rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc ⊢ by_cases h : c.1 = 0 · rw [h, hrow_zero] rw [h] at hc omega · rw [hrow_other c.1 h] exact hc/-- Proposition 5.5: Monotonicity of Latin tableaux count under cell addition to the top row. -/theorem latin_tableaux_count_mono_cell_addition_top_row (d₁ d₂ : YoungDiagram) (hwide₁ : is_wide_partition d₁) (hwide₂ : is_wide_partition d₂) (hlen : d₁.colLen 0 = d₂.colLen 0) (hrow_zero : d₁.rowLen 0 = d₂.rowLen 0 + 1) (hrow_other : ∀ j, j ≠ 0 → d₁.rowLen j = d₂.rowLen j) : (latin_tableaux d₂).card ≤ (latin_tableaux d₁).card := by let f := latin_tableaux_count_mono_cell_addition_top_row_extend d₁ d₂ apply Finset.card_le_card_of_injOn f · intro T hT rw [Finset.mem_coe] at hT ⊢ rw [latin_tableaux, Finset.mem_filter] refine ⟨?_, ?_, ?_⟩ · rw [Fintype.mem_piFinset] intro c rcases c with ⟨⟨r, k⟩, hc⟩ dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] split_ifs with h · simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT have hrange := hT.1 ⟨(r, k), h⟩ rw [mem_Icc] refine ⟨hrange.1, ?_⟩ have hle : d₂.rowLen r ≤ d₁.rowLen r := by by_cases hc0 : r = 0 · rw [hc0, hrow_zero] omega · rw [hrow_other r hc0] exact hrange.2.trans hle · rw [mem_Icc] have hc_not : (r, k) = (0, d₂.rowLen 0) := by have hc2 : (r, k) ∉ d₂.cells := h rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc2 have hc1 := hc rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc1 by_cases h0 : r = 0 · rw [h0, hrow_zero] at hc1 have hnot : ¬ (k < d₂.rowLen 0) := by rwa [h0] at hc2 have heq2 : k = d₂.rowLen 0 := by omega exact Prod.ext h0 heq2 · rw [hrow_other r h0] at hc1 exact False.elim (hc2 hc1) have hc0 : r = 0 := congr_arg Prod.fst hc_not have : d₁.rowLen r = d₁.rowLen 0 := by rw [hc0] rw [this, hrow_zero] omega · intro c₁ c₂ hr heq have hT_val : ∀ c : d₂.cells, T c < d₁.rowLen 0 := by intro c simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT have hrange := hT.1 c have hanti : d₂.rowLen c.1.1 ≤ d₂.rowLen 0 := d₂.rowLen_anti 0 c.1.1 (Nat.zero_le _) rw [hrow_zero] omega have hc_not : ∀ c : d₁.cells, c.1 ∉ d₂.cells → c.1 = (0, d₂.rowLen 0) := by intro c hc2 rcases c with ⟨⟨r, k⟩, hc⟩ rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc2 have hc1 := hc rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc1 by_cases h0 : r = 0 · rw [h0, hrow_zero] at hc1 have hnot : ¬ (k < d₂.rowLen 0) := by rwa [h0] at hc2 have heq2 : k = d₂.rowLen 0 := by omega exact Prod.ext h0 heq2 · rw [hrow_other r h0] at hc1 exact False.elim (hc2 hc1) dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] at heq by_cases h1 : c₁.1 ∈ d₂.cells <;> by_cases h2 : c₂.1 ∈ d₂.cells · rw [dif_pos h1, dif_pos h2] at heq simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT exact hT.2.1 ⟨c₁.1, h1⟩ ⟨c₂.1, h2⟩ hr heq · rw [dif_pos h1, dif_neg h2] at heq have hlt := hT_val ⟨c₁.1, h1⟩ omega · rw [dif_neg h1, dif_pos h2] at heq have hlt := hT_val ⟨c₂.1, h2⟩ omega · have e1 := hc_not c₁ h1 have e2 := hc_not c₂ h2 rw [e1, e2] · intro c₁ c₂ hc_col heq have hT_val : ∀ c : d₂.cells, T c < d₁.rowLen 0 := by intro c simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT have hrange := hT.1 c have hanti : d₂.rowLen c.1.1 ≤ d₂.rowLen 0 := d₂.rowLen_anti 0 c.1.1 (Nat.zero_le _) rw [hrow_zero] omega have hc_not : ∀ c : d₁.cells, c.1 ∉ d₂.cells → c.1 = (0, d₂.rowLen 0) := by intro c hc2 rcases c with ⟨⟨r, k⟩, hc⟩ rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc2 have hc1 := hc rw [YoungDiagram.mem_cells, YoungDiagram.mem_iff_lt_rowLen] at hc1 by_cases h0 : r = 0 · rw [h0, hrow_zero] at hc1 have hnot : ¬ (k < d₂.rowLen 0) := by rwa [h0] at hc2 have heq2 : k = d₂.rowLen 0 := by omega exact Prod.ext h0 heq2 · rw [hrow_other r h0] at hc1 exact False.elim (hc2 hc1) dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] at heq by_cases h1 : c₁.1 ∈ d₂.cells <;> by_cases h2 : c₂.1 ∈ d₂.cells · rw [dif_pos h1, dif_pos h2] at heq simp only [latin_tableaux, mem_filter, Fintype.mem_piFinset, mem_Icc] at hT exact hT.2.2 ⟨c₁.1, h1⟩ ⟨c₂.1, h2⟩ hc_col heq · rw [dif_pos h1, dif_neg h2] at heq have hlt := hT_val ⟨c₁.1, h1⟩ omega · rw [dif_neg h1, dif_pos h2] at heq have hlt := hT_val ⟨c₂.1, h2⟩ omega · have e1 := hc_not c₁ h1 have e2 := hc_not c₂ h2 rw [e1, e2] · intro T₁ hT₁ T₂ hT₂ heq ext c have hc1 : c.1 ∈ d₁.cells := latin_tableaux_count_mono_cell_addition_top_row_cell_subset d₁ d₂ hrow_zero hrow_other c.1 c.2 have h_eq_c := congr_fun heq ⟨c.1, hc1⟩ dsimp only [f, latin_tableaux_count_mono_cell_addition_top_row_extend] at h_eq_c rw [dif_pos c.2, dif_pos c.2] at h_eq_c exact h_eq_c