Documentation

MIPRE.Background.Repetition.CommutingRepetition.Prerounding.PriorAlignment

The Alice reverse datum as (base, interior cut) #

@[reducible, inline]

An Alice reverse base: the Alice block L_X ⊆ Dᶜ (the Bob block is L_Y⁺ = Dᶜ \ L_X), the forward order π_X, the reverse order π_Y, and the forward cut k_X. The interior cut k_Y ∈ Fin |L_Y⁺| is split off (aliceSigmaEquiv), because the alignment integrand is summed over it uniformly (eq IA-size-bias-calculation).

Equations
Instances For
    @[reducible, inline]
    abbrev CommutingRepetition.AliceBase.LX {n : } {D : Finset (Fin n)} (b : AliceBase n D) :

    The Alice block L_X.

    Equations
    Instances For
      @[reducible, inline]
      abbrev CommutingRepetition.AliceBase.LYp {n : } {D : Finset (Fin n)} (b : AliceBase n D) :

      The Bob block L_Y⁺ = Dᶜ \ L_X.

      Equations
      Instances For
        @[reducible, inline]
        abbrev CommutingRepetition.AliceBase.πY {n : } {D : Finset (Fin n)} (b : AliceBase n D) :
        Fin (D \ b.fst).card ↥(D \ b.fst)

        The reverse order of the Bob block.

        Equations
        Instances For
          @[reducible, inline]
          abbrev CommutingRepetition.AliceBase.SA {n : } {D : Finset (Fin n)} (b : AliceBase n D) :

          The Alice background S_A = D ∪ L_X of the scenario martingale.

          Equations
          Instances For
            @[reducible, inline]
            abbrev CommutingRepetition.AliceBase.SB {n : } {D : Finset (Fin n)} (b : AliceBase n D) :

            Bob's fixed revealed set D ∪ L_Y⁺ ∪ π_X^{≤ k_X} (the set behind aliceFixedBobEffect_eq_effectiveK, cut-independent).

            Equations
            Instances For
              noncomputable def CommutingRepetition.AliceBase.β {n : } {D : Finset (Fin n)} (b : AliceBase n D) :

              The base weight β = 2^{-m}·(1/|L_Y⁺|!)·(1/|L_X|!)·(1/(|L_X|+1)), so that the size-biased law is (2/m)·β (eq size-biased-partition).

              Equations
              Instances For
                def CommutingRepetition.mkAliceDatum {n : } {D : Finset (Fin n)} (b : AliceBase n D) (k : Fin (D \ b.fst).card) :

                The reverse datum with base b and interior cut k.

                Equations
                Instances For

                  The (base, cut) presentation of the Alice reverse datum.

                  Equations
                  • One or more equations did not get rendered due to their size.
                  Instances For
                    theorem CommutingRepetition.mkAliceDatum_law {n : } {D : Finset (Fin n)} (hm : D.card < n) (b : AliceBase n D) (k : Fin (D \ b.fst).card) :
                    (mkAliceDatum b k).law = 2 / ↑(n - D.card) * b.β

                    The size-biased law of a datum in (base, cut) form: (2/m)·β(base) (the factor (2N_A/m)·(1/N_A) = 2/m of eq size-biased-partition; the cut's existence forces N_A > 0).

                    theorem CommutingRepetition.AliceBase.sum_β {n : } (D : Finset (Fin n)) (hm : D.card < n) :
                    b : AliceBase n D, b.β = 1

                    The base weights sum to one: fair partitions 2^{-m} over the 2^m Alice blocks, uniform orders and forward cut (eq size-biased-partition, "the last sum of the fair-partition weights is at most one" — here exact).

                    Label identities for a datum in (base, cut) form (STEP 1 of the roadmap) #

                    theorem CommutingRepetition.mkAliceDatum_liveIdx {n : } {D : Finset (Fin n)} (b : AliceBase n D) (k : Fin (D \ b.fst).card) :
                    (mkAliceDatum b k).liveIdx = (b.πY k)
                    theorem CommutingRepetition.RevealDatum.CX_of_alice {n : } {D : Finset (Fin n)} (r : RevealDatum n D) (b : AliceBase n D) (k : Fin (D \ b.fst).card) (hLX : r.LX = (mkAliceDatum b k).LX) (hpre : r.prefixY = (mkAliceDatum b k).revealPrefix (mkAliceDatum b k).kY.castSucc) :
                    r.CX = b.SA ordPrefix b.πY k

