@ceccec/millennium-solutions 9.2.8 → 9.3.0

package/speedup.md

@@ -7,7 +7,7 @@ head:
 
 # The real uuidna advantage — measured, in all domains
 
-> Generated from the **2643** ledger entries and **59** domains, recomputed each build. The advantage is
+> Generated from the **2667** ledger entries and **59** domains, recomputed each build. The advantage is
 > real and **measured**, and honestly bounded: a **verification and reuse** complexity reduction — **not**
 > faster original compute, **not** faster than light, **not** quantum.
 
@@ -33,9 +33,9 @@ theorems verify in `⌈log₂N⌉`. So one proof covers the whole [rosetta](/):
 
 ## Discovery lead — a lead, never a verdict
 
-Grouping the 2643 receipts by the digital root of their address, the **sparsest** region is digit
-**3** (268) — a candidate area to examine, not a finding. The ledger is currently held
-at **2643** (the captain's cap): improving a theorem's name or proof heals the record without changing the
+Grouping the 2667 receipts by the digital root of their address, the **sparsest** region is digit
+**3** (270) — a candidate area to examine, not a finding. The ledger is currently held
+at **2667** (the captain's cap): improving a theorem's name or proof heals the record without changing the
 count; discovering a new one would require lifting the cap.
 
 A content-address proves **integrity, not truth**.

