Orthogonality-Preserving LOCC (OP-LOCC)

• OP-LOCC is a framework for local state discrimination where each measurement preserves the mutual orthogonality of surviving candidate states.
• One-way protocols prove that for two orthogonal pure states a nontrivial, orthogonality-preserving local measurement always exists, ensuring efficient discrimination.
• In multipartite and entangled scenarios, OP-LOCC highlights limits in local discrimination while enabling resource-assisted recovery of global identifiability.

to=arxiv_search ็ตทรูjson {"query":"orthogonality-preserving LOCC local distinguishability orthogonal states OP-LOCC", "max_results": 10} Orthogonality-preserving LOCC (OP-LOCC) is the branchwise formulation of local state discrimination in which each local measurement is required to preserve mutual orthogonality among the candidate states that survive a given classical outcome. For bipartite states expanded as $$|\tau\rangle_{AB}=\sum_e |e\rangle_A\otimes |\tau_B^{|e}\rangle$$, a local measurement in basis $\{|e\rangle\}$ is orthogonality-preserving for two candidates $|\psi\rangle,|\phi\rangle$ exactly when $\langle \phi^{|e}|\psi^{|e}\rangle=0$ for every outcome $e$ (George et al., 27 Jan 2026). In the literature, the same structural idea also appears under the names of non-disturbing first measurements, reducibility versus irreducibility, branchwise conditional orthogonality, or orthogonality-preserving rank-one POVMs. OP-LOCC is therefore best understood as a unifying constraint on perfect or near-perfect LOCC discrimination rather than as a single narrowly delimited protocol class.

1. Conceptual foundations and operational meaning

The central OP-LOCC question is whether a local party can take a nontrivial first step without destroying the orthogonality needed for later rounds. In one-way protocols this is especially transparent: Alice measures, communicates the outcome, and Bob must finish the task on the corresponding conditional states. For two states, the branchwise condition is simply $\langle \phi^{|e}|\psi^{|e}\rangle=0$ for all outcomes $e$; in multipartite iterative use, each successive party is required to preserve orthogonality on every branch that remains unresolved (George et al., 27 Jan 2026).

A closely related vocabulary is reducibility. In product-state discrimination, a set is reducible from a party’s side if that party has a nontrivial local move that splits the candidates into orthogonal blocks; it is irreducible if no such move exists. This is the same obstruction that OP-LOCC isolates: irreducibility means the absence of a nontrivial orthogonality-preserving local measurement for the current residual set (Mal et al., 2020). The distinction matters because perfect global discrimination of orthogonal states is trivial, whereas perfect local discrimination depends on whether orthogonality can be converted into locally accessible information without being disturbed.

A recurrent theme is that many results central to OP-LOCC are not stated using that exact label. Some papers study orthogonality-preserving rank-one POVMs, some conditional-state decompositions, some local irreducibility, and some operator-space criteria. The underlying issue is the same: whether each local branch can leave the remaining possibilities orthogonal enough for continuation.

2. One-way OP-LOCC and constructive characterizations

For two orthogonal pure states, OP-LOCC is completely understood in the one-way setting. Walgate, Short, Hardy, and Vedral proved the finite-dimensional result, and a later generalization extended it to infinite dimensions and multipartite systems with a simpler proof and a constructive finite-dimensional algorithm. The core fact is that every trace-class operator with vanishing trace admits a basis whose diagonal entries are all zero, and with $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$ this yields a basis $\{|e\rangle\}$ such that $\langle \phi^{|e}|\psi^{|e}\rangle=0$ for all $\{|e\rangle\}$0 (George et al., 27 Jan 2026). In finite dimensions $\{|e\rangle\}$1, the paper constructs such a one-way protocol in $\{|e\rangle\}$2 time (George et al., 27 Jan 2026). This gives an exact OP theorem for the simplest nontrivial case: an orthogonality-preserving first local measurement always exists for two orthogonal pure states.