                    C_X = S_A ∪ π_Y^{≤ k_Y} for the forward datum wired to (b, k).

                    theorem CommutingRepetition.RevealDatum.insert_CX_of_alice {n : } {D : Finset (Fin n)} (r : RevealDatum n D) (b : AliceBase n D) (k : Fin (D \ b.fst).card) (hi : r.i = (mkAliceDatum b k).liveIdx) (hLX : r.LX = (mkAliceDatum b k).LX) (hpre : r.prefixY = (mkAliceDatum b k).revealPrefix (mkAliceDatum b k).kY.castSucc) :
                    insert r.i r.CX = insert (↑(b.πY k)) (b.SA ordPrefix b.πY k)

                    {i} ∪ C_X = {π_Y(k_Y)} ∪ (S_A ∪ π_Y^{≤ k_Y}).

                    theorem CommutingRepetition.RevealDatum.insert_CY_of_alice {n : } {D : Finset (Fin n)} (r : RevealDatum n D) (b : AliceBase n D) (k : Fin (D \ b.fst).card) (hi : r.i = (mkAliceDatum b k).liveIdx) (hLY : r.LY = (mkAliceDatum b k).LYp.erase (mkAliceDatum b k).liveIdx) (hpx : r.prefixX = (mkAliceDatum b k).alicePrefixX) :
                    insert r.i r.CY = b.SB

                    The KEY collapse: {i} ∪ C_Y = D ∪ L_Y⁺ ∪ π_X^{≤ k_X} = S_B, independent of the cut k_Y (the set behind aliceFixedBobEffect_eq_effectiveK).

                    Unnormalizing the block law #

                    theorem CommutingRepetition.block_unnormalize {n : } {X Y : Type} [Fintype X] [Fintype Y] [DecidableEq X] [DecidableEq Y] [Nonempty X] [Nonempty Y] (μ : XY) ( : ∀ (x : X) (y : Y), 0 μ x y) (L : Finset (Fin n)) (yw : Fin nY) (F : (LX)) (Bnd : ) (h : c : L, v : X, μ v (yw c) 0ω : LX, ((∏ c : L, μ (ω c) (yw c)) / c : L, v : X, μ v (yw c)) * F ω Bnd) :
                    ω : LX, (∏ c : L, μ (ω c) (yw c)) * F ω (∏ c : L, v : X, μ v (yw c)) * Bnd

                    Convert a bound on a block-law-weighted sum (the martingale's normalized law ∏μ / M₀) into a bound on the raw-prior-weighted sum, with the block mass M₀ as the factor; at M₀ = 0 every raw block weight vanishes.

                    The size-bias identities (STEP 4 of the roadmap) #

                    theorem CommutingRepetition.TracialStrategy.sum_pairing_eq_coreMass {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] (S : TracialStrategy (Fin nX) (Fin nY) (Fin nA) (Fin nB)) {D : Finset (Fin n)} (μ : XY) ( : ∀ (x : X) (y : Y), 0 μ x y) (w : (Fin nX)(Fin nY)(DA)(DB)) (hwD : ∀ (xw xw' : Fin nX) (yw yw' : Fin nY) (zA : DA) (zB : DB), agreesOn D xw xw'agreesOn D yw yw'w xw yw zA zB = w xw' yw' zA zB) (SA SB : Finset (Fin n)) (hDA : DSA) (hDB : DSB) (hcov : SA SB = Finset.univ) :
                    xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * (S.M.τ (star S.σ * (S.setEffectA D SA μ xw yw (mkExtA D zD.1) * S.σ * S.setEffectB D SB μ xw yw (mkExtB D zD.2)))).re = xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * (S.M.τ (star S.σ * (S.coreEffectA D xw (mkExtA D zD.1) * S.σ * S.coreEffectB D yw (mkExtB D zD.2)))).re

                    Collapse of the prior pairing to the core mass: for revealed sets S_A ⊇ D, S_B ⊇ D covering every coordinate, the prior-weighted pairing of the two revealed-set effects, summed over the core answers with the D-measurable weight, is exactly the weighted core mass p (pairSum_eq_core, the tower property). This is the identity ∑ wᵢ hᵢ = p behind eq accepted-word-entropy.

                    theorem CommutingRepetition.TracialStrategy.sum_prior_weight_le {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] {D : Finset (Fin n)} (μ : XY) ( : ∀ (x : X) (y : Y), 0 μ x y) (hμsum : x : X, y : Y, μ x y = 1) (w : (Fin nX)(Fin nY)(DA)(DB)) (hw1 : ∀ (xw : Fin nX) (yw : Fin nY) (zA : DA) (zB : DB), w xw yw zA zB 1) :
                    xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 ((Fintype.card A) * (Fintype.card B)) ^ D.card

                    Total prior weight: the [0,1]-weighted prior mass over the core answers is at most the number (|A||B|)^{|D|} of core answer words ("the accepted words have number at most e^{s₀}").