package/src/proof/discovered.json

@@ -18966,5 +18966,125 @@
     "key": "lean_sequences_cassinis_identity_for_every_m",
     "name": "lean sequences.lean: cassinis_identity_for_every_m — ∀ m : Nat, fib (2 * m + 1) * fib (2 * m + 3) = fib (2 * m + 2) * fib (2 * m + 2) + 1 ∧ fib (2 * m) * fib (2 * m + 2) + 1 = fib (2 * m + 1) * fib (2 * m + 1) — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
     "receipt": "3b815427-42b8-8b8c-abf3-423b1e268af3"
+  },
+  {
+    "key": "lean_nim_the_xor_of_two_bits_is_their_sum_mod_two",
+    "name": "lean nim.lean: the_xor_of_two_bits_is_their_sum_mod_two — ∀ x y : Nat, (if x % 2 == y % 2 then 0 else 1) = (x + y) % 2 — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "322304d5-f000-8492-8e7f-69ce46520610"
+  },
+  {
+    "key": "lean_nim_xor_commutes_at_every_fuel",
+    "name": "lean nim.lean: xor_commutes_at_every_fuel — ∀ f a b : Nat, xorF f a b = xorF f b a — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "b091d5a8-4e4a-8116-93d8-199afc028903"
+  },
+  {
+    "key": "lean_nim_one_more_step_of_fuel_changes_nothing_once_it_exceeds_an_argument",
+    "name": "lean nim.lean: one_more_step_of_fuel_changes_nothing_once_it_exceeds_an_argument — ∀ f a b : Nat, (a < f ∨ b < f) → xorF (f + 1) a b = xorF f a b — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "adf2ef5a-ef44-84b9-adb7-2edf0437284f"
+  },
+  {
+    "key": "lean_nim_more_fuel_changes_nothing_once_it_exceeds_an_argument",
+    "name": "lean nim.lean: more_fuel_changes_nothing_once_it_exceeds_an_argument — ∀ k f a b : Nat, (a < f ∨ b < f) → xorF (f + k) a b = xorF f a b — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "961d94be-959c-80ff-b0c4-03079551178a"
+  },
+  {
+    "key": "lean_nim_xor_reads_the_same_at_every_sufficient_fuel",
+    "name": "lean nim.lean: xor_reads_the_same_at_every_sufficient_fuel — ∀ g a b : Nat, (a < g ∨ b < g) → xorN a b = xorF g a b — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "e04a45b5-8834-8bc4-bf6e-bff90869081f"
+  },
+  {
+    "key": "lean_nim_a_number_xored_with_itself_is_zero_at_every_fuel",
+    "name": "lean nim.lean: a_number_xored_with_itself_is_zero_at_every_fuel — ∀ f b : Nat, xorF f b b = 0 — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "409e78df-b565-83bb-b177-aa46813c1cb7"
+  },
+  {
+    "key": "lean_nim_xor_splits_off_its_lowest_bit",
+    "name": "lean nim.lean: xor_splits_off_its_lowest_bit — ∀ g x c : Nat, 1 ≤ g → xorF (g + 1) x c = (x + c) % 2 + 2 * xorF g (x / 2) (c / 2) — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "fa8f797f-95fa-80c6-9abc-4bd21309d53a"
+  },
+  {
+    "key": "lean_nim_xor_undoes_itself_when_the_fuel_suffices",
+    "name": "lean nim.lean: xor_undoes_itself_when_the_fuel_suffices — ∀ f a b g : Nat, (a < f ∨ b < f) → b < g → xorF g (xorF f a b) b = a — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "ba799e13-9cb6-8fc8-ab55-3879eb845e7f"
+  },
+  {
+    "key": "lean_nim_xor_is_commutative_for_every_a_b",
+    "name": "lean nim.lean: xor_is_commutative_for_every_a_b — ∀ a b : Nat, xorN a b = xorN b a — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "e62bc467-f90b-8bb2-844c-c44676d6fce0"
+  },
+  {
+    "key": "lean_nim_xor_zero_is_identity_for_every_a",
+    "name": "lean nim.lean: xor_zero_is_identity_for_every_a — ∀ a : Nat, xorN a 0 = a — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "667bd69d-51ea-88da-a48e-ce4a33c4f518"
+  },
+  {
+    "key": "lean_nim_xor_is_its_own_inverse_for_every_a_b",
+    "name": "lean nim.lean: xor_is_its_own_inverse_for_every_a_b — ∀ a b : Nat, xorN (xorN a b) b = a — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "e2f6d0f1-2fa2-832e-b884-1ac9f13160a0"
+  },
+  {
+    "key": "lean_nim_xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves",
+    "name": "lean nim.lean: xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves — ∀ a b : Nat, xorN a b = (a + b) % 2 + 2 * xorN (a / 2) (b / 2) — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "07c467d9-8226-81d4-93b2-612ebd433b2b"
+  },
+  {
+    "key": "lean_nim_xor_is_associative_below_every_bound",
+    "name": "lean nim.lean: xor_is_associative_below_every_bound — ∀ n a b c : Nat, a + b + c ≤ n → xorN (xorN a b) c = xorN a (xorN b c) — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "0c3ef805-2811-8035-a30b-cc03489171c1"
+  },
+  {
+    "key": "lean_nim_xor_is_associative_for_every_a_b_c",
+    "name": "lean nim.lean: xor_is_associative_for_every_a_b_c — ∀ a b c : Nat, xorN (xorN a b) c = xorN a (xorN b c) — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "d2a922d9-f09d-82db-ae54-2f92149512c9"
+  },
+  {
+    "key": "lean_nim_the_nim_sum_is_a_gf2_vector_addition_for_every_a_b_c",
+    "name": "lean nim.lean: the_nim_sum_is_a_gf2_vector_addition_for_every_a_b_c — ∀ a b c : Nat, xorN a b = xorN b a ∧ xorN (xorN a b) c = xorN a (xorN b c) ∧ xorN (xorN a b) b = a ∧ xorN a 0 = a — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "caa5c327-62cf-8171-9507-6453cfdaed0c"
+  },
+  {
+    "key": "lean_sequences_a_bit_sum_over_one_more_position_adds_that_bit",
+    "name": "lean sequences.lean: a_bit_sum_over_one_more_position_adds_that_bit — ∀ F m : Nat, (List.range (F + 1)).foldl (fun a i => a + m / 2 ^ i % 2) 0 = (List.range F).foldl (fun a i => a + m / 2 ^ i % 2) 0 + m / 2 ^ F % 2 — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "f5e71d05-068c-8beb-8470-541d3b4e8e8f"
+  },
+  {
+    "key": "lean_sequences_halving_shifts_every_bit_down_by_one",
+    "name": "lean sequences.lean: halving_shifts_every_bit_down_by_one — ∀ m i : Nat, m / 2 ^ (i + 1) = m / 2 / 2 ^ i — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "a5201a6e-8f86-8eba-b004-a9c565b42613"
+  },
+  {
+    "key": "lean_sequences_the_bits_of_a_number_are_its_low_bit_then_the_bits_of_its_half",
+    "name": "lean sequences.lean: the_bits_of_a_number_are_its_low_bit_then_the_bits_of_its_half — ∀ F m : Nat, (List.range (F + 1)).foldl (fun a i => a + m / 2 ^ i % 2) 0 = m % 2 + (List.range F).foldl (fun a i => a + m / 2 / 2 ^ i % 2) 0 — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "2c2dcdbf-8b9f-8899-a682-4e33aabab23a"
+  },
+  {
+    "key": "lean_sequences_bits_above_the_number_add_nothing",
+    "name": "lean sequences.lean: bits_above_the_number_add_nothing — ∀ k F n : Nat, n < 2 ^ F → (List.range (F + k)).foldl (fun a i => a + n / 2 ^ i % 2) 0 = (List.range F).foldl (fun a i => a + n / 2 ^ i % 2) 0 — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "77439845-748b-811f-9015-a2c92709f76b"
+  },
+  {
+    "key": "lean_sequences_a_number_is_below_two_to_its_own_power",
+    "name": "lean sequences.lean: a_number_is_below_two_to_its_own_power — ∀ n : Nat, n < 2 ^ n — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "4f5a709a-5de5-8424-881d-0c2675c278b0"
+  },
+  {
+    "key": "lean_sequences_a_number_is_below_two_to_its_own_successor",
+    "name": "lean sequences.lean: a_number_is_below_two_to_its_own_successor — ∀ n : Nat, n < 2 ^ (n + 1) — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "586b3088-ac85-8e47-84f8-d82edb5f5aba"
+  },
+  {
+    "key": "lean_sequences_doubling_preserves_the_bit_count_for_every_n",
+    "name": "lean sequences.lean: doubling_preserves_the_bit_count_for_every_n — ∀ n : Nat, popcount (2 * n) = popcount n — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "7006c6f5-3c8d-8158-a54e-6619e289159e"
+  },
+  {
+    "key": "lean_sequences_the_odd_step_sets_exactly_one_further_bit_for_every_n",
+    "name": "lean sequences.lean: the_odd_step_sets_exactly_one_further_bit_for_every_n — ∀ n : Nat, popcount (2 * n + 1) = popcount n + 1 — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "d9da9f38-b47d-836c-8b32-84ce53c93e9c"
+  },
+  {
+    "key": "lean_sequences_thue_morse_doubling_recurrence_for_every_n",
+    "name": "lean sequences.lean: thue_morse_doubling_recurrence_for_every_n — ∀ n : Nat, tm (2 * n) = tm n ∧ tm (2 * n + 1) = 1 - tm n — proved by the Lean kernel for every value (by\n  intro), standard axioms only — checked by lean.ts",
+    "receipt": "50c8ad36-d08a-835c-8d63-559ac849c2cc"
   }
 ]

