Physicists Shubhayan Sarkar, Armin Tavakoli Orthey Jr., and Remigiusz Augusiak have published a scheme in Nature Physics that can verify any quantum state or measurement without trusting the device performing the task. The method works inside a star-shaped quantum network and covers territory that prior self-testing protocols could not reach, including mixed states and non-projective measurements. If the technique proves practical under real-world noise, it could reshape how engineers and regulators confirm that quantum hardware actually does what its makers promise.
What Self-Testing Means for Quantum Trust
Self-testing is the strongest form of device-independent certification available in quantum information science. The idea is simple in principle: a user feeds inputs into a black-box device, collects output statistics, and from those statistics alone deduces what quantum state the device prepared and what measurement it performed. No assumptions about the internal workings of the hardware are needed. The concept traces back to foundational work on Bell nonlocality, including the CHSH inequality and the Mayers–Yao framework, as described in an Oxford reference on device-independent certification. Because the guarantee holds even against a dishonest or faulty manufacturer, self-testing sits at the top of the verification hierarchy for quantum technologies.
The catch has always been scope. Traditional self-testing results applied to narrow classes of quantum objects: specific entangled states, specific projective measurements, or specific Bell scenarios with two parties. Extending the technique to arbitrary states and arbitrary measurements, especially those that are mixed, composite, non-projective, or non-extremal, remained an open problem for years. That gap matters because practical quantum devices rarely operate with the clean, pure-state, projective-measurement setups that earlier theorems assumed.
How the Star Network Closes the Gap
The new scheme sidesteps the limits of standard two-party Bell tests by placing the certification task inside a star quantum network. In this topology, a central node shares entanglement with multiple peripheral parties. The correlations observed across the network carry enough structure to pin down quantum resources that a simple Bell experiment cannot distinguish. The Nature Physics article shows that this network-assisted approach can indirectly certify any quantum state, including mixed states, and any quantum measurement, including composite, non-projective, and non-extremal types, through a reduction argument that maps unknown devices to a canonical form.
The strategy builds on two lines of prior work. One line showed that quantum networks beyond standard Bell scenarios can self-test any pure entangled state shared among an arbitrary number of subsystems, establishing that complex multipartite resources are, in principle, characterizable from correlations alone. The other established that all real projective measurements are self-testable, broadening the catalogue of certifiable measurement devices. The 2026 result generalizes both directions at once. It removes the restriction to pure states and extends measurement certification past projective and extremal cases. The full derivations and proofs are also available as an open preprint on arXiv.
Why Mixed States and Non-Projective Measurements Matter
Most coverage of quantum self-testing treats it as an abstract mathematical achievement, but the practical stakes are concrete. Quantum computers, sensors, and communication links routinely deal with mixed states, which arise whenever a system interacts with its environment or when only partial information about a larger entangled system is available. A certification method that works only for pure states is like a food-safety test that works only in a sterile lab but not in an actual kitchen. Mixed states are the norm in any realistic device, especially in noisy intermediate-scale quantum processors.
Non-projective measurements, formally known as positive operator-valued measures or POVMs, are equally common in real devices. Quantum key distribution protocols, quantum state tomography routines, and certain quantum computing gates all rely on generalized measurements that fall outside the projective category. Previous self-testing results could not certify these operations device-independently. The new scheme handles them through a reduction technique: it embeds a general POVM into a larger star-network experiment where the effective measurements become equivalent, up to well-controlled transformations, to objects that the network structure can already self-test. In effect, the hard problem of characterizing an arbitrary measurement is converted into a network problem whose correlations uniquely identify the underlying operators.
For regulators, cloud customers, and national laboratories, this expansion of scope translates into a stronger notion of “quantum compliance.” Instead of taking a vendor’s word that a device implements a certain noisy gate set or measurement protocol, users could, in principle, run a carefully designed network experiment and infer the implemented operations directly from statistics, even if the internal design is proprietary or inaccessible.
Earlier Milestones That Set the Stage
The result did not emerge in isolation. A separate line of research tackled self-testing in higher-dimensional systems under tight measurement-count constraints, addressing arbitrary local dimension with minimal measurements. That work quantified how many distinct settings are needed to certify a system as its dimensionality grows, a scaling question that any universal scheme must eventually answer for practical deployment. It highlighted that, while universality is mathematically appealing, resource overheads can quickly become a bottleneck.
On the experimental side, researchers demonstrated self-testing of a single quantum system using a trapped calcium-40 ion, moving the concept from pure theory into the lab. This single-ion experiment expanded self-testing beyond multipartite Bell nonlocality into a contextuality setting and introduced a robustness-curve methodology that quantifies how much noise a protocol can tolerate before certification breaks down. Those robustness curves serve as a template for evaluating how theoretical guarantees fare when confronted with decoherence, miscalibration, and detector inefficiencies.
An earlier universal framework for the prepare-and-measure scenario also contributed key ideas about certifying overlaps between preparation states and measurement operators in finite dimensions. Unlike Bell-based schemes, prepare-and-measure protocols assume only classical control over state preparation and measurement choice, making them closer in spirit to many quantum communication and sensing architectures. Techniques developed in that context, such as semi-definite programming relaxations and dimension witnesses, inform how more complex network-based self-testing might be optimized.
The Noise Problem Still Looms
A theorem guaranteeing that any state or measurement can be self-tested is not the same as a protocol that works in a noisy lab with lossy detectors and imperfect sources. Robustness, the ability of a self-testing statement to degrade gracefully rather than collapse entirely when experimental imperfections are present, remains the main barrier between theory and deployment. In many existing schemes, even modest levels of noise can weaken the certification so much that only a vague statement about the device remains, undermining the value of device independence.
A very recent preprint on optimized robustness in self-testing focuses specifically on this gap, analyzing how universal schemes can be made noise-tolerant across multiple scenarios. It explores how to choose measurement settings, network topologies, and data-processing routines that maximize the certified fidelity of states and measurements for a given noise level. Ideas from that work could, in principle, be combined with the star-network construction to yield protocols that are both universal in scope and practical in realistic conditions.
For now, the new star-network theorem is best viewed as a conceptual ceiling: it tells researchers what is possible in an idealized limit and defines a target for more robust, resource-efficient protocols to approximate. Experimentalists will need to translate the abstract network into concrete architectures, perhaps using photonic entanglement sources, ion traps, or superconducting qubits, and then map out the noise thresholds at which meaningful certification is still achievable.
Even with these caveats, the work by Sarkar, Tavakoli Orthey Jr., and Augusiak marks a turning point. It closes a long-standing theoretical gap by showing that, at least in principle, every quantum state and every measurement can be placed under the harsh spotlight of device-independent scrutiny. As quantum technologies move from prototypes to infrastructure, such tools for building and auditing trust may prove as important as any speedup a quantum algorithm can offer.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
