Device-informed scalable priors for quantum channels

Determine which experimentally justified assumptions about locality, symmetries, calibrated noise, or restricted system-environment couplings are sufficient to derive informative quantum-channel priors whose kernels remain efficiently resolvable, and construct such priors for channel quantum Gaussian processes that are both provable and scalable.

Background

The paper derives a quantum Gaussian-process prior for quantum channels by placing the uniform Lebesgue measure over the set of channels, obtaining a kernel whose state-overlap structure is accompanied by a dimension-dependent scale. This prior is analytically tractable but becomes poorly suited to extensive channels because the kernel scale is exponentially suppressed. The authors introduce a rescaled channel kernel as an empirical Bayes heuristic, but it is not derived from a physical channel ensemble.

The unresolved problem is to move beyond the uninformative uniform prior by incorporating experimentally available structure—such as locality, symmetries, calibrated noise, and restricted system-environment interactions—while retaining both mathematical provability and efficient kernel resolution. Solving it would provide device-informed channel QGPs with stronger inductive bias and scalable learning guarantees.

References

The remaining question is how to construct channel priors that encode more of what is actually known about a physical application and how much we trust them. The Lebesgue ensemble is appropriate when essentially no channel-specific information is available, but experimentally relevant channels often come with additional structure from locality, symmetries, calibrated noise, or restricted system-environment couplings. The unitary setting already shows that such structure can lead to provable and scalable QGPs, as occurs for matchgate evolutions. For quantum channels, the corresponding problem is to determine which experimentally justified assumptions are sufficient to derive informative priors whose kernels remain efficiently resolvable, as necessary for channel QGPs that are both provable and scalable.

— Quantum Gaussian processes for prediction of channel observations  (2608.19306 - Jäger et al., 19 Aug 2026) in Summary and discussion, Section 6