                    Sum rearrangements #

                    theorem CommutingRepetition.sum4_rearrange {α β γ δ : Type} [Fintype α] [Fintype β] [Fintype γ] [Fintype δ] (Q : βγ) (w : βγδ) (X : αβγδ) :
                    a : α, b : β, c : γ, Q b c * d : δ, w b c d * X a b c d = c : γ, d : δ, b : β, Q b c * (w b c d * a : α, X a b c d)
                    theorem CommutingRepetition.sum3_rearrange {β γ δ : Type} [Fintype β] [Fintype γ] [Fintype δ] (Q : βγ) (w Y : βγδ) :
                    b : β, c : γ, Q b c * d : δ, w b c d * Y b c d = c : γ, d : δ, b : β, Q b c * (w b c d * Y b c d)
                    theorem CommutingRepetition.sum_split_comm {n : } {X : Type} [Fintype X] [DecidableEq X] [Nonempty X] (L : Finset (Fin n)) (F : (Fin nX)) :
                    xw : Fin nX, F xw = g : { j : Fin n // jL }X, ω : LX, F ((wordSplit L X).symm (ω, g))

                    Split a word sum at a block and put the off-block values outside.

                    The per-base telescoping bound (STEPS 2–3 of the roadmap) #

                    noncomputable def CommutingRepetition.TracialStrategy.alignIncA {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] (S : TracialStrategy (Fin nX) (Fin nY) (Fin nA) (Fin nB)) {D : Finset (Fin n)} {Af Bf : Type} [Fintype Af] [Fintype Bf] {Ffam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nA)AfS.M.A} {Gfam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nB)BfS.M.A} (R : ResolverArena S.M Ffam Gfam) (b : AliceBase n D) (k : Fin (D \ b.fst).card) (xw : Fin nX) (yw : Fin nY) (zD : (DA) × (DB)) :

                    The alignment increment at base b and interior cut k: the squared branch-vector increment between the Alice labels at S_A ∪ π_Y^{≤k} and at {π_Y(k)} ∪ S_A ∪ π_Y^{≤k}, against Bob's cut-independent label at S_B (the alignCostA integrand read through insert_CX_of_alice / insert_CY_of_alice).

                    Equations
                    • One or more equations did not get rendered due to their size.
                    Instances For
                      noncomputable def CommutingRepetition.TracialStrategy.alignEntA {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty A] [Nonempty B] (S : TracialStrategy (Fin nX) (Fin nY) (Fin nA) (Fin nB)) {D : Finset (Fin n)} (μ : XY) (b : AliceBase n D) (xw : Fin nX) (yw : Fin nY) (zD : (DA) × (DB)) :

                      The scalar entropy H₁ of the initial branch pairing at base b: H₁(re τ(σ* F_{S_A} σ G_{S_B}))H₁(p_z(U_A)) of eq random-martingale-increment.

