Local State Antidistinguishability (LSAD)

• Local State Antidistinguishability (LSAD) is a quantum framework where spatially separated parties use LOCC to definitively exclude the state that was not prepared.
• The paradigm contrasts global POVMs with LOCC measurements, uncovering nonlocality without entanglement and a measurable local-versus-global gap.
• LSAD extends to local state antimarking (LSAM), highlighting activation phenomena and offering a rich hierarchy compared to discrimination and marking tasks.

Local state antidistinguishability (LSAD) is the multipartite, LOCC-constrained version of quantum-state antidistinguishability: instead of identifying which state was prepared, the parties must identify a state that the system was definitively not prepared in. In this framework, the relevant operational contrast is between global POVMs on the full composite system and local operations with classical communication (LOCC) performed by spatially separated parties. Recent work places LSAD within a broader theory of state exclusion, shows that exclusion can exhibit nonlocality without entanglement, and extends the paradigm to local state antimarking (LSAM), where the object to be excluded is a sequence rather than a single state (Chatterjee et al., 10 May 2026, Manna et al., 17 Feb 2026).

1. Formal framework

For an ensemble $S=\{\rho_1,\dots,\rho_k\}$ on a Hilbert space $\mathcal H$, strong antidistinguishability means that there exists a POVM $M=\{\Pi_1,\dots,\Pi_k\}$ such that

$$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$

together with the outcome-relevance condition

$$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$

The first condition is perfect exclusion: if outcome $j$ occurs, then the state was definitely not $\rho_j$. The second condition excludes redundant POVM elements. Weak antidistinguishability drops the outcome-relevance requirement and demands only $\operatorname{Tr}[\rho_j\Pi_j]=0$. A parallel formulation in the exclusion literature writes the POVM as $\{M_j\}$ with $M_j\ge 0$, $\mathcal H$0, and $\mathcal H$1; in that language the set is exactly or perfectly antidistinguishable when exclusion succeeds with certainty (Chatterjee et al., 10 May 2026, Manna et al., 17 Feb 2026).

In the multipartite setting $\mathcal H$2, an ensemble $\mathcal H$3 of global states $\mathcal H$4 is locally antidistinguishable if there is a joint measurement implementable by LOCC whose elements satisfy the same exclusion conditions. Global antidistinguishability allows a single party holding the full system to perform the POVM; LSAD restricts the measurement to local operations and rounds of classical communication. This restriction is the source of the local-versus-global gap studied as exclusion-based nonlocality without entanglement (Chatterjee et al., 10 May 2026, Manna et al., 17 Feb 2026).

A further generalization is $\mathcal H$5-antidistinguishability, where each measurement outcome rules out an $\mathcal H$6-subset of labels rather than a single label. Weak $\mathcal H$7-antidistinguishability requires the exclusion constraints for each $\mathcal H$8-subset $\mathcal H$9; strong $M=\{\Pi_1,\dots,\Pi_k\}$0-antidistinguishability additionally requires that every $M=\{\Pi_1,\dots,\Pi_k\}$1-subset actually occur as an outcome, i.e. no $M=\{\Pi_1,\dots,\Pi_k\}$2 (Manna et al., 17 Feb 2026).

2. Mathematical criteria and analytic tools

A necessary and sufficient operator condition for strong antidistinguishability is the existence of positive weights $M=\{\Pi_1,\dots,\Pi_k\}$3 such that

$M=\{\Pi_1,\dots,\Pi_k\}$4

For pure states $M=\{\Pi_1,\dots,\Pi_k\}$5 on $M=\{\Pi_1,\dots,\Pi_k\}$6, this becomes

$M=\{\Pi_1,\dots,\Pi_k\}$7

For three pure states $M=\{\Pi_1,\dots,\Pi_k\}$8, if $M=\{\Pi_1,\dots,\Pi_k\}$9, strong antidistinguishability is characterized by the Caves–Fuchs–Schack conditions

$$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$0

and

$$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$1

These overlap inequalities recur in explicit exclusion and antimarking constructions (Chatterjee et al., 10 May 2026, Manna et al., 17 Feb 2026).