package/src/proof/nim.lean

@@ -16,13 +16,16 @@ set_option maxHeartbeats 2000000
 -- stated in each name rather than implied.
 --
 -- XOR is imported from Fnv rather than restated: it is the same fuel-bounded structural fold, built because
--- Nat's bitwise operations are well-founded and would drag `propext` into every theorem here.
+-- Nat's bitwise operations are well-founded and would drag `propext` into every theorem here. Its laws are
+-- proved below for every a, b, c, on the standard axioms; the decided rows above stay as their instances.
 
 namespace Nim
 
 open Fnv (xorF)
 
-def xorN (a b : Nat) : Nat := xorF 33 a b
+-- the fuel is a + b + 1, which always exceeds what the fold needs (each step halves both arguments and the fold
+-- stops at the first zero); the earlier fuel of 33 read XOR correctly only below 2³³ — a cap, now gone
+def xorN (a b : Nat) : Nat := xorF (a + b + 1) a b
 -- The board is SMALL and stated, because the game tree is exponential: deciding `lost` explores every play
 -- from every position, and at 8x8 the kernel exceeds its heartbeat budget. Six is what `decide` settles here
 -- in reasonable time. A larger board is not a harder theorem, it is a longer computation — and claiming the
@@ -42,7 +45,7 @@ def lost : Nat → Nat → Nat → Bool
 def isLost (a b : Nat) : Bool := lost (a + b + 1) a b
 
 -- ── BOUTON: a two-heap position is lost for the mover exactly when the heaps are equal — which is exactly
---    when their XOR is zero. Decided over every position on the 8×8 board. ──
+--    when their XOR is zero. Decided over every position on the N×N board, N = 6. ──
 theorem bouton_two_heaps_lost_iff_xor_zero :
   (List.range N).all (fun a => (List.range N).all (fun b =>
     isLost a b == (xorN a b == 0))) := by decide
@@ -102,4 +105,207 @@ set_option maxRecDepth 4000000 in
 theorem bouton_holds_at_every_heap_size_to_six :
   (List.range 7).all (fun a => (List.range 7).all (fun b => isLost a b == (xorN a b == 0))) := by decide
 
