Documentation

CsdLean4.RecordLayer.OnticMarginals

SigmaLayer/OnticMarginals: A6 steps 2–3 — ontic reduction maps, and marginal stability #

Category: 7-SigmaLayer (Paper C A6 — composite systems).

What this adds #

Step 1 (OnticComposite.lean) made non-factorisation a theorem. This file supplies the other half of A6's mathematical content:

⚠️ Scope #

References #

SigmaLayer/OnticComposite.lean (step 1); Mathlib/LinearAlgebra/Matrix/PartialTrace.lean (traceRight, traceLeft); LF4/ProjectedDynamics.lean (schrodingerUnitary); LF3/Projectors/TensorModel.lean, SigmaLayer/TensorSector.lean (the operational no-signalling this is the ontic form of); specs/reconstruction-status.md §2 (the A6 row).

Step 2: the ray density and its marginals #

noncomputable def CSD.RecordLayer.rayDensity {nA nB : } (p : Projectivization (EuclideanSpace (Fin nA × Fin nB))) :
Matrix (Fin nA × Fin nB) (Fin nA × Fin nB)

The density matrix of a composite ray: ρ(x,y) = v(x)·conj(v(y))/‖v‖² at the canonical representative. Ray-well-defined (rayDensity_mk), unit trace (rayDensity_trace).

Equations
Instances For
    theorem CSD.RecordLayer.rayDensity_ratio_smul {nA nB : } (c : ) (hc : c 0) (v : EuclideanSpace (Fin nA × Fin nB)) (x y : Fin nA × Fin nB) :
    (c v).ofLp x * star ((c v).ofLp y) / ↑(c v ^ 2) = v.ofLp x * star (v.ofLp y) / ↑(v ^ 2)

    The scale-invariance computation behind ray-well-definedness.

    theorem CSD.RecordLayer.rayDensity_mk {nA nB : } (v : EuclideanSpace (Fin nA × Fin nB)) (hv : v 0) :
    rayDensity (Projectivization.mk v hv) = Matrix.of fun (x y : Fin nA × Fin nB) => v.ofLp x * star (v.ofLp y) / ↑(v ^ 2)

    The density is well-defined on rays.

    theorem CSD.RecordLayer.normSq_eq_sum_mul_star {ι : Type u_1} [Fintype ι] (w : EuclideanSpace ι) :
    ↑(w ^ 2) = x : ι, w.ofLp x * star (w.ofLp x)

    ‖w‖² as the diagonal sum ∑ₓ w(x)·conj(w(x)) — the casting workhorse for the marginal computations.

    The ray density has unit trace — it is a genuine state.

    noncomputable def CSD.RecordLayer.reduceA {nA nB : } (p : Projectivization (EuclideanSpace (Fin nA × Fin nB))) :
    Matrix (Fin nA) (Fin nA)

    The A-marginal of a composite ray: trace out B.

    Equations
    Instances For
      noncomputable def CSD.RecordLayer.reduceB {nA nB : } (p : Projectivization (EuclideanSpace (Fin nA × Fin nB))) :
      Matrix (Fin nB) (Fin nB)

      The B-marginal of a composite ray: trace out A.

      Equations
      Instances For

        Step 3: local unitaries, and marginal stability #

        noncomputable def CSD.RecordLayer.actA {nA nB : } (U : Matrix (Fin nA) (Fin nA) ) (v : EuclideanSpace (Fin nA × Fin nB)) :

        The local A-action on composite vectors: (actA U v)(j,k) = ∑ₐ U j a · v(a,k) — the vector form of U ⊗ 1, with no Kronecker plumbing.

        Equations
        Instances For
          @[simp]
          theorem CSD.RecordLayer.actA_apply {nA nB : } (U : Matrix (Fin nA) (Fin nA) ) (v : EuclideanSpace (Fin nA × Fin nB)) (j : Fin nA) (k : Fin nB) :
          (actA U v).ofLp (j, k) = a : Fin nA, U j a * v.ofLp (a, k)
          theorem CSD.RecordLayer.actA_actA {nA nB : } (U W : Matrix (Fin nA) (Fin nA) ) (v : EuclideanSpace (Fin nA × Fin nB)) :
          actA U (actA W v) = actA (U * W) v
          theorem CSD.RecordLayer.actA_one {nA nB : } (v : EuclideanSpace (Fin nA × Fin nB)) :
          actA 1 v = v
          theorem CSD.RecordLayer.actA_zero {nA nB : } (U : Matrix (Fin nA) (Fin nA) ) :
          actA U 0 = 0
          theorem CSD.RecordLayer.actA_ne_zero {nA nB : } {U : Matrix (Fin nA) (Fin nA) } (hU : U Matrix.unitaryGroup (Fin nA) ) {v : EuclideanSpace (Fin nA × Fin nB)} (hv : v 0) :
          actA U v 0
          theorem CSD.RecordLayer.actA_column_sums {nA nB : } {U : Matrix (Fin nA) (Fin nA) } (hU : U Matrix.unitaryGroup (Fin nA) ) (v : EuclideanSpace (Fin nA × Fin nB)) (k k' : Fin nB) :
          j : Fin nA, (actA U v).ofLp (j, k) * star ((actA U v).ofLp (j, k')) = a : Fin nA, v.ofLp (a, k) * star (v.ofLp (a, k'))

          ★ The workhorse: local sums against a unitary collapse. For UᴴU = 1,

          ∑ⱼ (actA U v)(j,k) · conj((actA U v)(j,k')) = ∑ₐ v(a,k) · conj(v(a,k'))

          for every pair of B-indices — the invariance computation behind both the norm and the B-marginal.

          theorem CSD.RecordLayer.actA_row_sums {nA nB : } (U : Matrix (Fin nA) (Fin nA) ) (v : EuclideanSpace (Fin nA × Fin nB)) (j j' : Fin nA) :
          k : Fin nB, (actA U v).ofLp (j, k) * star ((actA U v).ofLp (j', k)) = a : Fin nA, b : Fin nA, U j a * star (U j' b) * k : Fin nB, v.ofLp (a, k) * star (v.ofLp (b, k))

          The bilinear expansion of a row sum — no unitarity needed.

          theorem CSD.RecordLayer.norm_actA {nA nB : } {U : Matrix (Fin nA) (Fin nA) } (hU : U Matrix.unitaryGroup (Fin nA) ) (v : EuclideanSpace (Fin nA × Fin nB)) :

          Local unitaries preserve the composite norm.

          noncomputable def CSD.RecordLayer.pointA {nA nB : } {U : Matrix (Fin nA) (Fin nA) } (hU : U Matrix.unitaryGroup (Fin nA) ) (p : Projectivization (EuclideanSpace (Fin nA × Fin nB))) :

          The local action on composite rays.

          Equations
          Instances For

            ★★ Marginal stability = ontic no-signalling. A local unitary on A leaves the B-marginal of the composite ray unchanged: acting on A changes nothing B can see, at the level of the single ontic point. This is Paper C A6's marginal-stability clause, and the ontic form of the corpus's operational tensorSector_no_signalling.

            The A-marginal evolves by conjugation — the Heisenberg transformation law for the reduction map.

            theorem CSD.RecordLayer.reduceB_local_flow_invariant {nA nB : } [NeZero nA] {HA : Matrix (Fin nA) (Fin nA) } (hHA : HA.IsHermitian) (t : ) (p : Projectivization (EuclideanSpace (Fin nA × Fin nB))) :

            ★★ Marginal stability under local flows. The Schrödinger flow of any A-Hamiltonian leaves the B-marginal fixed at every time — A6's marginal-stability clause in flow form, the ontic no-signalling statement.