A more general one-way framework was given for arbitrary orthogonal bipartite states, including mixed states, through the Hermitian subspace $\{|e\rangle\}$3 associated with the starting party $\{|e\rangle\}$4. If Alice starts, the states are distinguishable by one-way LOCC if and only if $\{|e\rangle\}$5 contains all elements of an extremal rank-one POVM; for projective first measurements this becomes the condition that $\{|e\rangle\}$6 contain a maximally abelian subspace (MAS) (Singal, 2015). This turns the OP-first-step problem into a linear-algebraic one: the rank-one positive operators in $\{|e\rangle\}$7 are exactly the candidates for orthogonality-preserving first outcomes.

The same framework yields sharp dimension-based criteria. If $\{|e\rangle\}$8, the starting party cannot initiate any one-way LOCC protocol; if $\{|e\rangle\}$9, distinguishability is equivalent to $|\psi\rangle,|\phi\rangle$0 being a MAS; and if $|\psi\rangle,|\phi\rangle$1, a projective orthogonality-preserving first measurement always exists (Singal, 2015). This clarifies a basic structural point: OP-LOCC at one round is not merely a heuristic but an exact criterion for one-way perfect discrimination.

3. Product-state sets, local irreducibility, and elimination phenomena

For larger sets of product states, OP-LOCC becomes substantially more rigid. Several constructive indistinguishability results proceed by showing that neither party can perform any nontrivial orthogonality-preserving first measurement. In $|\psi\rangle,|\phi\rangle$2, explicit families with $|\psi\rangle,|\phi\rangle$3, $|\psi\rangle,|\phi\rangle$4, and $|\psi\rangle,|\phi\rangle$5 orthogonal product states were shown to be LOCC-indistinguishable by exactly this method: the first local POVM element $|\psi\rangle,|\phi\rangle$6 is forced to have all off-diagonal entries zero and all diagonal entries equal, so $|\psi\rangle,|\phi\rangle$7 and every orthogonality-preserving first move is trivial (Zhang et al., 2015). These proofs are OP-LOCC arguments in their purest form.

The multipartite extension is even stronger in certain completable product bases. In $|\psi\rangle,|\phi\rangle$8, and then in $|\psi\rangle,|\phi\rangle$9 for $\langle \phi^{|e}|\psi^{|e}\rangle=0$0 and $\langle \phi^{|e}|\psi^{|e}\rangle=0$1, explicit sets $\langle \phi^{|e}|\psi^{|e}\rangle=0$2 were constructed with the property that if a complete orthogonal product basis $\langle \phi^{|e}|\psi^{|e}\rangle=0$3 contains $\langle \phi^{|e}|\psi^{|e}\rangle=0$4, then no state of $\langle \phi^{|e}|\psi^{|e}\rangle=0$5 can be eliminated by orthogonality-preserving measurements (Halder, 2016). This is stronger than ordinary LOCC-indistinguishability. The paper also exhibited a contrasting basis $\langle \phi^{|e}|\psi^{|e}\rangle=0$6 that remains LOCC-indistinguishable even though a nontrivial orthogonality-preserving first measurement can eliminate several states, showing that “no OP elimination” is a stricter notion than nonlocality without entanglement (Halder, 2016).

For arbitrary orthogonal product sets, OP-style elimination can nonetheless be effective in the many-copy regime. A polygon method showed that in bipartite systems a single copy is sufficient to exclude four states from any set of seven orthogonal product states, leading to the general bound that $\langle \phi^{|e}|\psi^{|e}\rangle=0$7 orthogonal product states are LOCC distinguishable with $\langle \phi^{|e}|\psi^{|e}\rangle=0$8 copies; in multipartite systems the bound becomes $\langle \phi^{|e}|\psi^{|e}\rangle=0$9 when $e$0 and $e$1 otherwise (Shu, 2020). These protocols are operationally OP-LOCC: each local projective step preserves orthogonality among the surviving candidates while eliminating a guaranteed number of states.

