A Formal Core of Causal Nonseparability: The Order Record, Dephasing, and the Witness Classification
Rolando P. Hong Enriquez
Two structural facts anchor the theory of indefinite causal order, and this paper proves both as explicit theorems and checks them line by line in a machine proof assistant. The first is a record fact: a process without a definite order must keep, on its control memory, a record of the orders it superposes, and erasing that record by blurring the control returns a plain mixture of definite orders. The second is a classification fact: on the canonical slice of processes, a single straight-line certificate decides exactly which processes are ordered and which are not. Working over an abstract algebra of order operators, so that each result holds in any faithful concrete model, the paper proves a short, connected set of statements. Blurring the order record always lands the process in the definite-order set, which is the mathematical heart of record-erasure. A process sits in that definite-order set exactly when it is unchanged by blurring, so that any surviving control coherence is an irreducible record. The number of orders a process populates is at most the size of its control memory, which is the record form of the memory cost studied elsewhere in the program. On the canonical slice the certificate takes a specific value that varies in a straight-line way across the slice, so that indefiniteness holds precisely on one side of a line, with an explicit decomposition into definite orders on the other side. And a tempting finer diagnostic built from the process's intrinsic flow is shown to over-count, because it tracks a product of the two dials rather than the true boundary. Every theorem is stated over an abstract ring so that it transfers to any faithful model, and every statement in the paper has been checked in a machine proof assistant with no assumptions beyond the three standard foundations of mathematics; the correspondence to the formal development is tabulated in an appendix. The contribution is a small, dependable, reusable core that turns the program's paper-level results about indefinite causal order into a durable, independently checkable formal library, and by forcing complete rigour it guards against hidden gaps of the kind that informal arguments can conceal. The paper is careful about scope, noting that the definite-order set reached by full blurring is contained in, but does not exhaust, the whole ordered set, and that the finer flow-based diagnostic is only a necessary sign of indefiniteness, not a sufficient one. What it secures is a machine-checked foundation on which the geometric, thermodynamic, and resource-theoretic parts of the program can safely rest.
•Aug 24, 2026•0 citations