+
+-- ── XOR's algebra for every a, b, c — no bound ───────────────────────────────────────────────────────────────
+-- xor_is_its_own_inverse and xor_is_commutative (a, b < 16), xor_zero_is_identity (a < 32) and
+-- the_nim_sum_is_a_gf2_vector_addition (a, b, c < 16) were decided over a range; these prove them for every value.
+-- The fold's fuel is a + b + 1, which always suffices: once it exceeds either argument, more fuel changes nothing,
+-- so xorN reads the same at any sufficient fuel. Everything follows from one step — the low bit of a XOR is the sum
+-- of the low bits mod 2, and the rest is twice the XOR of the halves — by induction. Bouton's theorem itself is
+-- still decided over the N×N board above; a proof of it for every heap is a game-tree induction not done here.
+theorem the_xor_of_two_bits_is_their_sum_mod_two :
+    ∀ x y : Nat, (if x % 2 == y % 2 then 0 else 1) = (x + y) % 2 := by
+  intro x y
+  split
+  · rename_i h
+    have := beq_iff_eq.mp h
+    omega
+  · rename_i h
+    have : x % 2 ≠ y % 2 := fun e => h (beq_iff_eq.mpr e)
+    omega
+
+theorem xor_commutes_at_every_fuel : ∀ f a b : Nat, xorF f a b = xorF f b a := by
+  intro f
+  induction f with
+  | zero => intro a b; cases a <;> cases b <;> rfl
+  | succ f ih =>
+    intro a b
+    cases a with
+    | zero => cases b <;> rfl
+    | succ a =>
+      cases b with
+      | zero => rfl
+      | succ b =>
+        show (if (a + 1) % 2 == (b + 1) % 2 then 0 else 1) + 2 * xorF f ((a + 1) / 2) ((b + 1) / 2) =
+          (if (b + 1) % 2 == (a + 1) % 2 then 0 else 1) + 2 * xorF f ((b + 1) / 2) ((a + 1) / 2)
+        rw [ih ((a + 1) / 2), the_xor_of_two_bits_is_their_sum_mod_two, the_xor_of_two_bits_is_their_sum_mod_two]
+        omega
+
+theorem one_more_step_of_fuel_changes_nothing_once_it_exceeds_an_argument :
+    ∀ f a b : Nat, (a < f ∨ b < f) → xorF (f + 1) a b = xorF f a b := by
+  intro f
+  induction f with
+  | zero => intro a b h; rcases h with h | h <;> exact absurd h (Nat.not_lt_zero _)
+  | succ f ih =>
+    intro a b h
+    cases a with
+    | zero => cases b <;> rfl
+    | succ a =>
+      cases b with
+      | zero => rfl
+      | succ b =>
+        show (if (a + 1) % 2 == (b + 1) % 2 then 0 else 1) + 2 * xorF (f + 1) ((a + 1) / 2) ((b + 1) / 2) =
+          (if (a + 1) % 2 == (b + 1) % 2 then 0 else 1) + 2 * xorF f ((a + 1) / 2) ((b + 1) / 2)
+        have h' : (a + 1) / 2 < f ∨ (b + 1) / 2 < f := by
+          rcases h with h | h
+          · exact Or.inl (by omega)
+          · exact Or.inr (by omega)
+        rw [ih _ _ h']
+
+theorem more_fuel_changes_nothing_once_it_exceeds_an_argument :
+    ∀ k f a b : Nat, (a < f ∨ b < f) → xorF (f + k) a b = xorF f a b := by
+  intro k
+  induction k with
+  | zero => intro f a b _; rfl
+  | succ k ih =>
+    intro f a b h
+    have h' : a < f + k ∨ b < f + k := by
+      rcases h with h | h
+      · exact Or.inl (by omega)
+      · exact Or.inr (by omega)
+    rw [← Nat.add_assoc, one_more_step_of_fuel_changes_nothing_once_it_exceeds_an_argument _ _ _ h', ih f a b h]
+
+theorem xor_reads_the_same_at_every_sufficient_fuel :
+    ∀ g a b : Nat, (a < g ∨ b < g) → xorN a b = xorF g a b := by
+  intro g a b h
+  unfold xorN
+  have h1 : a < a + b + 1 ∨ b < a + b + 1 := Or.inl (by omega)
+  rw [← more_fuel_changes_nothing_once_it_exceeds_an_argument g (a + b + 1) a b h1,
+    ← more_fuel_changes_nothing_once_it_exceeds_an_argument (a + b + 1) g a b h, Nat.add_comm g]
+
+theorem a_number_xored_with_itself_is_zero_at_every_fuel : ∀ f b : Nat, xorF f b b = 0 := by
+  intro f
+  induction f with
+  | zero => intro b; cases b <;> rfl
+  | succ f ih =>
+    intro b
+    cases b with
+    | zero => rfl
+    | succ b =>
+      show (if (b + 1) % 2 == (b + 1) % 2 then 0 else 1) + 2 * xorF f ((b + 1) / 2) ((b + 1) / 2) = 0
+      rw [ih, the_xor_of_two_bits_is_their_sum_mod_two]
+      omega
+
+theorem xor_splits_off_its_lowest_bit :
+    ∀ g x c : Nat, 1 ≤ g → xorF (g + 1) x c = (x + c) % 2 + 2 * xorF g (x / 2) (c / 2) := by
+  intro g x c hg
+  obtain ⟨g', rfl⟩ : ∃ g', g = g' + 1 := ⟨g - 1, by omega⟩
+  cases x with
+  | zero =>
+    rw [show xorF (g' + 1 + 1) 0 c = c by cases c <;> rfl, Nat.zero_div,
+      show xorF (g' + 1) 0 (c / 2) = c / 2 by cases (c / 2) <;> rfl]
+    omega
+  | succ x =>
+    cases c with
+    | zero =>