                      Equations
                      • One or more equations did not get rendered due to their size.
                      Instances For
                        theorem CommutingRepetition.TracialStrategy.cutSumA_block_le {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] (S : TracialStrategy (Fin nX) (Fin nY) (Fin nA) (Fin nB)) {D : Finset (Fin n)} {Af Bf : Type} [Fintype Af] [Fintype Bf] (μ : XY) ( : ∀ (x : X) (y : Y), 0 μ x y) {Ffam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nA)AfS.M.A} {Gfam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nB)BfS.M.A} (R : ResolverArena S.M Ffam Gfam) (hcol : R.ColEntropyBudget) (htotF : ∀ (s : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nA)), a : Af, Ffam s a = S.setEffectA D s.1 μ s.2.1 s.2.2.1 s.2.2.2) (htotG : ∀ (t : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nB)), b : Bf, Gfam t b = S.setEffectB D t.1 μ t.2.1 t.2.2.1 t.2.2.2) (w : (Fin nX)(Fin nY)(DA)(DB)) (hw0 : ∀ (xw : Fin nX) (yw : Fin nY) (zA : DA) (zB : DB), 0 w xw yw zA zB) (hwD : ∀ (xw xw' : Fin nX) (yw yw' : Fin nY) (zA : DA) (zB : DB), agreesOn D xw xw'agreesOn D yw yw'w xw yw zA zB = w xw' yw' zA zB) (b : AliceBase n D) (yw : Fin nY) (zD : (DA) × (DB)) (g : { j : Fin n // jb.LYp }X) :
                        ω : b.LYpX, (∏ j : Fin n, μ ((wordSplit b.LYp X).symm (ω, g) j) (yw j)) * (w ((wordSplit b.LYp X).symm (ω, g)) yw zD.1 zD.2 * k : Fin (D \ b.fst).card, S.alignIncA R b k ((wordSplit b.LYp X).symm (ω, g)) yw zD) ω : b.LYpX, (∏ j : Fin n, μ ((wordSplit b.LYp X).symm (ω, g) j) (yw j)) * (w ((wordSplit b.LYp X).symm (ω, g)) yw zD.1 zD.2 * S.alignEntA μ b ((wordSplit b.LYp X).symm (ω, g)) yw zD)

                        Block bound at a fixed Alice background (eqs random-martingale-increment summed over the uniform cut, via scenMartA_colBudget_mkALabel): for fixed Bob word yw, core answers zD and off-block Alice values g, the raw-prior-weighted sum over the Bob-block values ω of the cut-summed increments is at most the same weighted sum of the initial-pairing entropy.

                        theorem CommutingRepetition.TracialStrategy.cutSumA_le {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] (S : TracialStrategy (Fin nX) (Fin nY) (Fin nA) (Fin nB)) {D : Finset (Fin n)} {Af Bf : Type} [Fintype Af] [Fintype Bf] (μ : XY) ( : ∀ (x : X) (y : Y), 0 μ x y) {Ffam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nA)AfS.M.A} {Gfam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nB)BfS.M.A} (R : ResolverArena S.M Ffam Gfam) (hcol : R.ColEntropyBudget) (htotF : ∀ (s : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nA)), a : Af, Ffam s a = S.setEffectA D s.1 μ s.2.1 s.2.2.1 s.2.2.2) (htotG : ∀ (t : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nB)), b : Bf, Gfam t b = S.setEffectB D t.1 μ t.2.1 t.2.2.1 t.2.2.2) (w : (Fin nX)(Fin nY)(DA)(DB)) (hw0 : ∀ (xw : Fin nX) (yw : Fin nY) (zA : DA) (zB : DB), 0 w xw yw zA zB) (hwD : ∀ (xw xw' : Fin nX) (yw yw' : Fin nY) (zA : DA) (zB : DB), agreesOn D xw xw'agreesOn D yw yw'w xw yw zA zB = w xw' yw' zA zB) (b : AliceBase n D) :
                        k : Fin (D \ b.fst).card, xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * S.alignIncA R b k xw yw zD xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * S.alignEntA μ b xw yw zD

                        Per-base telescoped bound (STEPS 2–3 of the roadmap): at a fixed base, the uniform interior cut sums the alignment increments into the column budget, so the prior-weighted cut-sum is at most the prior-weighted initial-pairing entropy.

                        The size-bias entropy step (STEP 4 of the roadmap) #

                        theorem CommutingRepetition.TracialStrategy.sum_base_reshape {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] {D : Finset (Fin n)} (μ : XY) (w : (Fin nX)(Fin nY)(DA)(DB)) (F : AliceBase n D(Fin nX)(Fin nY)(DA) × (DB)) :
                        t : AliceBase n D × (Fin nX) × (Fin nY) × (DA) × (DB), t.1.β * ((∏ j : Fin n, μ (t.2.1 j) (t.2.2.1 j)) * w t.2.1 t.2.2.1 t.2.2.2.1 t.2.2.2.2) * F t.1 t.2.1 t.2.2.1 t.2.2.2 = b : AliceBase n D, b.β * xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * F b xw yw zD

                        Reshaping the base-indexed flat sums into nested sums.

