wild_triple_partition_large_order_mass_bound
For odd q>=100001, both canonical seal-event resource sums for q triples of vertices in K_(3q), with weight (21/20)^d/(3q)_d, are at most 1/50.
Statement
theorem wild_triple_partition_large_order_mass_bound (q : ℕ) (hq : 100001 ≤ q) (hodd : q % 2 = 1) : (∑ m ∈ Finset.range q, ((Nat.choose (q - 1) m : ℕ) : ℝ) * (if m + 1 ≤ q / 2 then (3 * q : ℝ) * (((3 * (m + 1)).factorial * Nat.choose (3 * q / 2) (3 * (m + 1) / 2) : ℕ) : ℝ) * wtc_wt (q := q) (3 * (m + 1)) else 0) ≤ 1 / 50) ∧ (∑ m ∈ Finset.range (q + 1), ((Nat.choose q m : ℕ) : ℝ) * (if 1 ≤ m ∧ m ≤ q / 2 then ((((3 * m).factorial * (if (3 * m) % 2 = 1 then Nat.choose (3 * q / 2) (3 * m / 2) else 0) + (3 * q - 1) * ((3 * m).factorial * Nat.choose (3 * q / 2 - 1) (3 * m / 2 - 1)) : ℕ) : ℝ)) * wtc_wt (q := q) (3 * m) else 0) ≤ 1 / 50)
- Source
- Uniform arithmetic estimate for OP-306. Vandermonde bounds give C(q,h)^3 <= C(3q,3h) and C(floor(3q/2),floor(3h/2))^2 <= C(3q,3h). Two exposed binomial factors and the geometric-square sum 330 give the threshold.
- Verified
- 29 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 3
- wild_triple_partition_mass_criterionFor an odd number q of triple blocks, q>=3, a proper edge coloring of K_(3q) with 3q colors and a bijection between colors and their…
- TotalPreorderA total preorder: a preorder where
≤is total (complete). This is weaker thanLinearOrder— it does NOT require antisymmetry or… - LinearOrder.toTotalPreorderEvery
LinearOrderis aTotalPreorder.
- Used by 6
- wild_triple_partition_term_geometric_tailFor q>=3 odd and any h>=5 with2*h<=q, the existing canonical triple-partition resource term wtc_term(q,h) is at most 3*h*(27/40)^h. The…
- wild_triple_partition_term_antitone_odd_stepFor every odd q>=3 and admissible fixed cut size h>=1 with2*h<=q, the canonical triple-partition resource term wtc_term(q,h) weakly…
- wild_triple_partition_mass_le_of_odd_ge_179For every odd number q of triple blocks with q>=179, the existing sufficient canonical seal-event resource sum wtc_mass(q) is at most1/50.…
- wild_partition_large_order_resource_boundsSuppose n is odd, q>=10000000, 3q<=n<=3q+2, and a proper n-coloring of K_n has a bijection between colors and their unique missing…
- wild_number_complete_graph_large_odd_multiple_three_eqFor every odd integer q>=100001, every natural number k equal to the chromatic number of the line graph of the complete graph K_(3q), and…
- wild_triple_partition_mass_179_baselineThe sum of the canonical triple-partition resource terms wtc_term(179, h) for cut sizes h from 1 to 39 is at most 989/50000 = 0.01978.
The proof, explained
To bound both canonical seal-event resource sums by 1/50 for odd q ≥ 100001, the proof identifies both sums with a common total mass wtc_mass q. The first term h = 1 is computed explicitly, while higher terms h ≥ 2 are bounded by (3/50000) * h^2 * (5/6)^h via combinatorial bounds on binomial cubes and squares. The geometric-square sum is bounded by 330, yielding a total mass of at most 99/5000 ≤ 1/50.
For any a, b, i, j in ℕ, (a.choose i) * (b.choose j) ≤ (a + b).choose (i + j).
By Vandermonde's convolution identity, (a + b).choose (i + j) expands as a sum of products of binomial coefficients over all pairs summing to i + j. Since all terms are non-negative, the single term corresponding to (i, j) is bounded by the full sum.
uses
Nat.add_choose_eq,Finset.single_le_sumrw [Nat.add_choose_eq] exact Finset.single_le_sum (a := (i, j)) (f := fun ij : ℕ × ℕ => a.choose ij.1 * b.choose ij.2) (fun _ _ => Nat.zero_le _) (Finset.HasAntidiagonal.mem_antidiagonal.mpr rfl)
For any q, h in ℕ, (q.choose h)^3 ≤ (3 * q).choose (3 * h).
Applying the product inequality wtc_choose_mul_le twice bounds (q.choose h)^2 by (2 * q).choose (2 * h), and multiplying by q.choose h gives a bound of (3 * q).choose (3 * h).
uses
wtc_choose_mul_lehave h1 := wtc_choose_mul_le q q h h have h2 := wtc_choose_mul_le (q + q) q (h + h) h calc _ = (q.choose h * q.choose h) * q.choose h := by ring _ ≤ (q + q).choose (h + h) * q.choose h := Nat.mul_le_mul_right _ h1 _ ≤ _ := by convert h2 using 1 <;> congr 1 <;> omega
For any m, s in ℕ, (m.choose (s / 2))^2 ≤ (2 * m + 1).choose s.
Splitting s into s / 2 and s / 2 + s % 2, a case split on the parity of s shows m.choose (s / 2) ≤ (m + 1).choose (s / 2 + s % 2). Multiplying by m.choose (s / 2) and applying wtc_choose_mul_le to m + (m + 1) = 2 * m + 1 completes the bound.
uses
wtc_choose_mul_lehave hs : s = s / 2 + (s / 2 + s % 2) := by omega have hc : m.choose (s / 2) ≤ (m + 1).choose (s / 2 + s % 2) := by have hmod := Nat.mod_lt s (by omega : 0 < 2) interval_cases hm : s % 2 · simpa [hm] using Nat.choose_le_succ m (s / 2) · rw [Nat.choose_succ_succ] omega calc _ = m.choose (s / 2) * m.choose (s / 2) := by ring _ ≤ m.choose (s / 2) * (m + 1).choose (s / 2 + s % 2) := Nat.mul_le_mul_left _ hc _ ≤ _ := by convert wtc_choose_mul_le m (m + 1) (s / 2) (s / 2 + s % 2) using 1 <;> congr 1 <;> omega
For 2 * h ≤ q, 2^h ≤ q.choose h.
