Learning Local Fermionic Lindbladians under Parity Superselection

2026-09-30

AI Summary

This study addresses the challenge that fermion parity superselection prohibits direct measurement of odd Majorana observables, thereby hindering the learning of fermionic Lindblad operators. To overcome this, we introduce internal markers to convert odd probes into even observables, combining paired measurements with signed Fierz inversion to recover the coefficients. We further propose a novel marker-calibrated mechanism for isolating dynamical contributions and employ semidefinite programming to fit physically valid Lindblad operators. This work breaks the superselection barrier, enabling local dynamics learning without external ancilla modes or prior knowledge of non-zero coefficients. On known interaction graphs, it achieves element-wise error recovery with near-optimal sample complexity. Moreover, for geometrically local models, predicting fixed-support observables requires only logarithmic sample complexity relative to system size.

Abstract

Parity superselection forbids direct measurement of odd Majorana observables. We learn time-independent, parity-covariant, $k$-mode-local Lindbladians on $m$ modes using even preparations and measurements with uninterrupted short-time evolution. Internal markers turn odd probes into even observables, and pair measurements among three separated markers calibrate their dynamical contributions. Signed Fierz inversion recovers canonical coefficients, while semidefinite fitting yields a valid generator. For finite-range models on known bounded-degree graphs with suitable marker access and a supplied weighted-strength bound $\barα$, entrywise error $\varepsilon$ is achieved using $\widetilde{\mathcal{O}}(\barα^2\varepsilon^{-2}\log(m/δ))$ samples, without external modes or known nonzero coefficient locations. Recovery to diamond-norm error $\varepsilon$ costs an additional factor $m^2$, matching lower bounds for short-time experiments on fresh systems up to logarithmic and fixed geometric factors. Without a supplied graph, one idle ancillary mode per system mode and potentially nonlocal pair operations give total absolute coefficient error at most $\varepsilon$ per mode using $\widetilde{\mathcal{O}}_k(\barα^2\mathsf d^2m^{\lfloor k/2\rfloor}\varepsilon^{-2}\log(m/δ))$ samples, under a supplied approximate coefficient-degree bound $\mathsf d$ and controlled weak-coefficient tails. All guarantees hold with probability at least $1-δ$. For geometric models, fixed-support even observables can be predicted with logarithmic system-size sample complexity at fixed time and accuracy. We also give finite-volume and exponential-tail extensions.

Problem

• Fermionic Lindbladians
• Parity superselection
• Quantum process learning
• Open quantum systems
• Majorana observables

Innovation

• Fermionic Lindbladians
• Parity superselection
• Internal markers
• Signed Fierz inversion
• Semidefinite fitting