No extension of the Quantum Tensor Product admits a Superposition principle

Abstract: Textbook quantum superposition refers to the feature that certain linear combinations of Hilbert space rays, each representing a valid quantum state, are themselves valid states. This notion is not operational, and it relies on the underlying Hilbert space formalism. Recent proposals for experimental tests of indefinite causal order, as well as tests probing the non-classicality of gravity, pivot on superposition, thereby calling for a theory-independent, operational formalisation of the concept. Here, we define superposition within the framework of Generalised Probabilistic Theories, based on observed statistics in prepare-and-measure experiments. Using this, we formulate three superposition principles to investigate which structural features of quantum theory carry over to other theories. We study conditions under which these principles carry over from subsystems to their compositions; to this end, we show that the quantum tensor product emerges as the largest composition rule for quantum systems respecting all three principles. Furthermore, we show how non-classical features such as entanglement and preparational uncertainty can be viewed as special forms of superposition.

• The paper defines superposition operationally in generalized probabilistic theories using maximally perfectly distinguishable sets and informationally complete measurements, recovering the Born-rule notion in finite-dimensional quantum theory.
• The paper proves that entanglement implies superposition under multiplicative operational dimension, while the converse fails, and shows that complete, uniform, and mutual superposition are logically independent across classical, polygonal, Boxworld, and toy theories.
• The paper establishes that no strict composition extending the quantum tensor product within the maximal no-signalling product can satisfy all three superposition principles under the no-restriction hypothesis, providing an operational constraint on quantum composition.

Operationalising superposition in generalised probabilistic theories

Textbook quantum superposition is a statement about rays in a Hilbert space: an extremal state $|\psi\rangle$ is a complex linear combination of other extremal states. This definition is not operational—it presupposes the Hilbert space formalism and offers no direct prescription for what statistics in a prepare-and-measure experiment certify the presence of superposition. This gap matters practically: proposed tests of gravitationally induced entanglement and of indefinite causal order both hinge on superposition, yet neither specifies which observed degrees of freedom are "in superposition" without importing the formalism whose validity (particularly for gravity) is precisely what is under test.

Fiorentino and Sengupta address this by defining superposition within the framework of Generalised Probabilistic Theories (GPTs). The definition is relational: given a maximally perfectly distinguishable (MPD) set $\mathsf{D}$ and its maximally distinguishing, extremal (MDE) measurement $\mathsf{M}$, an extremal state $r$ is a superposition of a subset $\mathsf{D}' \subseteq \mathsf{D}$ if $r$ is probabilistic with respect to exactly those effects in $\mathsf{M}$ that are deterministic on $\mathsf{D}'$. The maximality requirement is essential: with a merely perfectly distinguishable set, degenerate measurements would mislabel states as superpositions of incomplete collections. Superposition is attributed only to extremal states, mirroring its usage in quantum theory; a weakened variant allowing mixed states is analysed separately.

The work assumes throughout that MDE measurements form an informationally complete set—a nontrivial restriction that excludes some simplicial theories from tomographic completeness. Under this assumption, textbook superposition in finite-dimensional quantum theory is recovered as a special case: the Born-rule coefficients satisfy $|c_i|^2 = e_i(r)$ for the MDE discriminating the basis states.

Three inequivalent superposition principles

In quantum theory, three structural statements about pure states are equivalent: (i) every pure state is a superposition of others (complete), (ii) every pure state nondeterministic for a measurement basis is a superposition of the basis states (uniform), and (iii) superposition is symmetric—each member of a superposed collection is itself a superposition of another collection containing the original state (mutual). For arbitrary GPTs these statements decouple, and the paper formalises them as independent principles.

The resulting classification is instructive:

• Classical theory admits no superposition, but non-classicality does not imply it: gbits and their minimal composition (GLT) admit none.
• Spekkens' toy theory, designed to mimic quantum phenomena via epistemic restrictions, satisfies all three principles—an observation that complicates any claim that superposition is distinctively quantum.
• Boxworld exhibits superposition only through entanglement: PR boxes are superpositions of separable deterministic distributions relative to product MDEs, while no single-system superposition exists.
• Regular $n$-gon theories display a genuinely non-quantum feature: a state can be a superposition of a pair while being perfectly discriminable from one of its members, a consequence of violated outcome sharpness of MDEs. Odd $\mathsf{D}$0-gons inherit all three principles; even $\mathsf{D}$1-gons fail uniform superposition, with mutual superposition failing at $\mathsf{D}$2 but holding for larger even $\mathsf{D}$3. The paper concedes that no higher-level explanation of this parity-dependent structural gap is known.
• Two purpose-built fragments, GPT-1 and GPT-2, demonstrate that the three principles do not form a hierarchy: GPT-1 satisfies uniform and mutual but not complete superposition; GPT-2 satisfies only uniform superposition.