By induction on h. In the inductive step, multiplying by h + 1 and using the absorption identity (q + 1) * q.choose h = (q + 1).choose (h + 1) * (h + 1) lets us compare 2^(h + 1) * (h + 1) with (q + 1) * 2^h.
induction h generalizing q with | zero => simp | succ h ih => obtain ⟨q, rfl⟩ := Nat.exists_eq_succ_of_ne_zero (by omega : q ≠ 0) have hi := ih q (by omega) have he := Nat.add_one_mul_choose_eq q h have hz : 0 < h + 1 := by omega apply Nat.le_of_mul_le_mul_right (c := h + 1) (hc := hz) calc 2 ^ (h + 1) * (h + 1) = (2 * (h + 1)) * 2 ^ h := by rw [pow_succ]; ring _ ≤ (q + 1) * 2 ^ h := Nat.mul_le_mul_right _ hh _ ≤ (q + 1) * q.choose h := Nat.mul_le_mul_left _ hi _ = _ := he
For 1 ≤ h and 2 * h ≤ q, q * 2^(h - 1) ≤ h * q.choose h.
Shifting both indices by 1, this follows from 2^(h - 1) ≤ (q - 1).choose (h - 1) by multiplying both sides by q and applying the absorption identity.
uses
wtc_two_pow_le_chooseobtain ⟨h, rfl⟩ := Nat.exists_eq_succ_of_ne_zero (by omega : h ≠ 0) obtain ⟨q, rfl⟩ := Nat.exists_eq_succ_of_ne_zero (by omega : q ≠ 0) have hi := wtc_two_pow_le_choose q h (by omega) simpa [Nat.mul_comm] using (Nat.mul_le_mul_left (q + 1) hi).trans_eq (Nat.add_one_mul_choose_eq q h)
For all n, the sum of h^2 * (5/6)^h for h < n is at most 330.
We prove by induction on n that the partial sum plus the remainder (6 * n^2 + 60 * n + 330) * (5/6)^n exactly equals 330. Since the remainder term is always non-negative, the partial sum is at most 330.
uses
Finset.sum_range_succhave heq : ∀ n : ℕ, (∑ h ∈ Finset.range n, (h : ℝ)^2 * (5 / 6 : ℝ)^h) + (6 * (n : ℝ)^2 + 60 * n + 330) * (5 / 6 : ℝ)^n = 330 := by intro n induction n with | zero => norm_num | succ n ih => rw [Finset.sum_range_succ] push_cast rw [pow_succ (5 / 6 : ℝ) n] nlinarith only [ih] have hp : 0 ≤ (6 * (n : ℝ)^2 + 60 * n + 330) * (5 / 6 : ℝ)^n := by positivity linarith [heq n]
For 2 ≤ h and 2 * h ≤ q, q * (q - 1) * 2^(h - 2) ≤ h * (h - 1) * q.choose h.
Applying wtc_choose_lower to q - 1 and h - 1, multiplying by q, and substituting q * (q - 1).choose (h - 1) = h * q.choose h gives the second-order lower bound.
uses
wtc_choose_lowerhave hi := wtc_choose_lower (q - 1) (h - 1) (by omega) (by omega) have he := Nat.add_one_mul_choose_eq (q - 1) (h - 1) rw [Nat.sub_add_cancel (by omega : 1 ≤ q), Nat.sub_add_cancel (by omega : 1 ≤ h)] at he have he' : h - 1 - 1 = h - 2 := by omega rw [he'] at hi nlinarith only [Nat.mul_le_mul_left q hi, congrArg (fun x => (h - 1) * x) he]
For odd q ≥ 3, wtc_term q 1 = 3 * (21/20)^3 / (3 * q - 2).
At h = 1, (3 * q / 2).choose (3 * 1 / 2) becomes (3 * q - 1) / 2. Expressing (3 * q).choose 3 via descending factorials as 3 * q * (3 * q - 1) * (3 * q - 2) / 6, the factor 3 * q - 1 cancels, yielding the stated formula.
uses
Nat.descFactorial_eq_factorial_mul_choosehave hmN : 2 * (3 * q / 2) = 3 * q - 1 := by omega have hmR : 2 * ((3 * q / 2 : ℕ) : ℝ) = 3 * (q : ℝ) - 1 := by have ht : 2 * ((3 * q / 2 : ℕ) : ℝ) + 1 = 3 * (q : ℝ) := by exact_mod_cast (show 2 * (3 * q / 2) + 1 = 3 * q by omega) linarith have he := Nat.descFactorial_eq_factorial_mul_choose (3 * q) 3 norm_num only [Nat.descFactorial_succ, Nat.descFactorial_zero, Nat.sub_zero, Nat.factorial_succ, Nat.factorial_zero] at he have heR : (3 * (q : ℝ) - 2) * (3 * (q : ℝ) - 1) * (3 * (q : ℝ)) = 6 * ((3 * q).choose 3 : ℝ) := by have ht := congrArg (fun x : ℕ => (x : ℝ)) he simp only [Nat.cast_mul, Nat.cast_one] at ht rw [Nat.cast_sub (by omega : 2 ≤ 3 * q), Nat.cast_sub (by omega : 1 ≤ 3 * q)] at ht push_cast at ht nlinarith only [ht] have hc : ((3 * q).choose 3 : ℝ) ≠ 0 := by exact_mod_cast Nat.choose_ne_zero (by omega : 3 ≤ 3 * q) have hqR : (3 : ℝ) ≤ q := by exact_mod_cast hq unfold wtc_term norm_num only [Nat.choose_one_right, Nat.mul_one, Nat.reduceDiv] field_simp [hc, ne_of_gt (by linarith : (0 : ℝ) < 3 * (q : ℝ) - 2)] nlinarith only [heR, congrArg (fun x : ℝ => x * (3 * (q : ℝ)) * (3 * (q : ℝ) - 2)) hmR]
For odd q ≥ 100001 and h = 1, wtc_term q h ≤ (3 / 50000) * 1^2 * (5 / 6)^1.
