Covariant Channels and Symmetry-Restricted Recovery

Recovery under symmetry constraints asks whether lost information can be reconstructed by a covariant channel, not merely by an unrestricted mathematical map. Twirling can convert a recovery map into a covariant one for symmetric ensembles, while asymmetric inputs can suffer a performance cost. A finite reference frame can reduce that cost, but it must be included as a resource.

Required background. Symmetry-constrained operations defines covariance; recovery and approximate Markovianity supplies unrestricted recovery bounds.

Helpful background. Superselection and accessible entanglement supplies the sector-by-sector operational decomposition.

Covariant noise and recovery

Let $U_g$ act on the input and $V_g$ on the output. A noise channel $\mathcal N$ is covariant when

$$\mathcal N(U_g\rho U_g^\dagger) =V_g\mathcal N(\rho)V_g^\dagger.$$

A recovery $\mathcal R$ is covariant in the reverse direction when

$$\mathcal R(V_g X V_g^\dagger) =U_g\mathcal R(X)U_g^\dagger.$$

The recovery error should be reported for a state family or code ensemble, resolved by symmetry sector when relevant. An average fidelity can hide a sector with poor performance.

Covariant recovery is the downstream channel branch. It inherits the sector decomposition and reference budget from the resource theory rather than optimizing over all mathematical channels. Schematic and not to scale.

Twirling a recovery map

For compact $G$, any CPTP map $\mathcal R$ can be averaged to

$$\mathcal R_G(X) =\int_G dg\, U_g\,\mathcal R(V_g^\dagger X V_g)\,U_g^\dagger.$$

The result is CPTP and covariant. For a group-invariant ensemble and group-invariant average loss, this averaging preserves the ensemble performance. It need not preserve pointwise performance on one asymmetric state. Thus “twirl the optimal recovery” is justified only after the objective is shown to be symmetric.

The group-twirling and covariant-operation framework is Bartlett, Rudolph, and Spekkens 2007, §§ II.C–II.D, and asymmetry monotonicity under such channels is Marvian and Spekkens 2014, Theorem 1.

If $\rho$ and the reference $\sigma$ are invariant and $\mathcal N$ is covariant, rotated Petz maps can often be averaged without losing the relevant symmetric recovery guarantee. If the reference breaks the symmetry, its asymmetry is part of the construction.

The universal rotated recovery map and its relative-entropy guarantee are Junge et al. 2018, Theorem 2.1.

Sector-resolved error

Suppose

$$\rho=\bigoplus_q p_q\rho_q.$$

A covariant channel maps charge modes according to the group representation and may erase coherence between sectors while preserving sector populations. Useful diagnostics include the classical error in $p_q$, the conditional fidelity in each $q$, and the lost asymmetry

$$A_G(\rho)-A_G(\mathcal N\rho).$$

These quantities are not interchangeable. Perfect recovery of charge probabilities does not recover phases, and a high total fidelity in the dominant sector can hide failure in rare sectors.

Charge-erased Gaussian code

Encode a logical degree of freedom in two Gaussian modes with different charge-mode coherences. Let noise dephase the relative $U(1)$ phase while retaining number statistics. An unrestricted recovery supplied with an external phase standard may reprepare the coherence. A strictly covariant recovery without a reference cannot create asymmetry from a symmetric output.

Benchmark three cases:

1. unrestricted recovery with an idealized phase reference;
2. covariant recovery with no reference; and
3. covariant joint recovery using a finite reference state.

Report logical entanglement fidelity, sector-probability error, and the reference’s asymmetry loss. The third case should approach the first as the reference grows, while repeated use reveals degradation.

Locality, gauge constraints, and anomalies

Group covariance does not imply spacetime locality. A recovery averaged over a global symmetry can still require nonlocal access to the entire output. QFT applications must add localization, energy, and causal constraints. In gauge theories the map must act on the gauge-invariant algebra and respect its center. Anomalies can obstruct a local symmetric realization even when a formal group average exists.

Validity map for covariant recovery. The symmetry, algebra, operation class, reference state, and sector resolution must be fixed before recovery performance is compared. Covariance alone does not guarantee locality or physical implementability. Schematic and not to scale.

Common pitfalls

Assuming twirling preserves every optimum. It preserves symmetric average objectives, not arbitrary pointwise objectives on asymmetric states.

Hiding a reference in the decoder. A phase standard lets a decoder create apparent sector coherence. Count its preparation and degradation.

Equating covariance with locality. A globally covariant map may still be nonlocal, energy intensive, or incompatible with a gauge center.

References

• Bartlett, Stephen D., Terry Rudolph, and Robert W. Spekkens. “Reference Frames, Superselection Rules, and Quantum Information.” Reviews of Modern Physics 79 (2007): 555–609. DOI.
• Junge, Marius, Renato Renner, David Sutter, Mark M. Wilde, and Andreas Winter. “Universal Recovery Maps and Approximate Sufficiency of Quantum Relative Entropy.” Annales Henri Poincaré 19 (2018): 2955–2978. DOI.
• Marvian, Iman, and Robert W. Spekkens. “Extending Noether’s Theorem by Quantifying the Asymmetry of Quantum States.” Nature Communications 5 (2014): 3821. DOI.