We define the finite-dimensional one-slot deterministic superchannelcategory by its normalized Choi operators, and an environment-decorationfunctor on it by identity-wire insertion. A right adjoint to thisdecoration would internalize the environment universally, as a curryingoperation on processes. For unrestricted completely positive maps theanalogous construction exists through the compact-closed internal hom;we prove that probability normalization destroys it: the decorationfunctor has neither a right nor a left adjoint when the environment isnontrivial. The hom-sets of the category have explicit affine-dimensionpolynomials in the squared interface dimensions, and probabilitynormalization ties the constant term of each polynomial rigidly to onedistinguished coefficient. We call this the normalization-defect(intercept) principle. Environment decoration rescales precisely thedistinguished coefficient while leaving the constant term fixed, so norepresenting object satisfying its own normalization constraint canreproduce the decorated dimensions uniformly over all sources; hencethere is no right adjoint. A dual dimension count at the trivial objectforces a squared dimension strictly between consecutive squares,excluding every left adjoint. It follows that the canonical paralleltensor product is not closed at nontrivial environment objects. Anappendix constructs a category of deterministic combs of arbitraryfinite order and extends the same coefficient obstruction to show thatenvironment decoration has neither adjoint at any order. The obstructionis uniform over all sources and is consistent with the recoverability ofindividual processes.
We define the finite-dimensional one-slot deterministic superchannelcategory by its normalized Choi operators, and an environment-decorationfunctor on it by identity-wire insertion. A right adjoint to thisdecoration would internalize the environment universally, as a curryingoperation on processes. For unrestricted completely positive maps theanalogous construction exists through the compact-closed internal hom;we prove that probability normalization destroys it: the decorationfunctor has neither a right nor a left adjoint when the environment isnontrivial. The hom-sets of the category have explicit affine-dimensionpolynomials in the squared interface dimensions, and probabilitynormalization ties the constant term of each polynomial rigidly to onedistinguished coefficient. We call this the normalization-defect(intercept) principle. Environment decoration rescales precisely thedistinguished coefficient while leaving the constant term fixed, so norepresenting object satisfying its own normalization constraint canreproduce the decorated dimensions uniformly over all sources; hencethere is no right adjoint. A dual dimension count at the trivial objectforces a squared dimension strictly between consecutive squares,excluding every left adjoint. It follows that the canonical paralleltensor product is not closed at nontrivial environment objects. Anappendix constructs a category of deterministic combs of arbitraryfinite order and extends the same coefficient obstruction to show thatenvironment decoration has neither adjoint at any order. The obstructionis uniform over all sources and is consistent with the recoverability ofindividual processes.
We ask whether coupling a fixed finite-dimensional environment to a quantum process can be undone functorially: does tensoring with an environment E admit an adjoint in the category of quantum channels, and in related categories of instruments and stochastic maps? We prove that it does not, unless the environment is trivial. Our central new tool is an intercept principle: an enriched adjunction must preserve the affine dimension of hom-sets coefficient by coefficient, and probability normalization contributes a dimension-counting intercept that no environment-independent representing object can absorb. Specialized to the environment endofunctor, this gives a single coefficient-rigidity obstruction that governs channels, ordered instruments, finite stochastic maps, and higher-order superchannels uniformly, and yields an exact classification in each case: a left or right adjoint exists precisely when the environment is isomorphic to the tensor unit. We introduce an operational counterpart of this categorical obstruction. No finite-dimensional quantum memory with a fixed processor can store and exactly evaluate all channels on a nontrivial system, and any natural finite-memory encoding of processes would induce the adjoint just shown to be impossible. For approximate compression into insufficient memory we prove a new quantitative floor on the achievable diamond-norm error, sharp for the argument used and extended from linear to arbitrary continuous encoders by a Borsuk–Ulam argument. Finally, we place these obstructions against the standard theory of exact recovery: a channel is Petz-recoverable exactly when its complementary channel is constant, a morphism-level splitting that we show is strictly weaker than the functorial currying structure ruled out above. We delimit what these results do and do not establish, and accompany the mathematical claims with a machine-checkable verification suite.
