Section 10 commutativity points: answer-valued diagonal measurements #
This file proves the commutativity-at-points theorem using the answer-valued
diagonal-line verifier relation: the counterpart of
MIPRE/Background/LIDT/MIPStarRE/LDT/CommutativityPoints/AnswerTheorems.lean in the port of
planning/c6b-plan.md (milestone M1, section "Port conventions"). The conclusion concerns only
the point measurements, hence it can later be transferred to the ordinary carrier used by
self-improvement, but the proof does not use the carrier's inert diagonal measurement.
The strategy is an answer-valued symmetric strategy strategy : AnswerSymStrat params ๐ K: the
point answers are placed by strategy.state.L, the line answers by strategy.state.R, and the
joint operators are K โL[โ] K. The answer-valued route is ported as the vendored file has it,
parallel to the bridge theorems for SymStrat and not unified with them. The vendored file has
no classical content, so it is not imported; the classical names it uses come from the vendored
Defs, Approximation and SharedHelpers/SharedLine through the imports of
Co/CommutativityPoints/SharedHelpers/SharedLine.lean, named by an explicit
open MIPStarRE.LDT.CommutativityPoints (โฆ) list.
Not ported #
Every declaration of the vendored file has a counterpart here.
References #
In LionSR/MIPStarRE at commit 507e8122, not in this repository:
references/ldt-paper/commutativity-points.texblueprint/src/chapter/ch08_commutativity.tex
In this repository: lem:co-commutativity-points in blueprint/src/content/08_downstream.tex.
The point measurement, reindexed by a sampled diagonal line and a parameter on it, for an answer-valued strategy.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Evaluate an answer-valued diagonal-line measurement at the sampled parameter.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The ordered point product (A^u_a A^v_b) โ I for an answer-valued
strategy, placed by strategy.state.L.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The reversed point product (A^v_b A^u_a) โ I for an answer-valued
strategy, placed by strategy.state.L.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The mixed bridge A^u_a โ L^โ_[f(v)=b], placed by strategy.state.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The bridge I โ (L^โ_[f(v)=b] ยท L^โ_[f(u)=a]), placed by strategy.state.R.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The swapped bridge I โ (L^โ_[f(u)=a] ยท L^โ_[f(v)=b]), placed by strategy.state.R.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The mixed bridge A^v_b โ L^โ_[f(u)=a], placed by strategy.state, with outcome (a, b):
a indexes the line evaluation and b the point measurement.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Outcome (a, b) of the left mixed bridge: it is A^u_a โ L^โ_[f(v)=b], the point
measurement at u and the answer-valued diagonal-line evaluation at t_v, placed by
strategy.state.opTensor.
Outcome (a, b) of the right mixed bridge: it is A^v_b โ L^โ_[f(u)=a], the point
measurement at v and the answer-valued diagonal-line evaluation at t_u, placed by
strategy.state.opTensor, the post-processing having swapped the outcome pair back.
For a good answer-valued strategy, over the shared-diagonal-line distribution, the point
measurement at the second point v, placed left, and the answer-valued diagonal-line evaluation at
t_v, placed right, are pointDiagonalLineApproxError params gamma-close (SDDOpRel): the
first point is ignored, (โ, t_v) having the law of pointWithDiagonalLineDistribution.
For a good answer-valued strategy, over the shared-diagonal-line distribution, the point
measurement at the first point u, placed left, and the answer-valued diagonal-line evaluation at
t_u, placed right, are pointDiagonalLineApproxError params gamma-close (SDDOpRel): the
second point is ignored, (โ, t_u) having the law of pointWithDiagonalLineDistribution.
Lean-only: A local tensor-placement comparison in the answer-valued point-commutativity chain.
Paper origin: upstream references/ldt-paper/commutativity_points.tex; this is one of
the formal transport steps used to realize the mixed point/diagonal comparison
appearing in the paper. It is internal to the answer-valued implementation
tracked in upstream issue #1507 and is not a source theorem. Discharge: proved here from
the already formalized point-to-diagonal-line approximation and tensor-ordering
identities.
Lean-only: A local tensor-placement comparison from the mixed product to the ordered diagonal-line product.
Paper origin: upstream references/ldt-paper/commutativity_points.tex; this is an
internal reindexing and tensor-ordering step in the answer-valued
point-commutativity route tracked in upstream issue #1507. Discharge: proved here by
transporting the point-to-line comparison through the explicit ordered product
identities.
Lean-only: A local tensor-placement comparison from the ordered diagonal-line product to the reversed mixed product.
Paper origin: upstream references/ldt-paper/commutativity_points.tex; this is an
internal answer-valued implementation step for the point-commutativity argument
tracked in upstream issue #1507. Discharge: proved here from the reversed
point-to-line comparison and the explicit ordered/reversed product equality.
Lean-only: A local tensor-placement comparison from the reversed mixed product back to the reversed point product.
Paper origin: upstream references/ldt-paper/commutativity_points.tex; this is the last
internal answer-valued transport step in the point-commutativity chain tracked
in upstream issue #1507. Discharge: proved here from the line-to-point comparison and
the explicit tensor-placement identities.
Answer-valued form of thm:commutativity-points.
The proof is the paper's diagonal-line bridge argument, but the diagonal-line
measurement is the answer-valued measurement of strategy.