                        theorem CommutingRepetition.TracialStrategy.sizeBiasA {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] (S : TracialStrategy (Fin nX) (Fin nY) (Fin nA) (Fin nB)) {D : Finset (Fin n)} (μ : XY) ( : ∀ (x : X) (y : Y), 0 μ x y) (hμsum : x : X, y : Y, μ x y = 1) (w : (Fin nX)(Fin nY)(DA)(DB)) (hw0 : ∀ (xw : Fin nX) (yw : Fin nY) (zA : DA) (zB : DB), 0 w xw yw zA zB) (hw1 : ∀ (xw : Fin nX) (yw : Fin nY) (zA : DA) (zB : DB), w xw yw zA zB 1) (hwD : ∀ (xw xw' : Fin nX) (yw yw' : Fin nY) (zA : DA) (zB : DB), agreesOn D xw xw'agreesOn D yw yw'w xw yw zA zB = w xw' yw' zA zB) {p : } (hp : p = xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * (S.M.τ (star S.σ * (S.coreEffectA D xw (mkExtA D zD.1) * S.σ * S.coreEffectB D yw (mkExtB D zD.2)))).re) (hppos : 0 < p) (hm : D.card < n) :
                        b : AliceBase n D, b.β * xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * S.alignEntA μ b xw yw zD p * Real.log (((Fintype.card A) * (Fintype.card B)) ^ D.card / p)

                        The size-bias entropy bound (eqs accepted-word-logsum, accepted-word-entropy, in the weighted form of finite_weighted_entropy_le_of_weight_bound): the base-weighted prior average of the initial-pairing entropies is at most p·log(N/p), N = (|A||B|)^{|D|}, because the same weights average the pairings themselves to exactly p (sum_pairing_eq_coreMass) and have total mass at most N (sum_prior_weight_le, AliceBase.sum_β).

                        The Alice conjunct #

                        theorem CommutingRepetition.TracialStrategy.alignSumA_le {n : } {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B] [DecidableEq X] [DecidableEq Y] [DecidableEq A] [DecidableEq B] [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B] (S : TracialStrategy (Fin nX) (Fin nY) (Fin nA) (Fin nB)) {D : Finset (Fin n)} {Af Bf : Type} [Fintype Af] [Fintype Bf] (μ : XY) ( : ∀ (x : X) (y : Y), 0 μ x y) (hμsum : x : X, y : Y, μ x y = 1) {Ffam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nA)AfS.M.A} {Gfam : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nB)BfS.M.A} (R : ResolverArena S.M Ffam Gfam) (htotF : ∀ (s : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nA)), a : Af, Ffam s a = S.setEffectA D s.1 μ s.2.1 s.2.2.1 s.2.2.2) (htotG : ∀ (t : Finset (Fin n) × (Fin nX) × (Fin nY) × (Fin nB)), b : Bf, Gfam t b = S.setEffectB D t.1 μ t.2.1 t.2.2.1 t.2.2.2) (hcol : R.ColEntropyBudget) (w : (Fin nX)(Fin nY)(DA)(DB)) (hw0 : ∀ (xw : Fin nX) (yw : Fin nY) (zA : DA) (zB : DB), 0 w xw yw zA zB) (hw1 : ∀ (xw : Fin nX) (yw : Fin nY) (zA : DA) (zB : DB), w xw yw zA zB 1) (hwD : ∀ (xw xw' : Fin nX) (yw yw' : Fin nY) (zA : DA) (zB : DB), agreesOn D xw xw'agreesOn D yw yw'w xw yw zA zB = w xw' yw' zA zB) {p : } (hp : p = xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * (S.M.τ (star S.σ * (S.coreEffectA D xw (mkExtA D zD.1) * S.σ * S.coreEffectB D yw (mkExtB D zD.2)))).re) (hppos : 0 < p) (hm : D.card < n) :
                        r : RevealDatum n D, r.revealLaw * xw : Fin nX, yw : Fin nY, (∏ j : Fin n, μ (xw j) (yw j)) * zD : (DA) × (DB), w xw yw zD.1 zD.2 * R.branch S.σ (mkALabel D (insert r.i r.CX) xw yw zD.1) (mkBLabel D (insert r.i r.CY) xw yw zD.2) - R.branch S.σ (mkALabel D r.CX xw yw zD.1) (mkBLabel D (insert r.i r.CY) xw yw zD.2) ^ 2 2 * p * (Real.log p⁻¹ + D.card * Real.log ((Fintype.card A) * (Fintype.card B))) / (n - D.card)

                        The prior Alice alignment cost is at most 2p(t₀ + s₀)/m (node 1.2.9, eqs IA-size-bias-calculation, prior-alignment-bound; the body of alignCostA with the canonical labels mkALabel/mkBLabel).