Partial Stabilizer Learning under Fixed Commuting Constraints and the Absence of a Copy-Rate Discount
Masahito Hayashi, Yimin Lu
quant-ph
Sep 12, 2026 · v3
TL;DR
The main theorem on copy-rate thresholds for partial stabilizer learning is formalized in Lean 4, with source and verification records in a public repository.
Abstract
We consider an unknown pure stabilizer state of \(n\) qudits with a fixed prime local dimension \(p\), together with prescribed commuting Pauli observables. Rather than learning the whole state, the learner must recover only complementary stabilizers that commute with the prescribed observables, together with their eigenvalues. Once the prescribed measurement is performed, this information and the observed outcome determine the corresponding conditional stabilizer state. We compare the number \(k\) of copies available to the learner with the number \(n\) of qudits and refer to their ratio \(k/n\) as the copy rate. Suppose that the dimension \(m\) of the requested complementary stabilizer subspace satisfies \(m=βn+o(n)\) for a fixed \(0<β\leq1\). If the copy rate converges to a value below one, the optimal exact recovery probabilities and verification scores converge to zero, both in the worst case and under the uniform prior. Conversely, if the number \(k\) of copies minus the number \(n\) of qudits tends to positive infinity, the entire stabilizer state can be identified with probability tending to one, and the requested complementary information can then be extracted. Thus, restricting the learning target to the stabilizer information needed after the prescribed measurement yields no copy-rate discount.
Problem
The question is whether learning only the complementary stabilizer information needed after a fixed commuting Pauli measurement on an unknown n-qudit stabilizer state can use fewer copies than full stabilizer identification. The target subspace has dimension m = βn + o(n), with fixed prime local dimension p.
Approach
Two tasks are defined: exact learning and postmeasurement verification, each evaluated in the worst case and under a uniform prior. The learner is given an auxiliary Lagrangian frame, which reduces the problem to quadratic stabilizer coordinates (symmetric-matrix normal forms, with a Z4 enhancement for qubits). The resulting quotient ensemble is solved exactly at finite copy number using a pretty-good-measurement formula. A converse follows from moment-grouping, Cauchy–Schwarz and matrix-completion counting. The main theorem is formalized in Lean 4.
Results
If the copy rate k/n converges to a value below one, optimal exact recovery and verification scores tend to zero. If k − n tends to infinity, the full state can be identified with probability tending to one. Restricting the target therefore gives no copy-rate discount.