The exclusion literature also introduces an antidistinguishability figure of merit,

$$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$2

with $$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$3 if and only if the set is perfectly antidistinguishable. Perfect antidistinguishability and $$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$4-antidistinguishability can be cast as semidefinite programs, and explicit examples were analyzed by SDP methods. Alongside SDP, combinatorial and operational arguments are used to prove LOCC impossibility results for product-state ensembles (Manna et al., 17 Feb 2026).

3. Orthogonal multipartite pure states under LOCC exclusion

A central structural result is that any ensemble of mutually orthogonal multipartite pure states is locally antidistinguishable (Chatterjee et al., 10 May 2026). The proof proceeds from the two-state case established by Walgate and Hardy: any two orthogonal pure states $$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$5 can be perfectly distinguished by one-way LOCC, with Alice measuring in a suitable basis so that, conditioned on her outcome, Bob’s conditional states are orthogonal. Reinterpreted as exclusion, the outcome “$$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$6” rules out $$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$7, and the outcome “$$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$8” rules out $$\operatorname{Tr}[\rho_j \Pi_j]=0 \quad \text{for all } j=1,\dots,k,$$9.

From there, one chooses orthogonal pairs and runs their discrimination protocol. By randomly choosing pairs, using shared randomness, one can ensure that each global state $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$0 is excluded by some protocol run. A convex-combination argument then assembles these two-state exclusion protocols into a full $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$1-outcome LOCC measurement with $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$2 for every $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$3 and no redundant $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$4 (Chatterjee et al., 10 May 2026).

The significance of this theorem is comparative rather than merely formal. Because every orthogonal pure-state ensemble is LSAD-local, nonlocality without entanglement can survive discrimination yet vanish for exclusion. This suggests that, in the orthogonal pure-state regime, exclusion is operationally weaker than discrimination as a witness of local-versus-global separation (Chatterjee et al., 10 May 2026).

4. Exclusion-based nonlocality without entanglement

State exclusion nevertheless exhibits nonlocality without entanglement. A key result is that three bipartite product states can be globally antidistinguishable while failing to be LOCC antidistinguishable, and three is the minimal number of states required for this phenomenon (Manna et al., 17 Feb 2026). The construction considers three product states on $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$5,

$$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$6

for which the overlap parameters

$$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$7

satisfy the Caves–Fuchs–Schack inequalities, so a global exclusion POVM exists. At the same time, each party’s local ensemble fails antidistinguishability, and therefore no LOCC protocol can realize the same exclusion task.

For bipartite product states, LSAD also obeys a starter-independence theorem: if a set is LOCC-antidistinguishable with Alice starting, then the same protocol works with Bob starting. In fact, one need not communicate at all—each party’s local ensemble must already be antidistinguishable. This symmetry is specific to the $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$8 exclusion task and breaks down for higher-order $$\forall j\;\exists i \text{ with } \operatorname{Tr}[\rho_i \Pi_j]>0.$$9-antidistinguishability (Manna et al., 17 Feb 2026).

The same framework yields a genuine tripartite form of exclusion nonlocality. The three product states

$j$0

are globally antidistinguishable, yet across every bipartition at least one local ensemble is $j$1, which is not antidistinguishable. Consequently, no LOCC protocol succeeds even if two parties are allowed to join; this is genuine nonlocality without entanglement in the exclusion setting (Manna et al., 17 Feb 2026).

5. Local state antimarking and activation phenomena

Local state antimarking (LSAM) extends exclusion from single states to non-repetitive sequences drawn from a known set of multipartite states (Chatterjee et al., 10 May 2026). In local state marking, one is given $j$2 states drawn from a known set of orthogonal states and must identify their exact permutation. Antimarking inverts the goal: one must exhibit a permutation that definitely was not supplied.