+      rw [show xorF (g' + 1 + 1) (x + 1) 0 = x + 1 from rfl, Nat.zero_div,
+        show xorF (g' + 1) ((x + 1) / 2) 0 = (x + 1) / 2 by cases ((x + 1) / 2) <;> rfl]
+      omega
+    | succ c =>
+      show (if (x + 1) % 2 == (c + 1) % 2 then 0 else 1) + 2 * xorF (g' + 1) ((x + 1) / 2) ((c + 1) / 2) = _
+      rw [the_xor_of_two_bits_is_their_sum_mod_two]
+
+theorem xor_undoes_itself_when_the_fuel_suffices :
+    ∀ f a b g : Nat, (a < f ∨ b < f) → b < g → xorF g (xorF f a b) b = a := by
+  intro f
+  induction f with
+  | zero => intro a b g h; rcases h with h | h <;> exact absurd h (Nat.not_lt_zero _)
+  | succ f ih =>
+    intro a b g h hg
+    cases a with
+    | zero =>
+      rw [show xorF (f + 1) 0 b = b by cases b <;> rfl]
+      exact a_number_xored_with_itself_is_zero_at_every_fuel g b
+    | succ a =>
+      cases b with
+      | zero =>
+        obtain ⟨g', rfl⟩ : ∃ g', g = g' + 1 := ⟨g - 1, by omega⟩
+        rfl
+      | succ b =>
+        obtain ⟨g', rfl⟩ : ∃ g', g = g' + 1 := ⟨g - 1, by omega⟩
+        have hf : (a + 1) / 2 < f ∨ (b + 1) / 2 < f := by
+          rcases h with h | h
+          · exact Or.inl (by omega)
+          · exact Or.inr (by omega)
+        have ih' := ih ((a + 1) / 2) ((b + 1) / 2) g' hf (by omega)
+        rw [xor_splits_off_its_lowest_bit g' _ _ (by omega)]
+        rw [show xorF (f + 1) (a + 1) (b + 1) = (if (a + 1) % 2 == (b + 1) % 2 then 0 else 1) +
+          2 * xorF f ((a + 1) / 2) ((b + 1) / 2) from rfl, the_xor_of_two_bits_is_their_sum_mod_two]
+        generalize xorF f ((a + 1) / 2) ((b + 1) / 2) = y at ih' ⊢
+        have e : ((a + 1 + (b + 1)) % 2 + 2 * y) / 2 = y := by omega
+        rw [e, ih']
+        omega
+
+theorem xor_is_commutative_for_every_a_b : ∀ a b : Nat, xorN a b = xorN b a := by
+  intro a b
+  unfold xorN
+  rw [Nat.add_comm a b, xor_commutes_at_every_fuel]
+
+theorem xor_zero_is_identity_for_every_a : ∀ a : Nat, xorN a 0 = a := by
+  intro a
+  unfold xorN
+  cases a <;> rfl
+
+theorem xor_is_its_own_inverse_for_every_a_b : ∀ a b : Nat, xorN (xorN a b) b = a := by
+  intro a b
+  rw [xor_reads_the_same_at_every_sufficient_fuel (b + 1) (xorN a b) b (Or.inr (by omega))]
+  unfold xorN
+  exact xor_undoes_itself_when_the_fuel_suffices (a + b + 1) a b (b + 1) (Or.inl (by omega)) (by omega)
+theorem xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves :
+    ∀ a b : Nat, xorN a b = (a + b) % 2 + 2 * xorN (a / 2) (b / 2) := by
+  intro a b
+  rw [xor_reads_the_same_at_every_sufficient_fuel (a + b + 1 + 1) a b (Or.inl (by omega)),
+    xor_splits_off_its_lowest_bit (a + b + 1) a b (by omega),
+    ← xor_reads_the_same_at_every_sufficient_fuel (a + b + 1) (a / 2) (b / 2) (Or.inl (by omega))]
+
+theorem xor_is_associative_below_every_bound :
+    ∀ n a b c : Nat, a + b + c ≤ n → xorN (xorN a b) c = xorN a (xorN b c) := by
+  intro n
+  induction n with
+  | zero =>
+    intro a b c h
+    have ha : a = 0 := by omega
+    have hb : b = 0 := by omega
+    have hc : c = 0 := by omega
+    subst ha hb hc
+    rfl
+  | succ n ih =>
+    intro a b c h
+    if hz : a + b + c = 0 then
+      have ha : a = 0 := by omega
+      have hb : b = 0 := by omega
+      have hc : c = 0 := by omega
+      subst ha hb hc
+      rfl
+    else
+      have ihh := ih (a / 2) (b / 2) (c / 2) (by omega)
+      have hab := xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves a b
+      have hbc := xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves b c
+      have hx2 : xorN a b / 2 = xorN (a / 2) (b / 2) := by omega
+      have hy2 : xorN b c / 2 = xorN (b / 2) (c / 2) := by omega
+      rw [xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves (xorN a b) c,
+        xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves a (xorN b c), hx2, hy2, ihh]
+      omega
+
+theorem xor_is_associative_for_every_a_b_c : ∀ a b c : Nat, xorN (xorN a b) c = xorN a (xorN b c) := by
+  intro a b c
+  exact xor_is_associative_below_every_bound (a + b + c) a b c (Nat.le_refl _)
+
+theorem the_nim_sum_is_a_gf2_vector_addition_for_every_a_b_c :
+    ∀ a b c : Nat, xorN a b = xorN b a ∧ xorN (xorN a b) c = xorN a (xorN b c) ∧
+      xorN (xorN a b) b = a ∧ xorN a 0 = a := by
+  intro a b c
+  exact ⟨xor_is_commutative_for_every_a_b a b, xor_is_associative_for_every_a_b_c a b c,
+    xor_is_its_own_inverse_for_every_a_b a b, xor_zero_is_identity_for_every_a a⟩
+
 end Nim