Substituting the explicit formula for wtc_term q 1, the inequality simplifies to 27783 / 8000 / (3 * q - 2) ≤ 1 / 20000, which holds by linear arithmetic since q ≥ 100001.
uses
wtc_first_termby_cases heq : h = 1 · subst h rw [wtc_first_term q (by omega) hodd] norm_num have hqR : (100001 : ℝ) ≤ q := by exact_mod_cast hq rw [div_le_iff₀ (by linarith : (0 : ℝ) < 3 * (q : ℝ) - 2)] linarith
For odd q ≥ 100001, 2 ≤ h, and 2 * h ≤ q, wtc_term q h ≤ (3 / 50000) * h^2 * (5 / 6)^h.
Setting B = q.choose h, C = (3 * q).choose (3 * h), and D = (3 * q / 2).choose (3 * h / 2), the cube and square bounds yield B^3 ≤ C and D^2 ≤ C, so B * (B * D / C)^2 ≤ 1. Combining this with the lower bound on B and (21/20)^6 ≤ 2 * (5/6)^2 gives q * (q - 1) * (wtc_term q h)^2 ≤ 36 * h^4 * ((5/6)^h)^2. Since q * (q - 1) ≥ 10^10, taking square roots establishes the sharper bound.
uses
wtc_choose_cube,wtc_choose_square,wtc_choose_lower_twohave h2 : 2 ≤ h := by omega let B : ℝ := q.choose h let C : ℝ := (3 * q).choose (3 * h) let D : ℝ := (3 * q / 2).choose (3 * h / 2) have hb : 0 < B := by dsimp only [B]; exact_mod_cast Nat.choose_pos (by omega : h ≤ q) have hc : 0 < C := by dsimp only [C]; exact_mod_cast Nat.choose_pos (by omega : 3 * h ≤ 3 * q) have hd : 0 ≤ D := by positivity have hb3 : B^3 ≤ C := by dsimp only [B, C]; exact_mod_cast wtc_choose_cube q h have hn : 2 * (3 * q / 2) + 1 = 3 * q := by omega have hd2 : D^2 ≤ C := by dsimp only [D, C] exact_mod_cast (hn ▸ wtc_choose_square (3 * q / 2) (3 * h)) have hchoose : (q : ℝ) * ((q : ℝ) - 1) * 2^(h-2) ≤ (h : ℝ) * ((h : ℝ) - 1) * B := by dsimp only [B] have ht : ((q * (q - 1) * 2 ^ (h - 2) : ℕ) : ℝ) ≤ ((h * (h - 1) * q.choose h : ℕ) : ℝ) := by exact_mod_cast wtc_choose_lower_two q h h2 hh simp only [Nat.cast_mul] at ht rw [Nat.cast_sub (by omega : 1 ≤ q), Nat.cast_sub (by omega : 1 ≤ h)] at ht simpa using ht have hqR : (100001 : ℝ) ≤ q := by exact_mod_cast hq have hhR : 2 ≤ (h : ℝ) := by exact_mod_cast h2 have hhm : 0 ≤ (h : ℝ) - 1 := by linarith have hqm : 0 < (q : ℝ) - 1 := by linarith have hbase : (21 / 20 : ℝ)^6 ≤ 2 * (5 / 6 : ℝ)^2 := by norm_num have hpw := pow_le_pow_left₀ (by positivity : (0 : ℝ) ≤ (21 / 20)^6) hbase h have hpweq : ((21 / 20 : ℝ) ^ (3 * h))^2 ≤ 2^h * ((5 / 6 : ℝ)^h)^2 := by have he1 : 3 * h * 2 = 6 * h := by omega have he2 : 2 * h = h * 2 := by omega simpa only [mul_pow, ← pow_mul, he1, he2] using hpw have h2pow : (2 : ℝ)^h = 4 * (2 : ℝ)^(h-2) := by calc (2 : ℝ)^h = 2^(h-2+2) := by congr 1; omega _ = _ := by rw [pow_add]; norm_num; ring have hratio : B * (B * D / C)^2 ≤ 1 := by have hmul : B^3 * D^2 ≤ C^2 := by simpa only [pow_two] using mul_le_mul hb3 hd2 (by positivity : 0 ≤ D^2) (by positivity : 0 ≤ C) have he : B * (B * D / C)^2 = (B^3 * D^2) / C^2 := by ring rw [he, div_le_one (sq_pos_of_pos hc)] exact hmul have hratio' : (q : ℝ) * ((q : ℝ)-1) * 2^(h-2) * (B * D / C)^2 ≤ (h : ℝ) * ((h : ℝ)-1) := by calc _ ≤ ((h : ℝ) * ((h : ℝ)-1) * B) * (B * D / C)^2 := mul_le_mul_of_nonneg_right hchoose (sq_nonneg _) _ = (h : ℝ) * ((h : ℝ)-1) * (B * (B * D / C)^2) := by ring _ ≤ _ := by simpa using mul_le_mul_of_nonneg_left hratio (by positivity : 0 ≤ (h : ℝ)*((h : ℝ)-1)) have hs := mul_le_mul hratio' hpweq (by positivity) (by positivity : 0 ≤ (h : ℝ)*((h : ℝ)-1)) rw [h2pow] at hs have hs' : (q : ℝ) * ((q : ℝ)-1) * (B * D / C)^2 * ((21 / 20 : ℝ) ^ (3*h))^2 ≤ 4 * (h : ℝ) * ((h : ℝ)-1) * ((5/6 : ℝ)^h)^2 := by apply (mul_le_mul_iff_of_pos_left (by positivity : (0 : ℝ) < 2^(h-2))).mp nlinarith only [hs] have hquot : (wtc_term q h)^2 = 9 * (h : ℝ)^2 * (B * D / C)^2 * ((21 / 20 : ℝ)^(3*h))^2 := by dsimp [wtc_term, B, C, D]; push_cast; ring have hupper : (q : ℝ) * ((q : ℝ)-1) * (wtc_term q h)^2 ≤ 36 * (h : ℝ)^4 * ((5/6 : ℝ)^h)^2 := by rw [hquot] have hmul := mul_le_mul_of_nonneg_left hs' (by positivity : 0 ≤ 9*(h : ℝ)^2) have hhh : (h : ℝ) * ((h : ℝ)-1) ≤ (h : ℝ)^2 := by nlinarith have hmul2 := mul_le_mul_of_nonneg_left hhh (by positivity : 0 ≤ 36*(h : ℝ)^2*((5/6 : ℝ)^h)^2) nlinarith only [hmul, hmul2] have hqq : (10000000000 : ℝ) ≤ (q : ℝ) * ((q : ℝ)-1) := by nlinarith [sq_nonneg ((q : ℝ)-100001)] have hsq : (wtc_term q h)^2 ≤ ((3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h)^2 := by apply (mul_le_mul_iff_of_pos_left (by positivity : (0 : ℝ) < (q : ℝ)*((q : ℝ)-1))).mp calc _ ≤ 36 * (h : ℝ)^4 * ((5/6 : ℝ)^h)^2 := hupper _ ≤ _ := by have hx := mul_le_mul_of_nonneg_right hqq (by positivity : 0 ≤ ((3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h)^2) nlinarith only [hx] have hnon : 0 ≤ (3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h := by positivity nlinarith [sq_nonneg (wtc_term q h + (3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h)]
For odd q ≥ 100001, wtc_mass q ≤ 1 / 50.
