An arXiv preprint lays out an exact way to represent quaternionic quantum states and channels in doubled complex spaces, and an exact criterion for quaternionic error correction. It also compares octonionic models without selecting one global continuation. On the quaternionic side, the representation keeps an antiunitary symplectic structure, a fixed operator algebra and a normalization that divides the embedded state by two; on the octonionic side, the alternatives preserve different operational structures.
The front matter labels the document arXiv:2608.20259v1 [math.QA] 20 Aug 2026, and the paper states that no experimental data are used.
A precise route through complex space
On the quaternionic side, the authors use a doubled complex Hilbert space as an exact representation. The embedding is constrained: it retains an antiunitary symplectic structure—the rule that selects the quaternionic sector—along with a fixed operator algebra and the required state normalization.
One clear algebraic identity is that the ordinary complex rank of an embedded quaternionic state is twice its quaternionic rank. This is a formal relation, not an empirical estimate.
The same representation carries a quaternionic Kraus channel into complex space while preserving the quaternionic state sector. That gives the construction an exact complex-space simulation of the channel action, subject to the fixed-sector assumptions used in the paper.
To test whether a complex channel is liftable to the quaternionic framework, the paper uses one linear fixed-point condition on its Choi matrix. When the condition holds, a compatible quaternionic family of Kraus operators can be reconstructed from a fixed eigenbasis.
An exact code test, with a boundary
For a finite-dimensional right-quaternionic code, the paper gives an exact error-correction criterion: the compression coefficients for the relevant errors must lie in the real center of the quaternionic operator algebra. In ordinary terms, the compressed error information has to take a restricted scalar form for the criterion to hold.
The recovery construction also covers singular error families. It omits error combinations associated with zero eigenvalues and uses only the remaining nonzero syndrome sectors.
The one-dimensional code is treated as a separate boundary. Its normalized logical state space is a single state, so discard-and-prepare correction carries no logical information.
The missing piece is composition
Those results do not by themselves specify how two right-quaternionic modules form a composite system. The paper says a compatible left action by quaternionic scalars, an inner product and a discard map—the rule for removing one part—are also required; two right-module structures alone do not determine the composite.
That choice also sets the terms for erasure. Once a composite is selected, exact erasure is governed by the ordinary complex Knill–Laflamme compression condition for block errors. For approximate correction, the relevant control is the distance of the entire complementary channel from replacement channels.
Where octonions stop behaving like operators
Octonions introduce a different boundary. Their multiplication is nonassociative, meaning that the grouping of three factors can matter; the paper shows that associative operator composition cannot globally coincide with octonion multiplication on every triple.
Rather than treating the alternatives as interchangeable, the paper compares what each one keeps. Para-linear models retain octonionic amplitudes but lack an intrinsic complete package for channels and quantum error correction; cochain twisting untwists to ordinary complex graded theory; and Jordan, Clifford-envelope and Moufang models preserve different state, operator or multiplication structures.
There is a narrower route inside the octonions: a fixed associative quaternionic sector inherits the quaternionic operational theory when amplitudes, operators and errors remain in that sector. The result does not cover processes that move among incompatible sectors.
A boundary map, not a physical test
The preprint gives exact quaternionic constructions inside a constrained complex representation, identifies the extra data needed for composites and discard, and leaves octonionic models dependent on an explicit choice of operational framework. No experimental system is tested.
Paper data and sources
Original title: Operational Foundations for Quaternionic and Octonionic Quantum Models: Exact Quaternionic Channels and Error Correction, Para-Linear Operators, and Categorical Closure Boundaries Beyond Associativity
Authors: Santiago Pineda Montoya, Johan H. Rua Munoz
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text