package/src/proof/sequences.lean

@@ -1,5 +1,6 @@
 import Fnv
 set_option maxRecDepth 4000000
+set_option maxHeartbeats 2000000
 -- title: Sequences
 -- wing: the ring
 -- prior_art: named
@@ -79,13 +80,15 @@ theorem pisano_twentyfour_is_four_sixes :
 -- Withdrawn for want of a proof, not because it is false — and it HAS a decidable form. t(n) is the parity
 -- of the 1-bits of n. Doubling shifts every bit one place and introduces no new one, so the popcount is
 -- unchanged and t(2n) = t(n). 2n+1 sets exactly one further bit, so the parity flips: t(2n+1) = 1 − t(n).
--- Both are decided below over a stated finite domain, which is what the withdrawal said was missing.
-def popcount (n : Nat) : Nat := (List.range 12).foldl (fun a i => a + n / 2 ^ i % 2) 0
+-- Both are decided below over a stated finite domain, which is what the withdrawal said was missing, and
+-- proved for every n at the end of this file.
+-- The fuel is n + 1, which covers every bit of every n (n < 2^(n+1)); the earlier fuel of 12 counted only the
+-- low twelve bits — a cap every statement here had to stay under.
+def popcount (n : Nat) : Nat := (List.range (n + 1)).foldl (fun a i => a + n / 2 ^ i % 2) 0
 def tm (n : Nat) : Nat := popcount n % 2
 
--- Domain: n < 200, so 2n+1 < 400 < 2^12 = 4096 and the twelve-bit popcount is exact over every value the
--- statement touches. The bound is stated rather than assumed, because a popcount that silently truncated
--- would make this hold for the wrong reason.
+-- Domain: n < 200 — the decided instance of the law proved for every n at the end of this file. The popcount
+-- no longer stops at twelve bits, so no statement here can hold for the wrong reason at any size.
 theorem thue_morse_doubling_recurrence :
   (List.range 200).all (fun n => tm (2 * n) == tm n && tm (2 * n + 1) == 1 - tm n) := by decide
 
@@ -133,4 +136,84 @@ theorem cassinis_identity_for_every_m :
       try simp only [Nat.mul_comm _ (OfNat.ofNat _), Nat.mul_left_comm _ (OfNat.ofNat _), Nat.reduceMul] at *
       all_goals omega
 