Each non-zero term in wtc_mass q is bounded by (3 / 50000) * h^2 * (5 / 6)^h. Factoring out 3 / 50000 and bounding the remaining sum by 330 using wtc_geometric_square_sum yields wtc_mass q ≤ (3 / 50000) * 330 = 99 / 5000 ≤ 1 / 50.
uses
wtc_term_bound_sharper,wtc_geometric_square_sumhave hs : wtc_mass q ≤ (3 / 50000 : ℝ) * ∑ h ∈ Finset.range (q + 1), (h : ℝ)^2 * (5 / 6 : ℝ)^h := by unfold wtc_mass rw [Finset.mul_sum] refine Finset.sum_le_sum fun h hh => ?_ split_ifs with hcond · simpa only [mul_assoc] using wtc_term_bound_sharper q h hq hodd hcond.1 (by omega) · positivity have hg := wtc_geometric_square_sum (q + 1) nlinarith
For all q, d, d! * wtc_wt d = (21/20)^d / (3 * q).choose d.
By definition, wtc_wt d has denominator (3 * q).descFactorial d. Using (3 * q).descFactorial d = d! * (3 * q).choose d, the factor of d! cancels from the numerator.
uses
Nat.descFactorial_eq_factorial_mul_chooseunfold wtc_wt rw [Nat.descFactorial_eq_factorial_mul_choose] push_cast have hf : (d.factorial : ℝ) ≠ 0 := by positivity by_cases hc : ((3 * q).choose d : ℝ) = 0 · simp [hc] · field_simp
For q, h ≥ 1, the position summand for h equals wtc_term q h.
Using the binomial absorption identity q * (q - 1).choose (h - 1) = h * q.choose h and substituting the simplified form of (3 * h)! * wtc_wt (3 * h), the product rearranges directly into wtc_term q h.
uses
wtc_factorial_weighthave hi := Nat.add_one_mul_choose_eq (q - 1) (h - 1) rw [Nat.sub_add_cancel hq, Nat.sub_add_cancel h1] at hi have hir : (q : ℝ) * ((q - 1).choose (h - 1) : ℝ) = (q.choose h : ℝ) * (h : ℝ) := by exact_mod_cast hi push_cast calc _ = 3 * ((q : ℝ) * ((q - 1).choose (h - 1) : ℝ)) * ((3 * q / 2).choose (3 * h / 2) : ℝ) * (( (3 * h).factorial : ℝ) * wtc_wt (q := q) (3 * h)) := by ring _ = _ := by rw [hir, wtc_factorial_weight]; unfold wtc_term; push_cast; ring
For odd n ≥ 3 and s ≥ 3, the vertex degree coefficient identity holds.
Since n is odd, n - 1 = 2 * (n / 2). Using absorption on (n / 2 - 1).choose (s / 2 - 1) and splitting on whether s is odd or even, the two terms combine to s * (n / 2).choose (s / 2).
have hi := Nat.add_one_mul_choose_eq (n / 2 - 1) (s / 2 - 1) rw [Nat.sub_add_cancel (by omega : 1 ≤ n / 2), Nat.sub_add_cancel (by omega : 1 ≤ s / 2)] at hi have hn' : n - 1 = 2 * (n / 2) := by omega rw [hn'] split_ifs with hmod · have hs' : s = 2 * (s / 2) + 1 := by omega nlinarith only [hi, congrArg (fun x => x * (n / 2).choose (s / 2)) hs'] · have hs' : s = 2 * (s / 2) := by omega nlinarith only [hi, congrArg (fun x => x * (n / 2).choose (s / 2)) hs']
For q ≥ 1, the canonical position sum equals wtc_mass q.
Re-indexing the sum from m to h = m + 1, the h = 0 term vanishes and every term with h ≤ q / 2 simplifies to wtc_term q h by wtc_position_identity, exactly matching wtc_mass q.
uses
wtc_position_identityunfold wtc_mass rw [Finset.sum_range_succ'] simp only [not_le.mpr (by omega : 0 < 1), false_and, if_false, add_zero] refine Finset.sum_congr rfl fun m hm => ?_ by_cases hb : m + 1 ≤ q / 2 · simp only [hb, if_true, Nat.add_sub_cancel, show 1 ≤ m + 1 by omega, true_and] simpa only [Nat.add_sub_cancel, mul_assoc] using wtc_position_identity q (m + 1) hq (by omega) · simp only [hb, if_false, and_false, mul_zero]
For odd q ≥ 1, the canonical vertex sum equals wtc_mass q.
Applying wtc_vertex_coefficient with n = 3 * q and s = 3 * h simplifies the coefficient in the vertex sum to 3 * h * (3 * q / 2).choose (3 * h / 2). Multiplying by the factorial weight simplifies the summand to wtc_term q h, matching wtc_mass q.
uses
wtc_vertex_coefficient,wtc_factorial_weightunfold wtc_mass refine Finset.sum_congr rfl fun h hh => ?_ by_cases hcond : 1 ≤ h ∧ h ≤ q / 2 · simp only [if_pos hcond] have hn : (3 * q) % 2 = 1 := by omega have hi := wtc_vertex_coefficient (3 * q) (3 * h) hn (by omega) (by omega) have hi' : (3 * h).factorial * (if (3 * h) % 2 = 1 then (3 * q / 2).choose (3 * h / 2) else 0) + (3 * q - 1) * ((3 * h).factorial * (3 * q / 2 - 1).choose (3 * h / 2 - 1)) = (3 * h).factorial * (3 * h * (3 * q / 2).choose (3 * h / 2)) := by nlinarith [hi] rw [hi'] push_cast calc _ = (3 * (h : ℝ)) * (q.choose h : ℝ) * ((3 * q / 2).choose (3 * h / 2) : ℝ) * (((3 * h).factorial : ℝ) * wtc_wt (q := q) (3 * h)) := by ring _ = _ := by rw [wtc_factorial_weight]; unfold wtc_term; push_cast; ring · simp only [if_neg hcond, mul_zero]
Both canonical seal-event resource sums are at most 1/50 for odd q ≥ 100001.
