Which quantum channels destroy or preserve measurement incompatibility?

Incompatibility breaking means every measurement family becomes simulable

A family of measurements is compatible when one parent measurement plus classical post-processing can reproduce all of its statistics. Incompatible measurements cannot be written that way, so implementing them costs extra apparatus, control hardware, or quantum memory, which is why the paper treats incompatibility as a resource.

A channel acts on measurements in the Heisenberg picture, sending an assemblage {M_a|x} to {N†(M_a|x)}. The paper separates three strengths of destruction: M-incompatibility breaking for one specified assemblage, PVM-incompatibility breaking for every assemblage of projective measurements, and full incompatibility breaking for every measurement assemblage. The breaking results focus mainly on projective measurements, while the preservation results cover the most general POVM assemblages.

Zonoids turn dichotomic compatibility into a geometric inclusion

The technical engine is the theory of zonoids and zonotopes. For dichotomic measurements the approach gives a complete characterization of compatibility, and in the qubit case it reduces the problem to centrally symmetric zonoids in three dimensions, where Bloch vectors of POVMs live in the unit ball.

This geometric rewriting is what makes the rest of the paper possible. It yields a compatibility criterion stated as zonoid inclusion, gives a comparatively simple proof that compatibility survives convergent limits of compatible assemblages, and produces an analytic proof of the conjectured compatibility threshold for arbitrary finite families of noisy planar qubit PVMs.

The unital criterion is exact, but nonunital channels only get sufficiency

For a general qubit channel the paper derives a geometric criterion guaranteeing that a prescribed family of spin measurements becomes compatible after passing through the channel. For unital channels this condition is also necessary, so applied to the full family of qubit PVMs it becomes an exact PVM-incompatibility-breaking criterion for unital qubit channels.

For nonunital channels the same condition remains sufficient but not necessary. That asymmetry matters for reading the result: the exact boundary is established for unital qubit channels, while for nonunital ones the criterion can certify breaking but cannot be used to conclude that breaking fails.

At the unital boundary the parent POVM is essentially unique

At the unital incompatibility-breaking boundary the paper proves a rigidity result that effectively fixes the parent POVM. A concrete consequence is that the local hidden state model for two-qubit Werner states at the compatibility threshold r = 1/2 is essentially unique.

The authors describe this as the first uniqueness result for any local hidden state or hidden variable model, while noting that a somewhat similar uniqueness result was announced independently while the manuscript was being finalized. The uniqueness is tied to the boundary case, not to Werner states at arbitrary noise.

Two incompatibility-breaking channels can activate incompatibility in parallel

Incompatibility breaking is not stable under tensor products. The qubit depolarizing channel at 1/2 is incompatibility breaking, yet its tensor product with itself is not, so two copies can activate incompatibility even though each individual channel destroys it.

Through the channel–state correspondence the same construction gives a two-copy activation of quantum steering. Two copies of the two-qubit Werner state can be steered at noise threshold 1/2, even though steering is impossible with just one copy. The threshold 1/2 belongs to the two-copy scenario; it is not a single-copy steering threshold.

Matching dimensions force incompatibility preservation to be unitary

For preservation the paper proves that a channel with matching input and output dimensions preserves general incompatibility if and only if it is a unitary transformation. The qubit case is first proved geometrically through the zonoid picture, then the classification is extended to arbitrary finite dimensions.

The matching-dimension condition is essential. When the output dimension is larger the equivalence no longer holds, and the paper offers a conjectured form of a general incompatibility-preserving channel rather than a theorem. Mapped back to steering, a bipartite state with equal local dimensions is incompatibility preserving from Alice to Bob if and only if it is pure and has full Schmidt rank.

The decisive split is between one copy and two

The most consequential result here is not a single threshold but the failure of a natural assumption: breaking incompatibility on each channel separately does not imply breaking it on their tensor product. The depolarizing channel at 1/2 is the concrete witness, and the Werner-state steering statement at threshold 1/2 is its operational translation.

Read alongside the preservation theorem, the paper draws a sharp line. Preservation with matching dimensions collapses to unitarity, so no genuinely noisy channel preserves general incompatibility there, whereas destruction is fragile under parallel use. For a reader deciding what to trust, the exact unital breaking criterion and the two-copy activation are the results with the clearest scope; the nonunital criterion is one-directional, and the general preservation classification beyond matching dimensions remains conjectural.

Sources used in this answer