Assumptions, results and research status
Exact finite operator calculations with explicit reduction hypotheses
Assumptions
- An ambient represented unital star algebra and an orthogonal projection are given.
- Products use ordinary operator multiplication unless an alternative effective product is explicitly defined.
- Unbounded operators are handled on a declared common invariant domain.
Results
- The product defect is precisely the contribution that leaves and re-enters the retained space.
- Compressing a commutative three-state algebra can produce a noncommuting operator system that is not closed under products.
- Reducing representations, corners and pinching have different algebraic and physical roles.
Open questions
- Which observables and their products are predictively retained by the completion rule?
- What physical process implements a conditional expectation or an effective product?
- How do algebra, dynamics and spectral domains remain compatible through continuum reduction?
Calculate the missing intermediate sector
Let u:K→H be an isometry, P=uu† and Q=I−P. A bounded ambient observable a compresses to Φ(a)=u†au on K. This map preserves the identity and adjoints. Its failure to preserve multiplication is exactly
Insert I=P+Q between a and b to obtain the identity. The right-hand term describes a route that enters the eliminated space between the two operations. Multiplying already compressed operators removes that route. Thus the two possible meanings of “perform a and then b” need not agree after reduction.
For a self-adjoint a, the defect is positive: Φ(a²)−Φ(a)²=(Qau)†(Qau). It measures the squared amplitude of an excursion out of the retained space. This also explains why compression is compatible with positivity while failing to be a homomorphism. The distinction is essential when turning a family of compressed measurements into a proposed observable algebra.
Source: Noncommutative Geometry, PDF pp. 16–17, 30. The product-defect calculation is an explicit audit of the compressed-algebra interface.
A two-state example separates two operational questions
Take H=C², retain only the first basis vector and choose a=σx=[0,1;1,0]. Its compression is the scalar zero. Yet a²=I, whose compression is the identity on the retained line.
Two ambient swaps return a retained state exactly. Inserting the projection between them deletes the intermediate second state. An observed record that needs to predict the unmeasured two-swap experiment cannot treat the first compression as a complete replacement for a.
This example does not mean that a one-dimensional effective algebra is inconsistent. It means that the intended test family and operation protocol must be stated. If every step includes an actual retained-outcome measurement, the compressed composition is the right conditional amplitude. If hidden propagation occurs between observations, memory or an effective interaction can be required. Predictive sufficiency determines which of these histories the record must reproduce.
Source: Completed Records, PDF pp. 3–5, 25–26. The two-swap matrix is a worked example developed here.
Compress a commutative algebra into three effects
Let A be the diagonal algebra on C³. Retain the plane orthogonal to (1,1,1), with frame columns v₁=(1,−1,0)/√2 and v₂=(1,1,−2)/√6. Compressing the three diagonal coordinate projections gives the following two-by-two matrices, with a=1/√12:
E₃=[0,0;0,2/3].
Each Eᵢ is positive, E₁+E₂+E₃=I₂, and Eᵢ²=(2/3)Eᵢ. They are measurement effects rather than orthogonal projections. The image Φ(A) is the complex linear span of I₂, σx and σz, of dimension three.
The original algebra has commuting sharp outcomes. After selecting a nonreducing plane, those outcomes are represented by overlapping effects. This is a complete finite example of an operator system: a linear, adjoint-closed space containing the identity. No multiplication closure has yet been established, and positivity alone does not provide it.
Source: Noncommutative Geometry, PDF pp. 16–17, 30. This three-state compression is an original worked counterexample.
Exhibit the product outside the image
Multiplying the first two effects gives E₁E₂=[1/6,−1/(6√3);1/(6√3),−1/18]. Every matrix in Φ(A) is symmetric under ordinary transpose, including complex linear combinations. This product is not symmetric, so it is not in Φ(A).
The generated star algebra is all of M₂(C), because adding products supplies the missing Pauli direction. There are therefore two mathematically legitimate objects: the three-dimensional compressed operator system and the four-dimensional algebra generated by it. They should not receive the same name without explaining the enlargement.