By wtc_position_sum_eq and wtc_vertex_sum_eq, both sums equal wtc_mass q. Applying the mass bound wtc_mass q ≤ 1/50 proves both inequalities simultaneously.
uses
wtc_mass_bound_sharper,wtc_position_sum_eq,wtc_vertex_sum_eqhave hmass := wtc_mass_bound_sharper q hq hodd exact ⟨(wtc_position_sum_eq q (by omega)).trans_le hmass, (wtc_vertex_sum_eq q (by omega) hodd).trans_le hmass⟩
Nat.add_choose_eq- Vandermonde's convolution formula: (m + n).choose k equals the sum of m.choose i * n.choose j over all pairs (i, j) with i + j = k.
Finset.single_le_sum- A single non-negative term in a finite sum is bounded above by the total sum.
wtc_choose_mul_le- For all natural numbers a, b, i, j, a.choose i * b.choose j <= (a + b).choose (i + j).
wtc_choose_cube- For all q, h, (q.choose h)^3 <= (3 * q).choose (3 * h).
wtc_choose_square- For all m, s, m.choose (s / 2)^2 <= (2 * m + 1).choose s.
wtc_two_pow_le_choose- If 2 * h <= q, then 2^h <= q.choose h.
wtc_choose_lower- If 1 <= h and 2 * h <= q, then q * 2^(h - 1) <= h * q.choose h.
Finset.sum_range_succ- Splitting off the last term of a sum over Finset.range (n + 1).
wtc_choose_lower_two- If 2 <= h and 2 * h <= q, then q * (q - 1) * 2^(h - 2) <= h * (h - 1) * q.choose h.
wtc_first_term- For odd q >= 3, the h = 1 term simplifies to 3 * (21/20)^3 / (3q - 2).
wtc_term_bound_sharper- For odd q >= 100001, 1 <= h, and 2 * h <= q, wtc_term q h <= (3 / 50000) * h^2 * (5 / 6)^h.
wtc_geometric_square_sum- The partial sum of h^2 * (5/6)^h over h < n is at most 330 for all n.
wtc_mass_bound_sharper- For odd q >= 100001, the total mass wtc_mass q is at most 1/50.
Nat.descFactorial_eq_factorial_mul_choose- The descending factorial n.descFactorial k equals k! * n.choose k.
wtc_factorial_weight- The weight wtc_wt d multiplied by d! equals (21/20)^d / (3q).choose d.
wtc_position_identity- The canonical position summand for h equals wtc_term q h.
wtc_vertex_coefficient- An algebraic identity relating binomial coefficients across parity cases for odd n.
wtc_position_sum_eq- The canonical position sum over all m equals wtc_mass q.
wtc_vertex_sum_eq- The canonical vertex sum over all m equals wtc_mass q.
- The definitions wtc_term and wtc_mass use integer division by 2, where for odd n, n / 2 represents (n - 1) / 2.
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.
set_option maxHeartbeats 2000000set_option maxRecDepth 100000lemma wtc_choose_mul_le (a b i j : ℕ) : a.choose i * b.choose j ≤ (a + b).choose (i + j) := by rw [Nat.add_choose_eq] exact Finset.single_le_sum (a := (i, j)) (f := fun ij : ℕ × ℕ => a.choose ij.1 * b.choose ij.2) (fun _ _ => Nat.zero_le _) (Finset.HasAntidiagonal.mem_antidiagonal.mpr rfl)lemma wtc_choose_cube (q h : ℕ) : q.choose h ^ 3 ≤ (3 * q).choose (3 * h) := by have h1 := wtc_choose_mul_le q q h h have h2 := wtc_choose_mul_le (q + q) q (h + h) h calc _ = (q.choose h * q.choose h) * q.choose h := by ring _ ≤ (q + q).choose (h + h) * q.choose h := Nat.mul_le_mul_right _ h1 _ ≤ _ := by convert h2 using 1 <;> congr 1 <;> omegalemma wtc_choose_square (m s : ℕ) : m.choose (s / 2) ^ 2 ≤ (2 * m + 1).choose s := by have hs : s = s / 2 + (s / 2 + s % 2) := by omega have hc : m.choose (s / 2) ≤ (m + 1).choose (s / 2 + s % 2) := by have hmod := Nat.mod_lt s (by omega : 0 < 2) interval_cases hm : s % 2 · simpa [hm] using Nat.choose_le_succ m (s / 2) · rw [Nat.choose_succ_succ] omega calc _ = m.choose (s / 2) * m.choose (s / 2) := by ring _ ≤ m.choose (s / 2) * (m + 1).choose (s / 2 + s % 2) := Nat.mul_le_mul_left _ hc _ ≤ _ := by convert wtc_choose_mul_le m (m + 1) (s / 2) (s / 2 + s % 2) using 1 <;> congr 1 <;> omegalemma wtc_two_pow_le_choose (q h : ℕ) (hh : 2 * h ≤ q) : 2 ^ h ≤ q.choose h := by induction h generalizing q with | zero => simp | succ h ih => obtain ⟨q, rfl⟩ := Nat.exists_eq_succ_of_ne_zero (by omega : q ≠ 0) have hi := ih q (by omega) have he := Nat.add_one_mul_choose_eq q h have hz : 0 < h + 1 := by omega apply Nat.le_of_mul_le_mul_right (c := h + 1) (hc := hz) calc 2 ^ (h + 