A further reformulation appears through irreducibility and generalized CNOT. For full product bases, entanglement is generated by CNOT if and only if the basis contains an irreducible subspace from the control side; equivalently, the side acting as control lacks a nontrivial orthogonality-preserving local reduction on that subspace (Mal et al., 2020). This recasts OP-LOCC obstruction as a dynamical entanglement-generation criterion.

4. Entangled orthogonal sets and the limits of OP arguments

The jump from product states to entangled states changes both the possibilities and the obstructions. For three maximally entangled states $e$2, perfect one-way LOCC is possible if and only if there exist vectors $e$3 and positive numbers $e$4 such that $e$5 and $e$6 whenever $e$7 (Nathanson, 2013). This is an explicit branchwise OP condition: after Alice’s first outcome $e$8, Bob’s residual states $e$9 must be orthogonal. The same work produced triples of mutually orthogonal maximally entangled states in $\langle \phi^{|e}|\psi^{|e}\rangle=0$0 that cannot be perfectly distinguished by one-way LOCC for every even $\langle \phi^{|e}|\psi^{|e}\rangle=0$1 and every $\langle \phi^{|e}|\psi^{|e}\rangle=0$2, while two-way LOCC does distinguish them (Nathanson, 2013). It also showed that any three orthogonal maximally entangled states are perfectly distinguishable by PPT measurements and are distinguishable by one-way LOCC with error probability at most $\langle \phi^{|e}|\psi^{|e}\rangle=0$3 (Nathanson, 2013).

For larger entangled sets, OP preservation is not sufficient. A constructive family of $\langle \phi^{|e}|\psi^{|e}\rangle=0$4 orthogonal maximally entangled states in $\langle \phi^{|e}|\psi^{|e}\rangle=0$5 with $\langle \phi^{|e}|\psi^{|e}\rangle=0$6 was shown to have PPT success probability at most $\langle \phi^{|e}|\psi^{|e}\rangle=0$7, implying PPT and hence LOCC indistinguishability whenever $\langle \phi^{|e}|\psi^{|e}\rangle=0$8; for $\langle \phi^{|e}|\psi^{|e}\rangle=0$9, this gives explicit examples with $e$0 (Cosentino et al., 2013). The OP lesson is negative but important: globally orthogonal maximally entangled states may remain locally indistinguishable even when the set size is strictly below the local dimension.

Entanglement assistance changes the picture again. For a complete orthonormal maximally entangled basis $e$1 assisted by a pure resource state $e$2, the optimal LOCC success probability is exactly the fully entangled fraction,

$e$3

and this equals the PPT optimum as well, so separable and PPT measurements give no advantage over LOCC for that task (Bandyopadhyay et al., 2024). Perfect discrimination is possible if and only if the resource is maximally entangled. The protocol is teleportation-based rather than explicitly branchwise OP-LOCC, and the paper is careful not to claim that every intermediate local step preserves orthogonality in the formal OP sense (Bandyopadhyay et al., 2024). It is nevertheless highly relevant because it quantifies how shared entanglement restores local distinguishability of an orthogonal maximally entangled basis.

5. Asymptotic, many-copy, and resource-assisted OP viewpoints

The asymptotic version of OP-LOCC asks what remains necessary when finite protocols are replaced by arbitrarily accurate approximations. A necessary condition for perfect asymptotic LOCC discrimination of orthogonal states is that, for every $e$4, there exist product effects $e$5 satisfying $e$6, $e$7, and

$e$8

The last condition is an asymptotic OP requirement: the product filter $e$9 must preserve mutual orthogonality of the filtered states (Kleinmann et al., 2011). For complete product bases, this implies that unlimited asymptotic resources provide no advantage over finite LOCC (Kleinmann et al., 2011).