Inheritance from subsystems to composites

A central technical question is when superposition properties transfer to compositions. The key hypothesis is multiplicativity of the operational dimension: the composite's MPD cardinality equals the product $\mathsf{D}$4, so that products of local MDEs remain MDEs. Notably, this fails in general—the paper constructs five perfectly distinguishable product states in any no-signalling composition of two pentagons, exceeding the multiplicative bound of four.

Under dimension multiplicativity, the paper proves that existence and completeness of superposition carry over from one subsystem alone (the other may be classical), and uniform superposition carries over given outcome sharpness of MDEs. Mutual superposition conspicuously does not carry over—and this failure is the pivot of the main result.

On the converse direction, the paper shows that entanglement implies superposition: whenever the composition strictly contains the minimal tensor product and has multiplicative operational dimension, every extremal entangled state is a superposition of product states. The proof exploits informational completeness of MDEs to show an entangled state cannot be deterministic for all product MDEs. The reverse implication fails: product states such as $\mathsf{D}$5 are superpositions in the minimal tensor product of quantum systems. Entanglement and superposition are therefore related by a strict one-way implication within this class of theories.

Maximality of the quantum tensor product

The main theorem states that no composition rule $\mathsf{D}$6 with $\mathsf{D}$7 admits mutual superposition. The proof strategy is constructive: any state $\mathsf{D}$8 outside the quantum tensor product is not positive semidefinite, hence possesses a non-separable eigenvector $\mathsf{D}$9 with negative expectation value. This rank-one projector is extremal in any intermediate composition, is a superposition relative to a product MPD set, yet cannot belong to any MPD set—because the only effect giving it probability one is the projector itself, which is excluded as an effect since $\mathsf{M}$0 is a valid state. Mutual superposition therefore fails.

Combined with the fact that the maximal tensor product retains complete and uniform superposition, this yields the titular corollary: no strict extension of the quantum tensor product simultaneously admits all three superposition principles together with the no-restriction hypothesis. Since the argument applies pairwise, it extends to multipartite systems. This provides an operational characterisation of the quantum composition rule, replacing the postulated tensor product structure with a physically interpretable principle—an alternative to prior reconstruction attempts that were either non-operational or failed to single out the quantum composition uniquely.

Two caveats qualify this result. First, it assumes the subsystems are already quantum; the theorem constrains only the composition rule, not the single-system state spaces. Second, the no-restriction hypothesis is load-bearing, so compositions with restricted effect spaces are not directly covered.

Preparational uncertainty as a form of superposition

Preparational uncertainty—the impossibility of finding a state deterministic for two MDEs $\mathsf{M}$1, i.e. $\mathsf{M}$2—is formulated relationally, with weak and strong variants abstracting features of general qudits and qubits respectively. Unlike the superposition principles, the uncertainty principles do form a hierarchy, and strong uncertainty holds only for individual qubits among quantum systems (where projectors do not intertwine, in Gleason's terminology).

The paper establishes that preparational uncertainty is strictly stronger than superposition: its existence implies complete superposition unconditionally, and uniform superposition given outcome sharpness of MDEs. Under outcome sharpness, existence and weak uncertainty can be restated entirely in superpositional terms—for instance, weak uncertainty becomes the requirement that for every MPD set there exists another MPD set all of whose elements are superpositions of the first. No analogous characterisation of strong uncertainty is known. Within polygonal theories, preparational uncertainty appears only beyond $\mathsf{M}$3 (weak) and only in the continuous limit (strong), reinforcing that strong uncertainty effectively isolates real qubit quantum theory in that class.

Limitations and open questions

Several assumptions delimit the scope of the results. The informational completeness of MDEs is imposed globally and excludes certain simplicial theories; the no-restriction hypothesis is required for the maximality theorem; and outcome sharpness of MDEs is needed for several inheritance results and for the superpositional reformulation of uncertainty. The framework is confined to finite dimensions, and the analysis of composite systems presumes multiplicative operational dimensions—which, as the pentagon example shows, is a substantive restriction rather than a generic property. Whether an operational condition singling out the quantum state space for single systems can be found, thereby removing the assumption of quantum subsystems, remains open, as does the extension to infinite-dimensional systems.

Conclusion

This paper supplies a theory-independent, statistical definition of superposition and demonstrates that it carries genuine structural content: mutual superposition singles out the quantum tensor product as the maximal no-signalling composition of quantum systems consistent with all three superposition principles under no-restriction. Along the way, entanglement and preparational uncertainty are subsumed as special cases of operational superposition, with the implications running in specific, proved directions rather than as equivalences. The result contributes a concrete, operational principle toward the reconstruction of quantum theory, contingent on the subsystems being quantum and on the stated informational assumptions.