The Signal
Quantum paper bounds local measurement gap for structured states
Nathaniel Johnston and Vincent Russo published a study of how well local measurements can distinguish structured quantum states. It reduces the size of related optimization problems and bounds the gap between two measurement strategies.

Bounds for local measurements
A paper by Nathaniel Johnston and Vincent Russo puts a bound on how far local measurements can lag behind positive-partial-transpose measurements when distinguishing a broad class of structured quantum states. Published in Quantum on November 18, 2026, the study also shows how symmetry can shrink the optimization problems used to analyze such measurements.
The work examines locally diagonal orthogonally invariant (LDOI) states, which remain unchanged under a specific local averaging operation called diagonal orthogonal twirling. The class includes Werner states, isotropic states, X-states and mixtures of Dicke states. The authors study how successfully measurements can identify a state from an ensemble when the state is shared between two parties.
Reducing the search
Local operations with classical communication, or LOCC, allow the parties to measure their own parts of a shared state and communicate their results. The paper compares these measurements with separable measurements and positive-partial-transpose (PPT) measurements, which are broader measurement classes defined by mathematical conditions on the measurement operators.
For ensembles of LDOI states, the authors show that optimal PPT and separable measurements can themselves be chosen to have the same LDOI structure. The best LOCC success probability may not be attained by a single measurement, but it can be approached arbitrarily closely by LDOI LOCC measurements, the paper states.
That structure reduces the associated optimization search from a general space scaling as n to the fourth power to one scaling as n squared. For distinguishing orthonormal LDOI bases, the study provides computable upper and lower bounds. It says the LOCC lower bound can be calculated in polynomial time by turning part of the calculation into an assignment problem; the PPT upper bound uses a smaller semidefinite program.
How close are the strategies?
For a uniform ensemble drawn from an orthonormal LDOI basis in local dimension n, the authors bound the difference between PPT and LOCC success probabilities by (n − 2)/(2n²). The bound reaches its maximum of 1/16, or 0.0625, at n = 4 and vanishes as the dimension grows.
The two strategies reach the same optimum for a broad class of LDOI bases, including every two-qubit case, according to the paper. In those cases, the LOCC, separable and PPT success probabilities are equal. The result extends the equality beyond particular previously studied families, the authors write.
It identifies that as an open question and gives an example where the derived upper and lower bounds do not match, leaving open the possibility of a gap in higher dimensions.
Sources
Related stories
- A New Paper Sets Lower Bounds on Quantum Measurement Memory
- Researchers introduce a framework to upper-bound entanglement from a state's spectrum
- At Intermediate Frequencies, Resonance Alone May Not Stabilize Floquet States
- Researchers define states whose entanglement dimension resists change
- Three Researchers Derive a Unique Quantum Bayes Rule