In the many-copy binary setting, orthogonal-state discrimination can become exponentially reliable even when one-shot perfect local discrimination is not the central issue. For testing an arbitrary multipartite entangled pure state against its orthogonal complement under LOCC, the optimal average error probability always decays exponentially with the number of copies; the paper also gave a sufficient condition under which LOCC, separable, and PPT operations have exactly the same performance (Cheng et al., 2020). Maximally entangled states versus their orthogonal complements, and extremal Werner states, satisfy that condition (Cheng et al., 2020). This is not a direct OP-LOCC theorem, but it shows that locality-restricted discrimination of orthogonal states has a robust asymptotic regime beyond one-shot OP-first-step analyses.

Resource assistance for product states provides a different extension. For certain multipartite orthogonal product sets constructed by Wang et al., shared entanglement enlarges the local Hilbert spaces and enables new local measurements on $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$0, $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$1, and analogous extended systems, so that the surviving branches remain orthogonal and the states become perfectly distinguishable by LOCC (Li et al., 2018). In a related construction, a single Bell pair shared between any two parties is sufficient to distinguish the completable multipartite family $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$2 discussed above, and this sufficiency does not depend on the local dimension or the number of parties (Halder, 2016). These results show that entanglement can restore branchwise orthogonality-preserving discrimination that bare LOCC forbids.

6. Structured classes, classification, and current directions

Recent work has extended OP-relevant thinking from explicit protocol trees to symmetry-reduced optimization and conversion hierarchies. For locally diagonal orthogonally invariant (LDOI) states, optimal PPT and separable measurements for discrimination can always be chosen LDOI, and the LOCC supremum can be approached by LDOI LOCC POVMs, reducing the associated optimization from $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$3 variables to $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$4 (Johnston et al., 14 Apr 2026). For orthonormal LDOI bases, efficiently computable bounds were derived, and for a broad class—including all two-qubit cases—the LOCC supremum equals the PPT and separable optima (Johnston et al., 14 Apr 2026). More generally, the gap between PPT and LOCC distinguishability is at most $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$5 (Johnston et al., 14 Apr 2026). These results are not explicit OP-LOCC characterizations, but they provide a structured search space in which OP-style questions can be sharpened.

A further shift is to treat OP-LOCC itself as a transformation resource. Locally distinguishable sets need not be “useless”: some can be converted into locally indistinguishable sets under OP-LOCC. A 2025 classification introduced hierarchies among locally distinguishable sets by identifying structures that do or do not allow such conversion, and in multipartite systems it singled out sets that cannot be converted to locally indistinguishable ones by OP-LOCC across any bipartition, a phenomenon described as “no activation across bi-partitions” (Bhunia et al., 14 Jul 2025). This moves OP-LOCC from a purely diagnostic role toward a resource-theoretic one.

Several misconceptions are corrected by the accumulated literature. Global orthogonality does not imply the existence of a nontrivial orthogonality-preserving local first move; two-state pure discrimination is exceptional rather than generic (George et al., 27 Jan 2026). Orthogonality preservation is necessary for perfect continuation, but it is not by itself sufficient for successful local discrimination of larger entangled sets (Cosentino et al., 2013). Nor is LOCC always strictly weaker than its convex relaxations in a task-independent way: some resource-assisted basis discrimination problems have $T=S_{|\phi\rangle}^*S_{|\psi\rangle}$6 at the optimum (Bandyopadhyay et al., 2024), whereas other structured families exhibit only a small but nonzero PPT–LOCC gap (Johnston et al., 14 Apr 2026).

Taken together, these results portray OP-LOCC as a structural language for local quantum discrimination. In the two-state pure case it gives an exact constructive theory; for larger product sets it identifies local irreducibility and no-elimination phenomena; for entangled sets it captures sharp one-way obstructions but also reveals limits of purely OP reasoning; and in asymptotic, resource-assisted, and symmetry-reduced settings it remains the natural lens through which the relation between global orthogonality and locally accessible information is analyzed.