The density operator of an ontic preparation, and the QIT layer applied to it #
Category: 3-Local (the spine from a preparation on Σ to the quantum-information layer; W2 of
specs/qit-chain-scoping.md).
Two theorems of LF2 compose to "a preparation on Σ determines a density operator":
OperationalPackage.fromPreparation (LF2/Preparation.lean) turns a preparation measure μprep on
Σ into an operational package, and effect_gleason_representation (LF2/EffectGleason.lean) turns
any operational package into the unique density operator whose trace form is its effect
probabilities. Until 2026-09-11 the composition was never stated, its density operator was consumed
by no theorem of Mathlib/QuantumInfo/, and every entropy, distance and data-processing theorem
there was a statement about a bare matrix nothing CSD-side ever supplied. This module composes them
and applies the layer.
preparationDensity— the density operator of a preparation,(fromPreparation …).qdensity;- ★★
preparation_traceForm/preparation_qdensity_unique— the composition: the effect probabilities of a preparation are the trace form ofpreparationDensity, and it is the unique density operator with that property (effect_gleason_representationonfromPreparation); preparationDensity_posSemidef,preparationDensity_trace_one,preparationDensity_isHermitian— the instantiation lemmas, in the exact hypothesis forms the QIT layer takes;- ★
vonNeumannEntropy_preparation_nonneg, ★vonNeumannEntropy_preparation_le_log— the von Neumann entropy of a preparation onΣis well-defined, non-negative, and at mostlog N; - ★
channel_traceDist_preparation_le— the data-processing inequality for two preparations: no channel increases their trace distance.
Every theorem below is effect_gleason_representation plus one exact; that is the point. The
spine was two proved theorems apart, and the QIT layer's hypotheses are DensityOperator's fields.
Honest scope #
⚠️ What "preparation" means here. A SectorData on an abstract Σ, a MeasureBridgeData, a
probability measure μprep on Σ, and a unit-norm measurable representative rep : P → ℂᴺ — the
data fromPreparation takes. The sector is posited (specs/POSITS.md Posit 2); this module does not
derive it. What it proves is: given the posited sector, the QIT quantities of its preparations
obey the QIT theorems.
Gleason witness here, barycentre one module up. preparationDensity is characterised below by
its trace form (uniquely); that it is the barycentre ∫ |ψ⟩⟨ψ| d(π_* μprep) — entrywise, and in the
ρ_ep dμ_FS form under absolute continuity — is LF2/PreparationBarycenter.lean (W3,
preparationDensity_eq_barycenter, preparationDensity_apply). Nothing below depends on it.
⚠️ Channels are still bare. channel_traceDist_preparation_le takes any QuantumInfo.Channel;
that the channel comes from a Σ-flow is W5/W6.
References: specs/qit-chain-scoping.md (W2–W4); LF2/Preparation.lean, LF2/EffectGleason.lean;
Mathlib/QuantumInfo/{Entropy,DataProcessing,TraceDistance}.lean.
The composition #
The density operator of a preparation on Σ: the Gleason witness of the operational
package fromPreparation builds from it.
Equations
- CSD.LF2.preparationDensity D μFS bridge μprep rep hrep_unit hrep_meas = (CSD.LF2.OperationalPackage.fromPreparation D μFS bridge μprep rep hrep_unit hrep_meas).qdensity
Instances For
★★ The effect probabilities of a preparation are the trace form of its density operator.
fromPreparation composed with p_eq_traceForm_qdensity.
★★ Uniqueness: preparationDensity is the only density operator whose trace form is the
preparation's effect probabilities — effect_gleason_representation on fromPreparation.
The instantiation lemmas, in the QIT layer's hypothesis forms #
The QIT layer applied to a preparation #
The von Neumann entropy of a preparation on Σ.
Equations
- CSD.LF2.preparationEntropy D μFS bridge μprep rep hrep_unit hrep_meas = QuantumInfo.vonNeumannEntropy ⋯
Instances For
★ The entropy of a preparation is non-negative (vonNeumannEntropy_nonneg).
★ The entropy of a preparation on the N-dimensional sector is at most log N
(vonNeumannEntropy_le_log_card).
★ Data processing for two preparations on Σ: for any channel Φ, the trace distance of
the channel outputs is at most the trace distance of the preparations' density operators
(channel_traceDist_le).