For a known set $j$3, a referee draws a length-$j$4 sequence of distinct indices $j$5 without replacement and distributes $j$6 among $j$7 parties. An LOCC protocol $j$8 outputs $j$9 distinct candidate sequences $\rho_j$0, each guaranteed not to equal the actual $\rho_j$1. If $\rho_j$2, the task reduces to LSAD on the length-1 case; if $\rho_j$3 and perfect, it coincides with global antimarking (Chatterjee et al., 10 May 2026).

The principal LSAM example uses the parent set

$\rho_j$4

on $\rho_j$5. This set is not globally antidistinguishable: applying the Caves overlap criterion shows that no strong exclusion POVM exists. However, choosing states from $\rho_j$6 without replacement and passing to length-2 sequences produces a child set $\rho_j$7 of six ordered pairs. These six states can be partitioned into three triples, each satisfying the Caves conditions, so there is a global 6-outcome POVM that excludes one child state per outcome; in the terminology of the paper, $\rho_j$8 admits $\rho_j$9-global antimarking (Chatterjee et al., 10 May 2026).

The same child set fails locally. Alice’s and Bob’s two-qubit reduced ensembles are again $\operatorname{Tr}[\rho_j\Pi_j]=0$0, which violate the Caves condition. No strong exclusion POVM exists even on one party’s side, so no LOCC protocol can exclude any single child sequence with certainty. The example therefore reveals a form of nonlocality without entanglement that is activated only at the antimarking level. A plausible implication is that local-versus-global gaps can remain hidden at the level of single-state exclusion and become visible only when exclusion is lifted to structured sequence tasks (Chatterjee et al., 10 May 2026).

6. Relations to discrimination, marking, and open directions

The LSAD and LSAM frameworks are compared directly with conclusive local state discrimination (CLSD) and conclusive local state marking (CLSM). In CLSD one seeks unambiguous discrimination under LOCC via a POVM $\operatorname{Tr}[\rho_j\Pi_j]=0$1 such that $\operatorname{Tr}[\rho_j\Pi_j]=0$2 for $\operatorname{Tr}[\rho_j\Pi_j]=0$3 and $\operatorname{Tr}[\rho_j\Pi_j]=0$4 for $\operatorname{Tr}[\rho_j\Pi_j]=0$5, with an inconclusive outcome allowed. CLSM is the marking analogue, where one must identify the exact order of a permutation unambiguously under LOCC (Chatterjee et al., 10 May 2026).

No strict hierarchy exists among these paradigms. The four Bell states are CLSD-nonlocal but LSAD-local. Duan’s four-state product ensemble is CLSD-nonlocal and LSAD-nonlocal, yet it is locally antimarkable, so CLSM-nonlocality is not the same notion as LSAM-nonlocality. More generally, there exist product-state ensembles that permit one task while strictly forbidding the other, and vice versa; the cited work formulates this as the absence of any strict hierarchy among LSD, LSM, LSAD, LSAM, CLSD, and CLSM (Chatterjee et al., 10 May 2026).

Several broader implications follow. Exclusion tasks reveal striking differences from state discrimination, and the breakdown of starter-symmetry for $\operatorname{Tr}[\rho_j\Pi_j]=0$6 points to a richer hierarchy of LOCC capabilities in exclusion tasks than in discrimination. Operationally, these tasks may inform cryptographic primitives where one needs to deny rather than assert the presence of given states or sequences. Open questions include whether there exist mixed-state ensembles that remain LSAD-nonlocal, whether there are bipartite examples of hidden LSAM-nonlocality analogous to the tripartite activation result, how to characterize the full LOCC-versus-global gap for general $\operatorname{Tr}[\rho_j\Pi_j]=0$7 and $\operatorname{Tr}[\rho_j\Pi_j]=0$8, and how separable measurements compare with LOCC in exclusion problems (Chatterjee et al., 10 May 2026, Manna et al., 17 Feb 2026).