1) * (h + 1) = (2 * (h + 1)) * 2 ^ h := by rw [pow_succ]; ring _ ≤ (q + 1) * 2 ^ h := Nat.mul_le_mul_right _ hh _ ≤ (q + 1) * q.choose h := Nat.mul_le_mul_left _ hi _ = _ := helemma wtc_choose_lower (q h : ℕ) (h1 : 1 ≤ h) (hh : 2 * h ≤ q) : q * 2 ^ (h - 1) ≤ h * q.choose h := by obtain ⟨h, rfl⟩ := Nat.exists_eq_succ_of_ne_zero (by omega : h ≠ 0) obtain ⟨q, rfl⟩ := Nat.exists_eq_succ_of_ne_zero (by omega : q ≠ 0) have hi := wtc_two_pow_le_choose q h (by omega) simpa [Nat.mul_comm] using (Nat.mul_le_mul_left (q + 1) hi).trans_eq (Nat.add_one_mul_choose_eq q h)lemma wtc_geometric_square_sum (n : ℕ) : ∑ h ∈ Finset.range n, (h : ℝ)^2 * (5 / 6 : ℝ)^h ≤ 330 := by have heq : ∀ n : ℕ, (∑ h ∈ Finset.range n, (h : ℝ)^2 * (5 / 6 : ℝ)^h) + (6 * (n : ℝ)^2 + 60 * n + 330) * (5 / 6 : ℝ)^n = 330 := by intro n induction n with | zero => norm_num | succ n ih => rw [Finset.sum_range_succ] push_cast rw [pow_succ (5 / 6 : ℝ) n] nlinarith only [ih] have hp : 0 ≤ (6 * (n : ℝ)^2 + 60 * n + 330) * (5 / 6 : ℝ)^n := by positivity linarith [heq n]noncomputable def wtc_term (q h : ℕ) : ℝ := (3 * h : ℝ) * (q.choose h : ℝ) * ((3 * q / 2).choose (3 * h / 2) : ℝ) / ((3 * q).choose (3 * h) : ℝ) * (21 / 20 : ℝ) ^ (3 * h)noncomputable def wtc_mass (q : ℕ) : ℝ := ∑ h ∈ Finset.range (q + 1), if 1 ≤ h ∧ h ≤ q / 2 then wtc_term q h else 0set_option maxHeartbeats 2000000set_option maxRecDepth 100000lemma wtc_choose_lower_two (q h : ℕ) (h2 : 2 ≤ h) (hh : 2 * h ≤ q) : q * (q - 1) * 2 ^ (h - 2) ≤ h * (h - 1) * q.choose h := by have hi := wtc_choose_lower (q - 1) (h - 1) (by omega) (by omega) have he := Nat.add_one_mul_choose_eq (q - 1) (h - 1) rw [Nat.sub_add_cancel (by omega : 1 ≤ q), Nat.sub_add_cancel (by omega : 1 ≤ h)] at he have he' : h - 1 - 1 = h - 2 := by omega rw [he'] at hi nlinarith only [Nat.mul_le_mul_left q hi, congrArg (fun x => (h - 1) * x) he]lemma wtc_first_term (q : ℕ) (hq : 3 ≤ q) (hodd : q % 2 = 1) : wtc_term q 1 = 3 * (21 / 20 : ℝ)^3 / (3 * (q : ℝ) - 2) := by have hmN : 2 * (3 * q / 2) = 3 * q - 1 := by omega have hmR : 2 * ((3 * q / 2 : ℕ) : ℝ) = 3 * (q : ℝ) - 1 := by have ht : 2 * ((3 * q / 2 : ℕ) : ℝ) + 1 = 3 * (q : ℝ) := by exact_mod_cast (show 2 * (3 * q / 2) + 1 = 3 * q by omega) linarith have he := Nat.descFactorial_eq_factorial_mul_choose (3 * q) 3 norm_num only [Nat.descFactorial_succ, Nat.descFactorial_zero, Nat.sub_zero, Nat.factorial_succ, Nat.factorial_zero] at he have heR : (3 * (q : ℝ) - 2) * (3 * (q : ℝ) - 1) * (3 * (q : ℝ)) = 6 * ((3 * q).choose 3 : ℝ) := by have ht := congrArg (fun x : ℕ => (x : ℝ)) he simp only [Nat.cast_mul, Nat.cast_one] at ht rw [Nat.cast_sub (by omega : 2 ≤ 3 * q), Nat.cast_sub (by omega : 1 ≤ 3 * q)] at ht push_cast at ht nlinarith only [ht] have hc : ((3 * q).choose 3 : ℝ) ≠ 0 := by exact_mod_cast Nat.choose_ne_zero (by omega : 3 ≤ 3 * q) have hqR : (3 : ℝ) ≤ q := by exact_mod_cast hq unfold wtc_term norm_num only [Nat.choose_one_right, Nat.mul_one, Nat.reduceDiv] field_simp [hc, ne_of_gt (by linarith : (0 : ℝ) < 3 * (q : ℝ) - 2)] nlinarith only [heR, congrArg (fun x : ℝ => x * (3 * (q : ℝ)) * (3 * (q : ℝ) - 2)) hmR]lemma wtc_term_bound_sharper (q h : ℕ) (hq : 100001 ≤ q) (hodd : q % 2 = 1) (h1 : 1 ≤ h) (hh : 2 * h ≤ q) : wtc_term q h ≤ (3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h := by by_cases heq : h = 1 · subst h rw [wtc_first_term q (by omega) hodd] norm_num have hqR : (100001 : ℝ) ≤ q := by exact_mod_cast hq rw [div_le_iff₀ (by linarith : (0 : ℝ) < 3 * (q : ℝ) - 2)] linarith have h2 : 2 ≤ h := by omega let B : ℝ := q.choose h let C : ℝ := (3 * q).choose (3 * h) let D : ℝ := (3 * q / 2).choose (3 * h / 2) have hb : 0 < B := by dsimp only [B]; exact_mod_cast Nat.choose_pos (by omega : h ≤ q) have hc : 0 < C := by dsimp only [C]; exact_mod_cast Nat.choose_pos (by omega : 3 * h ≤ 3 * q) have hd : 0 ≤ D := by positivity have hb3 : B^3 ≤ C := by dsimp only [B, C]; exact_mod_cast wtc_choose_cube q h have hn : 2 * (3 * q / 2) + 1 = 3 * q := by omega have hd2 : D^2 ≤ C := by dsimp only [D, C] exact_mod_cast (hn ▸ wtc_choose_square (3 * q / 2) (3 * h)) have hchoose : (q : ℝ) * ((q : ℝ) - 1) * 2^(h-2) ≤ (h : ℝ) * ((h : ℝ) - 1) * B := by dsimp only [B] have ht : ((q * (q - 1) * 2 ^ (h - 2) : ℕ) : ℝ) ≤ ((h * (h - 1) * q.choose h : ℕ) : ℝ) := by exact_mod_cast wtc_choose_lower_two q h h2 hh simp only [Nat.cast_mul] at ht rw [Nat.cast_sub (by omega : 1 ≤ q), Nat.cast_sub (by omega : 1 ≤ h)] at ht simpa using ht have hqR : (100001 : ℝ) ≤ q := by exact_mod_cast hq have hhR : 2 ≤ (h : ℝ) := by exact_mod_cast h2 have hhm : 0 ≤ (h : ℝ) - 1 := by linarith have hqm : 0 < (q : ℝ) - 1 := by linarith have hbase : (21 / 20 : ℝ)^6 ≤ 2 * (5 / 6 : ℝ)^2 := by norm_num have hpw := pow_le_pow_left₀ (by positivity : (0 : ℝ) ≤ (21 / 20)^6) hbase h have hpweq : ((21 / 20 : ℝ) ^ (3 * h))^2 ≤ 2^h * ((5 / 6 : ℝ)^h)^2 := by have he1 : 3 * h * 2 = 6 * h := by omega have he2 : 2 * h = h * 2 := by omega simpa only [mul_pow, ← pow_mul, he1, he2] using hpw have h2pow : (2 : ℝ)^h = 4 * (2 : ℝ)^(h-2) := by calc (2 : ℝ)^h = 2^(h-2+2) := by congr 1; omega _ = _ := by rw [pow_add]; norm_num; ring have hratio : B * (B * D / C)^2 ≤ 1 := by have hmul : B^3 * D^2 ≤ C^2 := by simpa only [pow_two] using mul_le_mul hb3 hd2 (by positivity : 0 ≤ D^2) (by positivity : 0 ≤ C) have he : B * (B * D / C)^2 = (B^3 * D^2) / C^2 := by ring rw [he, div_le_one (sq_pos_of_pos hc)] exact hmul have hratio' : (q : ℝ) * ((q : ℝ)-1) * 2^(h-2) * (B * D / C)^2 ≤ (h : ℝ) * ((h : ℝ)-1) := by calc _ ≤ ((h : ℝ) * ((h : ℝ)-1) * B) * (B * D / C)^2 := mul_le_mul_of_nonneg_right hchoose (sq_nonneg _) _ = (h : ℝ) * ((h : ℝ)-1) * (B * (B * D / C)^2) := by ring _ ≤ _ := by simpa using mul_le_mul_of_nonneg_left hratio (by positivity : 0 ≤ (h : ℝ)*((h : ℝ)-1)) have hs := mul_le_mul hratio' hpweq (by positivity) (by positivity : 0 ≤ (h : ℝ)*((h : ℝ)-1)) rw [h2pow] at hs have hs' : (q : ℝ) * ((q : ℝ)-1) * (B * D / C)^2 * ((21 / 20 : ℝ) ^ (3*h))^2 ≤ 4 * (h : ℝ) * ((h : ℝ)-1) * ((5/6 : ℝ)^h)^2 := by apply (mul_le_mul_iff_of_pos_left (by positivity : (0 : ℝ) < 2^(h-2))).mp nlinarith only [hs] have hquot : (wtc_term q h)^2 = 9 * (h : ℝ)^2 * (B * D / C)^2 * ((21 / 20 : ℝ)^(3*h))^2 := by dsimp [wtc_term, B, C, D]; push_cast; ring have hupper : (q : ℝ) * ((q : ℝ)-1) * (wtc_term q h)^2 ≤ 36 * (h : ℝ)^4 * ((5/6 : ℝ)^h)^2 := by rw [hquot] have hmul := mul_le_mul_of_nonneg_left hs' (by positivity : 0 ≤ 9*(h : ℝ)^2) have hhh : (h : ℝ) * ((h : ℝ)-1) ≤ (h : ℝ)^2 := by nlinarith have hmul2 := mul_le_mul_of_nonneg_left hhh (by positivity : 0 ≤ 36*(h : ℝ)^2*((5/6 : ℝ)^h)^2) nlinarith only [hmul, hmul2] have hqq : (10000000000 : ℝ) ≤ (q : ℝ) * ((q : ℝ)-1) := by nlinarith [sq_nonneg ((q : ℝ)-100001)] have hsq : (wtc_term q h)^2 ≤ ((3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h)^2 := by apply (mul_le_mul_iff_of_pos_left (by positivity : (0 : ℝ) < (q : ℝ)*((q : ℝ)-1))).mp calc _ ≤ 36 * (h : ℝ)^4 * ((5/6 : ℝ)^h)^2 := hupper _ ≤ _ := by have hx := mul_le_mul_of_nonneg_right hqq (by positivity : 0 ≤ ((3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h)^2) nlinarith only [hx] have hnon : 0 ≤ (3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h := by positivity nlinarith [sq_nonneg (wtc_term q h + (3 / 50000 : ℝ) * (h : ℝ)^2 * (5 / 6 : ℝ)^h)]lemma wtc_mass_bound_sharper (q : ℕ) (hq : 100001 ≤ q) (hodd : q % 2 = 1) : wtc_mass q ≤ 1 / 50 := by have hs : wtc_mass q ≤ (3 / 50000 : ℝ) * ∑ h ∈ Finset.range (q + 1), (h : ℝ)^2 * (5 / 6 : ℝ)^h := by unfold wtc_mass rw [Finset.mul_sum] refine Finset.sum_le_sum fun h hh => ?