Noncommuting compressed effects do not prove that the original diagonal observables failed to commute. Their new sequential products implement operations with an intermediate retained-space restriction. A physical claim about noncommutative gauge observables must identify that operation, its reference resources and the tests it preserves; a change of representation alone does not settle those issues.
Source: Noncommutative Geometry, PDF pp. 16–17, 30. The displayed product and generated algebra are calculated here.
Three conditions give different valid reductions
If P commutes with every element of A, then Qbu=0 and the product defect vanishes. Compression is a star representation on the reducing subspace. This is a particularly strong condition: both the retained and eliminated spaces are invariant under the full observable algebra.
If instead P belongs to A, the corner PAP is an algebra with identity P, since (PaP)(PbP)=P(aPb)P and aPb remains in A. The corner can be valid even when the map a↦PaP on the original algebra is not multiplicative. The full-matrix two-swap example illustrates that distinction.
A third option is to generate the smallest star algebra containing all compressed observables. That changes the allowed observable set and needs an explicit operational interpretation. Alternatively one may retain the operator system and work with its positive measurement effects. None of these choices follows just from naming a spectral cutoff, and each preserves a different amount of the original product structure.
Source: Noncommutative Geometry, PDF pp. 16–17, 30. These alternatives specify the missing algebraic hypotheses.
Pinching is a conditional expectation on the full space
For the full matrix algebra B(H), the block pinching E(a)=PaP+QaQ is unital, completely positive and idempotent. Its image is the block-diagonal algebra B(PH)⊕B(QH). If b and c are block diagonal, direct multiplication gives E(bac)=bE(a)c.
This is a conditional expectation onto an actual subalgebra. It removes cross-block coherences while retaining both diagonal sectors. Compression to PH instead changes the Hilbert space and discards the Q block. The operations have different information content, even though both contain a term PaP.
The operational manuscript uses explicit conditional expectations and channels to specify what is forgotten. Such maps need not be reversible unitary conjugations on the same algebra. An exact selector expectation can erase a nonzero observable, while every reversible conjugation preserves its norm. A declared physical channel, with any required environment, is therefore additional content beyond a passive change of frame.
Source: Completed Records, PDF pp. 7–8.
Check the algebra before checking the Dirac operator
A spectral-triple construction needs a represented algebra, a self-adjoint operator with the required domain and compactness properties, and bounded commutators on a suitable dense algebra. A list of matrices labelled PaP is insufficient when the list fails multiplication closure.
Suppose P is a spectral projection of D and reduces it. On a compatible common domain, DP=D|PH is self-adjoint and
That commutator identity is useful, but it does not make a↦PaP multiplicative. If one takes the generated algebra, commutators must be checked on a specified smooth subalgebra; norm closure alone does not preserve every unbounded-commutator domain.
The source’s compactified temporal model supplies a particular algebra and operator before proving its spectral properties. A later low-energy construction must repeat the relevant compatibility checks. Finite rank makes many operator conditions easy, but it also changes heat asymptotics and cannot preserve an infinite spectral dimension automatically.
Closed observables need not have closed dynamics
Even a valid retained algebra does not imply that the physical propagator closes on its Hilbert space. For a time-independent self-adjoint H and U(t)=e−itH/ℏ, define T(t)=u†U(t)u. Its failure of unitarity obeys
Leakage is nonnegative. Exact unitary group evolution on the retained space requires that the projection reduce the dynamics; otherwise elimination can create memory and an energy-dependent effective operator. Those effects do not equal spatial curvature and should not be inferred from it.
A reviewable reduced model therefore identifies four items: the retained space, the observable products, the operation or measurement protocol, and the effective time evolution. The finite examples above let each claim fail independently. Resolving them is a concrete step toward a completion-based theory, while the remaining choice of physical algebra and dynamics stays explicit.
Source: Completed Records, PDF pp. 25–26.
Source: Vacuum Obstructions, PDF pp. 23–25.
Continue reading
Sources for this guide
Source manuscripts, version notes, and standard references are listed in the reference library.