+
+-- ── the popcount rows above, proved for every n — no bound ───────────────────────────────────────────────
+-- thue_morse_doubling_recurrence, doubling_preserves_the_parity_of_the_bits and the_odd_step_sets_exactly_one_
+-- further_bit were decided over n < 200; these prove them for every n. The bits of m are its low bit and then the
+-- bits of m / 2; positions above the number add nothing; so doubling keeps the count and the odd step adds one.
+theorem a_bit_sum_over_one_more_position_adds_that_bit :
+    ∀ F m : Nat, (List.range (F + 1)).foldl (fun a i => a + m / 2 ^ i % 2) 0 =
+      (List.range F).foldl (fun a i => a + m / 2 ^ i % 2) 0 + m / 2 ^ F % 2 := by
+  intro F m
+  rw [List.range_succ, List.foldl_append]
+  rfl
+
+theorem halving_shifts_every_bit_down_by_one : ∀ m i : Nat, m / 2 ^ (i + 1) = m / 2 / 2 ^ i := by
+  intro m i
+  rw [Nat.div_div_eq_div_mul, Nat.pow_succ, Nat.mul_comm]
+
+theorem the_bits_of_a_number_are_its_low_bit_then_the_bits_of_its_half :
+    ∀ F m : Nat, (List.range (F + 1)).foldl (fun a i => a + m / 2 ^ i % 2) 0 =
+      m % 2 + (List.range F).foldl (fun a i => a + m / 2 / 2 ^ i % 2) 0 := by
+  intro F
+  induction F with
+  | zero => intro m; show 0 + m / 2 ^ 0 % 2 = m % 2 + 0; rw [Nat.pow_zero, Nat.div_one]; omega
+  | succ F ih =>
+    intro m
+    rw [a_bit_sum_over_one_more_position_adds_that_bit (F + 1) m, ih m,
+      a_bit_sum_over_one_more_position_adds_that_bit F (m / 2), halving_shifts_every_bit_down_by_one]
+    omega
+
+theorem bits_above_the_number_add_nothing :
+    ∀ k F n : Nat, n < 2 ^ F → (List.range (F + k)).foldl (fun a i => a + n / 2 ^ i % 2) 0 =
+      (List.range F).foldl (fun a i => a + n / 2 ^ i % 2) 0 := by
+  intro k
+  induction k with
+  | zero => intro F n _; rfl
+  | succ k ih =>
+    intro F n h
+    rw [← Nat.add_assoc, a_bit_sum_over_one_more_position_adds_that_bit, ih F n h]
+    have hp : 2 ^ F ≤ 2 ^ (F + k) := Nat.pow_le_pow_right (by decide) (Nat.le_add_right F k)
+    rw [Nat.div_eq_of_lt (by omega), Nat.zero_mod, Nat.add_zero]
+
+theorem a_number_is_below_two_to_its_own_power : ∀ n : Nat, n < 2 ^ n := by
+  intro n
+  induction n with
+  | zero => decide
+  | succ n ih =>
+    rw [Nat.pow_succ]
+    omega
+
+theorem a_number_is_below_two_to_its_own_successor : ∀ n : Nat, n < 2 ^ (n + 1) := by
+  intro n
+  have h := a_number_is_below_two_to_its_own_power n
+  rw [Nat.pow_succ]
+  omega
+
+theorem doubling_preserves_the_bit_count_for_every_n : ∀ n : Nat, popcount (2 * n) = popcount n := by
+  intro n
+  unfold popcount
+  rw [the_bits_of_a_number_are_its_low_bit_then_the_bits_of_its_half (2 * n) (2 * n),
+    show 2 * n / 2 = n by omega, show 2 * n % 2 = 0 by omega, Nat.zero_add]
+  cases n with
+  | zero => rfl
+  | succ m =>
+    have hlt : m + 1 < 2 ^ (m + 1 + 1) := a_number_is_below_two_to_its_own_successor (m + 1)
+    rw [show 2 * (m + 1) = m + 1 + 1 + m by omega, bits_above_the_number_add_nothing m (m + 1 + 1) (m + 1) hlt]
+
+theorem the_odd_step_sets_exactly_one_further_bit_for_every_n : ∀ n : Nat, popcount (2 * n + 1) = popcount n + 1 := by
+  intro n
+  unfold popcount
+  rw [the_bits_of_a_number_are_its_low_bit_then_the_bits_of_its_half (2 * n + 1) (2 * n + 1),
+    show (2 * n + 1) / 2 = n by omega, show (2 * n + 1) % 2 = 1 by omega]
+  have hlt : n < 2 ^ (n + 1) := a_number_is_below_two_to_its_own_successor n
+  rw [show 2 * n + 1 = n + 1 + n by omega, bits_above_the_number_add_nothing n (n + 1) n hlt]
+  omega
+
+theorem thue_morse_doubling_recurrence_for_every_n : ∀ n : Nat, tm (2 * n) = tm n ∧ tm (2 * n + 1) = 1 - tm n := by
+  intro n
+  unfold tm
+  rw [doubling_preserves_the_bit_count_for_every_n, the_odd_step_sets_exactly_one_further_bit_for_every_n]
+  exact ⟨rfl, by omega⟩
+
 end Sequences