_ split_ifs with hcond · simpa only [mul_assoc] using wtc_term_bound_sharper q h hq hodd hcond.1 (by omega) · positivity have hg := wtc_geometric_square_sum (q + 1) nlinarithset_option maxHeartbeats 2000000set_option maxRecDepth 100000lemma wtc_factorial_weight (q d : ℕ) : (d.factorial : ℝ) * wtc_wt (q := q) d = (21 / 20 : ℝ)^d / ((3 * q).choose d : ℝ) := by unfold wtc_wt rw [Nat.descFactorial_eq_factorial_mul_choose] push_cast have hf : (d.factorial : ℝ) ≠ 0 := by positivity by_cases hc : ((3 * q).choose d : ℝ) = 0 · simp [hc] · field_simplemma wtc_position_identity (q h : ℕ) (hq : 1 ≤ q) (h1 : 1 ≤ h) : ((q - 1).choose (h - 1) : ℝ) * (3 * q : ℝ) * (((3 * h).factorial * (3 * q / 2).choose (3 * h / 2) : ℕ) : ℝ) * wtc_wt (q := q) (3 * h) = wtc_term q h := by have hi := Nat.add_one_mul_choose_eq (q - 1) (h - 1) rw [Nat.sub_add_cancel hq, Nat.sub_add_cancel h1] at hi have hir : (q : ℝ) * ((q - 1).choose (h - 1) : ℝ) = (q.choose h : ℝ) * (h : ℝ) := by exact_mod_cast hi push_cast calc _ = 3 * ((q : ℝ) * ((q - 1).choose (h - 1) : ℝ)) * ((3 * q / 2).choose (3 * h / 2) : ℝ) * (( (3 * h).factorial : ℝ) * wtc_wt (q := q) (3 * h)) := by ring _ = _ := by rw [hir, wtc_factorial_weight]; unfold wtc_term; push_cast; ringlemma wtc_vertex_coefficient (n s : ℕ) (hodd : n % 2 = 1) (hn : 3 ≤ n) (hs : 3 ≤ s) : (if s % 2 = 1 then (n / 2).choose (s / 2) else 0) + (n - 1) * (n / 2 - 1).choose (s / 2 - 1) = s * (n / 2).choose (s / 2) := by have hi := Nat.add_one_mul_choose_eq (n / 2 - 1) (s / 2 - 1) rw [Nat.sub_add_cancel (by omega : 1 ≤ n / 2), Nat.sub_add_cancel (by omega : 1 ≤ s / 2)] at hi have hn' : n - 1 = 2 * (n / 2) := by omega rw [hn'] split_ifs with hmod · have hs' : s = 2 * (s / 2) + 1 := by omega nlinarith only [hi, congrArg (fun x => x * (n / 2).choose (s / 2)) hs'] · have hs' : s = 2 * (s / 2) := by omega nlinarith only [hi, congrArg (fun x => x * (n / 2).choose (s / 2)) hs']lemma wtc_position_sum_eq (q : ℕ) (hq : 1 ≤ q) : (∑ m ∈ Finset.range q, ((q - 1).choose m : ℝ) * (if m + 1 ≤ q / 2 then (3 * q : ℝ) * (((3 * (m + 1)).factorial * (3 * q / 2).choose (3 * (m + 1) / 2) : ℕ) : ℝ) * wtc_wt (q := q) (3 * (m + 1)) else 0)) = wtc_mass q := by unfold wtc_mass rw [Finset.sum_range_succ'] simp only [not_le.mpr (by omega : 0 < 1), false_and, if_false, add_zero] refine Finset.sum_congr rfl fun m hm => ?_ by_cases hb : m + 1 ≤ q / 2 · simp only [hb, if_true, Nat.add_sub_cancel, show 1 ≤ m + 1 by omega, true_and] simpa only [Nat.add_sub_cancel, mul_assoc] using wtc_position_identity q (m + 1) hq (by omega) · simp only [hb, if_false, and_false, mul_zero]lemma wtc_vertex_sum_eq (q : ℕ) (hq : 1 ≤ q) (hodd : q % 2 = 1) : (∑ m ∈ Finset.range (q + 1), (q.choose m : ℝ) * (if 1 ≤ m ∧ m ≤ q / 2 then ((((3 * m).factorial * (if (3 * m) % 2 = 1 then (3 * q / 2).choose (3 * m / 2) else 0) + (3 * q - 1) * ((3 * m).factorial * (3 * q / 2 - 1).choose (3 * m / 2 - 1)) : ℕ) : ℝ)) * wtc_wt (q := q) (3 * m) else 0)) = wtc_mass q := by unfold wtc_mass refine Finset.sum_congr rfl fun h hh => ?_ by_cases hcond : 1 ≤ h ∧ h ≤ q / 2 · simp only [if_pos hcond] have hn : (3 * q) % 2 = 1 := by omega have hi := wtc_vertex_coefficient (3 * q) (3 * h) hn (by omega) (by omega) have hi' : (3 * h).factorial * (if (3 * h) % 2 = 1 then (3 * q / 2).choose (3 * h / 2) else 0) + (3 * q - 1) * ((3 * h).factorial * (3 * q / 2 - 1).choose (3 * h / 2 - 1)) = (3 * h).factorial * (3 * h * (3 * q / 2).choose (3 * h / 2)) := by nlinarith [hi] rw [hi'] push_cast calc _ = (3 * (h : ℝ)) * (q.choose h : ℝ) * ((3 * q / 2).choose (3 * h / 2) : ℝ) * (((3 * h).factorial : ℝ) * wtc_wt (q := q) (3 * h)) := by ring _ = _ := by rw [wtc_factorial_weight]; unfold wtc_term; push_cast; ring · simp only [if_neg hcond, mul_zero]theorem wild_triple_partition_large_order_mass_bound (q : ℕ) (hq : 100001 ≤ q) (hodd : q % 2 = 1) : (∑ m ∈ Finset.range q, ((Nat.choose (q - 1) m : ℕ) : ℝ) * (if m + 1 ≤ q / 2 then (3 * q : ℝ) * (((3 * (m + 1)).factorial * Nat.choose (3 * q / 2) (3 * (m + 1) / 2) : ℕ) : ℝ) * wtc_wt (q := q) (3 * (m + 1)) else 0) ≤ 1 / 50) ∧ (∑ m ∈ Finset.range (q + 1), ((Nat.choose q m : ℕ) : ℝ) * (if 1 ≤ m ∧ m ≤ q / 2 then ((((3 * m).factorial * (if (3 * m) % 2 = 1 then Nat.choose (3 * q / 2) (3 * m / 2) else 0) + (3 * q - 1) * ((3 * m).factorial * Nat.choose (3 * q / 2 - 1) (3 * m / 2 - 1)) : ℕ) : ℝ)) * wtc_wt (q := q) (3 * m) else 0) ≤ 1 / 50) := by have hmass := wtc_mass_bound_sharper q hq hodd exact ⟨(wtc_position_sum_eq q (by omega)).trans_le hmass, (wtc_vertex_sum_eq q (by omega) hodd).trans_le hmass⟩