Can an external environment be absorbed universally into a quantum process? This article identifies trace normalization as the obstruction and tracks it through higher orders. Exact ranks are derived for sequential-comb normalization at every order. Two concrete families of higher-order maps are defined, and their transformation-space dimensions are computed. In the family stable under arbitrary auxiliary combs, adding an environment to an input has neither a left nor a right adjoint from order three onward. Together with the channel and superchannel cases, this yields an all-order family of no-go theorems. The claim is restricted to specified process types: broader normalized type systems may still admit internal homs. Without normalization, a two-sided adjunction returns for normal completely positive maps. Exactly the finite-dimensional Hilbert spaces are dualizable, but trace scaling and rank-one evaluation prevent restriction to deterministic channels. The continuous scalar laws underlying composition- and tensor-additive dynamics are proved to be logarithmic boundary terms. Nonadditive quadratic actions are analyzed through exact stationarity equations and several exact qubit phase classifications. Finally, continuous channel encoding below the full channel-space dimension has worst-case diamond-norm error at least 1/(dAdB ), while normalized weighted cups have sharp defect 1 − 1/δ2. The results identify three ways around the obstruction: enlarge the type universe, remove normalization, or accept controlled approximation
Can a process with a fixed environmental input be represented naturally as one into anotherphysical system? Categorically, does tensoring with the environment have a right adjoint,or causal internal hom? The dimension-profile method introduced here records the numberof independent parameters in each family of processes. Any adjunction preserves convexmixtures and therefore these counts. The intercept principle turns systematic violations intogeometric or arithmetic no-go tests.Applied to quantum and classical channels, instruments, comb states, and superchan-nels, this yields complete or conditional classifications. Quantum channels fail an all-inputarithmetic test; even exceptional dimension matches have incompatible boundary geometry.Classical channel dimensions match, but products of simplices are not simplices. Nontrivialinstruments fail grade by grade. Adding a fixed channel type admits a right adjoint onlywhen the added type is trivial. Comb-state dimensions are derived separately from arbitraryhigher-order process spaces.Dimension is only the first layer. The analysis separates affine shape, positivity withancillas, and naturality; constructs the exact linear candidate for a causal internal hom; andproves that universal evaluators assemble into an adjunction. Finite tomography, symmetry,and robustness make the converse testable. Extensions treat generalized probabilistic the-ories, composites without local tomography, positive extension, and finite operator-systemrealization
A recurring question in quantum information theory is whether the coupling of a system toan environment can be reversed in a structural, universal way: given a channel that processesa system together with an environment, can the effect of the environment be factored out andinverted? Working in the category Chan of finite-dimensional Hilbert spaces and completelypositive trace-preserving maps, and under an affine-enrichment hypothesis, we study thegeometry of the obstruction to such a reversal.The obstruction carries a precise numerical signature. Universal reversal would force thedimension of a representing object to equal d4E − d2E + 1, an integer that is never a perfectsquare for any environment dimension dE = dimE > 1; the gap to the nearest square isexactly d2E −1, the number of independent parameters of the environment’s state space S(E).At the geometric level, the Choi spectrahedron — the convex body of Choi operatorsof channels — contains proper faces whose dimensions cannot occur in any quantum statespace, and is therefore not affinely equivalent to one; this facial argument is decisive evenwhere a global dimension count is silent.Operationally, a channel discards its environment exactly when its Choi state has vanishing quantum mutual information across the environment, and its diamond-norm distanceto the set of environment-independent channels is bounded below by this mutual information. This characterization connects to the structure of quantum Markov chains [17] andto information-flow analyses of open-system dynamics; the quantitative diamond-norm formgiven here is the present contribution.Finally, we show that every regular, additive, refinement-invariant measure of environment size is logarithmic, equal to the entropy of the maximally mixed state; the underlyingcompletely additive arithmetic structure is classical (see e.g. [18]).The results are finite-dimensional and kinematic; they map the structure of the normalization obstruction and motivate a companion paper on whether this geometry can supporta dynamics.
Storing an arbitrary quantum channel in a finite-dimensional memory for later recoveryis impossible—a fact known as the normalization obstruction and closely related to the noprogramming theorem. This paper asks whether the framework can still support a meaningfuldynamics or variational principle.We show that the answer is essentially no for additive dynamics. In the category ofenvironment-decorated quantum channels, every composition-additive valuation is forced tobe a boundary term; under natural regularity assumptions, the only nontrivial edge actionis proportional to the logarithm of the environment dimension. The resulting variationalproblem is trivial, ruling out a rich additive dynamics.On the positive side, we classify the natural quadratic invariants of quantum channelsunder local unitary symmetry and prove there are exactly four independent ones. We alsoestablish a universal lower bound on approximate process storage: by a Borsuk–Ulam bottleneck argument, any continuous compression of the full channel space into a memory thatis too small must incur a worst-case diamond-norm error of at least 2/d2, where d is thesystem dimension.These results show that the framework is kinematically rigid but operationally informative: although no nontrivial additive dynamics exists, the obstruction itself yields sharpstructural and operational constraints.
Standard finite quantum de Finetti theorems approximate the k-system marginals of permutation-invariant states of n-systems by mixtures of independent and identically distributed (iid) states, usually in trace distance. We prove both standard and Renner's exponential de Finetti theorems in the stronger form of operator inequalities, implying bounds in every Schatten norm and for every quantum Rényi divergence satisfying data processing. In the standard case, for fixed local dimension, our k/n error bound in max-relative entropy improves on the previously best known k2/n scaling, even in the classical setting. The operator-inequality approach is particularly suited to study channel de Finetti representations because operator order between Choi states is equivalent to completely positive (CP) order between the underlying channels. For permutation-covariant channels N(n):A⊗n→B⊗n, where dA=dimA, we prove that the k-system reduced channel is CP-dominated by a mixture of tensor-power channels with error O(k/n) and polynomial dependence on the local dimensions, addressing a question raised by Berta et al. [Math. Program. 194, 781-829 (2022)]. Under the no-signalling condition, we also prove an exponential channel de Finetti theorem where the approximating mixture consists of Choi-almost-iid channels, whose normalized Choi states are almost-iid in the sense of Mazzola-Sutter-Renner. In the case of r defects, the representation error is at most poly(n)(2dA4k3/(nr2))(r+1)/2 and decays exponentially in n for a suitable choice of parameters r and k.