package/src/proof/trial-all.json

@@ -15856,5 +15856,149 @@
     "verdict": "UNVERIFIED",
     "ground": "no candidate in the current space — nothing to recompute",
     "receipt": "3b815427-42b8-8b8c-abf3-423b1e268af3"
+  },
+  {
+    "key": "lean_nim_the_xor_of_two_bits_is_their_sum_mod_two",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "322304d5-f000-8492-8e7f-69ce46520610"
+  },
+  {
+    "key": "lean_nim_xor_commutes_at_every_fuel",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "b091d5a8-4e4a-8116-93d8-199afc028903"
+  },
+  {
+    "key": "lean_nim_one_more_step_of_fuel_changes_nothing_once_it_exceeds_an_argument",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "adf2ef5a-ef44-84b9-adb7-2edf0437284f"
+  },
+  {
+    "key": "lean_nim_more_fuel_changes_nothing_once_it_exceeds_an_argument",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "961d94be-959c-80ff-b0c4-03079551178a"
+  },
+  {
+    "key": "lean_nim_xor_reads_the_same_at_every_sufficient_fuel",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "e04a45b5-8834-8bc4-bf6e-bff90869081f"
+  },
+  {
+    "key": "lean_nim_a_number_xored_with_itself_is_zero_at_every_fuel",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "409e78df-b565-83bb-b177-aa46813c1cb7"
+  },
+  {
+    "key": "lean_nim_xor_splits_off_its_lowest_bit",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "fa8f797f-95fa-80c6-9abc-4bd21309d53a"
+  },
+  {
+    "key": "lean_nim_xor_undoes_itself_when_the_fuel_suffices",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "ba799e13-9cb6-8fc8-ab55-3879eb845e7f"
+  },
+  {
+    "key": "lean_nim_xor_is_commutative_for_every_a_b",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "e62bc467-f90b-8bb2-844c-c44676d6fce0"
+  },
+  {
+    "key": "lean_nim_xor_zero_is_identity_for_every_a",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "667bd69d-51ea-88da-a48e-ce4a33c4f518"
+  },
+  {
+    "key": "lean_nim_xor_is_its_own_inverse_for_every_a_b",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "e2f6d0f1-2fa2-832e-b884-1ac9f13160a0"
+  },
+  {
+    "key": "lean_nim_xor_is_the_sum_of_the_bits_plus_twice_the_xor_of_the_halves",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "07c467d9-8226-81d4-93b2-612ebd433b2b"
+  },
+  {
+    "key": "lean_nim_xor_is_associative_below_every_bound",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "0c3ef805-2811-8035-a30b-cc03489171c1"
+  },
+  {
+    "key": "lean_nim_xor_is_associative_for_every_a_b_c",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "d2a922d9-f09d-82db-ae54-2f92149512c9"
+  },
+  {
+    "key": "lean_nim_the_nim_sum_is_a_gf2_vector_addition_for_every_a_b_c",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "caa5c327-62cf-8171-9507-6453cfdaed0c"
+  },
+  {
+    "key": "lean_sequences_a_bit_sum_over_one_more_position_adds_that_bit",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "f5e71d05-068c-8beb-8470-541d3b4e8e8f"
+  },
+  {
+    "key": "lean_sequences_halving_shifts_every_bit_down_by_one",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "a5201a6e-8f86-8eba-b004-a9c565b42613"
+  },
+  {
+    "key": "lean_sequences_the_bits_of_a_number_are_its_low_bit_then_the_bits_of_its_half",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "2c2dcdbf-8b9f-8899-a682-4e33aabab23a"
+  },
+  {
+    "key": "lean_sequences_bits_above_the_number_add_nothing",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "77439845-748b-811f-9015-a2c92709f76b"
+  },
+  {
+    "key": "lean_sequences_a_number_is_below_two_to_its_own_power",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "4f5a709a-5de5-8424-881d-0c2675c278b0"
+  },
+  {
+    "key": "lean_sequences_a_number_is_below_two_to_its_own_successor",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "586b3088-ac85-8e47-84f8-d82edb5f5aba"
+  },
+  {
+    "key": "lean_sequences_doubling_preserves_the_bit_count_for_every_n",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "7006c6f5-3c8d-8158-a54e-6619e289159e"
+  },
+  {
+    "key": "lean_sequences_the_odd_step_sets_exactly_one_further_bit_for_every_n",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "d9da9f38-b47d-836c-8b32-84ce53c93e9c"
+  },
+  {
+    "key": "lean_sequences_thue_morse_doubling_recurrence_for_every_n",
+    "verdict": "UNVERIFIED",
+    "ground": "no candidate in the current space — nothing to recompute",
+    "receipt": "50c8ad36-d08a-835c-8d63-559ac849c2cc"
   }
 ]