An arXiv preprint reports an exact mathematical link between a projection’s positivity and the structure of the subspace it reaches. For the positive contractive projections covered by the paper, complete positivity is equivalent to the range being completely isometric to a noncommutative Lp-space. A projection’s range is the subspace selected by the map; the result says that this range carries enough structural information to characterize the map’s stronger positivity property, but only under the theorem’s assumptions.
This is not a study with participants, measurements or a statistical effect. It is a theoretical operator-algebra analysis of noncommutative Lp-spaces, von Neumann algebras, positive contractive projections, closed subspaces, complete isometries and W*-TROs, also called ternary rings of operators. In this context, “complete isometry” is the paper’s required form of structural matching: the paper contrasts it with ordinary isometry and shows that the weaker condition is not enough.
A conditional equivalence
The first characterization is narrow in scope. It concerns a positive contractive projection on a sigma-finite noncommutative Lp-space and applies under the paper’s stated restriction on p. Within that frame, the two sides of the theorem are complete positivity of the projection and a range that is completely isometric to a noncommutative Lp-space. The claim is an if-and-only-if statement, so the structural range condition is not merely a consequence; it is equivalent to complete positivity.
That equivalence does not license a broader reading. The paper gives a counterexample showing that positivity of the projection is an essential hypothesis. It therefore does not say that every contractive projection is completely positive. Nor does it allow ordinary isometry to stand in for complete isometry, even when the ordinary isometry preserves order.
Five formal ways to recognize the same case
A corollary makes the result more detailed without changing its assumptions. Under the same conditions, it states a five-way equivalence linking complete positivity, 2-positivity, a completely isometric noncommutative Lp-range, a surjective 2-isometry, and complete order/isometric isomorphism. These are formal descriptions within the theory, not five empirical tests or five separately observed effects.
For a general reader, the plain meaning is that the paper supplies several interchangeable mathematical labels for the same situation. The central label is still structural: whether the projection’s range has the complete-isometric form required by the theorem. The corollary does not remove the need for positivity, the stated p restriction or the other hypotheses that define the setting.
A rectangular version
The paper also gives a characterization for contractively decomposable complementability. For a nonzero closed subspace under the stated assumptions, being the range of such a projection is equivalent to having a surjective complete isometry from a rectangular corner—and equivalently from a rectangular Lp-space associated with a W*-TRO. In other words, the theorem identifies the rectangular structure of the subspaces that can appear in this way.
The second characterization also has explicit conditions. It assumes sigma-finiteness and a separable predual for the ambient algebra, in addition to the conditions stated for the class of spaces and maps under study. The supplied analysis excludes p=2 from the results. The paper therefore describes a defined mathematical class; it does not claim a characterization for every noncommutative Lp-space.
How the proofs move
The first proof leans on the Junge–Ruan–Sherman classification. It factors the relevant map and absorbs a partial-isometry factor into the von Neumann-algebraic part of that factorization. That step connects complete positivity with the completely isometric form of the range through the structure of the factorization.
An intermediate structural result says that a completely positive complete isometry has a range that is a bimodule over the von Neumann algebra image π(N). Put less formally, the range has the two-sided module relationship with that image that the proof uses. This is one of the links between the abstract map and the subspace it produces.
For the rectangular result, the proof uses S_p amplification, then converts the problem into an ordinary noncommutative Lp-space, applies the classification theorem and compresses the answer back to a matrix corner. The sequence is a proof strategy for transporting the rectangular question into a setting covered by the classification; it is not a numerical approximation or a data-analysis step.
What the paper does—and does not—establish
Taken together, the theorems are characterizations within their hypotheses, not universal statements about projections. The positivity counterexample marks one boundary; the isometry counterexample marks another. Complete positivity is not shown for arbitrary contractive projections, and ordinary order-preserving isometry is not shown to be sufficient.
The document is an arXiv version-one preprint dated 20 Aug 2026. Its supplied end matter reports no funding statement, says the author has no competing interests and says no datasets were generated. Those disclosures fit the study’s design: the paper examines theoretical mathematical objects and contains no empirical participant sample.
The open questions are correspondingly mathematical. The supplied analysis asks whether similar statements can be extended beyond the sigma-finiteness and separable-predual assumptions, whether a replacement characterization is possible in the excluded p=2 case, and whether weaker structural assumptions could still support narrower conclusions. For now, the paper’s contribution is a precise map of what the stated complete-isometric conditions do characterize—and where they stop.
Paper data and sources
Original title: Completely isometric subspaces of noncommutative $\mathrm{L}^p$-spaces and contractive projections
Authors: Cédric Arhancet
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text