Symmetry, Superselection, and Information Resources

Symmetry does more than label states: it restricts which observables, operations, ancillas, and communication protocols are physically available. The same density operator can therefore carry several inequivalent quantities—uncertainty in a charge sector, entanglement within a sector, coherence relative to a missing reference frame, gauge-edge data, or information recoverable by a covariant decoder. None of these quantities is an operational resource until the subsystem algebra and the allowed task have been declared.

This chapter builds a working dictionary for those declarations. It emphasizes finite regulators and exactly checkable models before taking a continuum limit, separates global symmetries from gauge redundancies, and keeps emerging soft-sector and generalized-symmetry diagnostics within their present evidence boundaries.

Helpful background. Factorization failure and local algebras supplies the continuum subsystem boundary; symmetry of a QFT supplies global action data; edge modes and subregion factorization supplies the gauge-boundary language; topological entanglement diagnostics supplies a downstream material application.

Enter this chapter

Start with symmetry-constrained operations, because “free” has meaning only relative to an operation class. Superselection and accessible entanglement then separates charge-label uncertainty from entanglement usable under charge-preserving local operations. Fermionic graded subsystems explains why an ordinary qubit tensor product does not faithfully represent odd operators without the corresponding Jordan–Wigner algebra map.

The reference and resolution branch continues through reference frames and asymmetry, charge-resolved entanglement, and entanglement-asymmetry restoration. These pages distinguish coherence supplied by a finite frame, conditional entropies reconstructed from charged moments, and local symmetry restoration under a stated order of limits.

The gauge branch fixes a regional algebra in gauge subregions and centers, matches continuum bulk and edge terms in gauge-field edge contributions, and isolates the part usable under gauge-invariant local operations in centers, edges, and distillable entanglement.

Finally, soft-sector information makes finite detector resolution part of the question, generalized-symmetry information treats defect-conditioned observables without reclassifying generalized symmetries, and covariant channels and recovery asks what a symmetry-respecting decoder can actually reconstruct.

From task specification to an information resource

The chapter’s first diagram is deliberately a branching dictionary rather than a causal chain. A block-diagonal sector state, a pre-twirled asymmetric state, and a gauge algebra with a center are different inputs; one should not feed one formula into another merely because all of them contain a charge label.

Symmetry–information dictionary. Sector entanglement, asymmetry, charged or defect moments, gauge or soft data, and covariant recovery are parallel outputs of a declared task. A reference frame, edge system, regulator, or detector response is an input to that task, not an invisible background. Schematic and not to scale.

Read the diagram from top to bottom:

1. Fix the subsystem. Name the regional algebra, including its center or fermionic grading, rather than assuming a Hilbert-space factorization.
2. Fix the symmetry data and state. Specify the group action, charge or defect labels, and whether the state is already invariant or still has coherence between sectors.
3. Fix the operational resources. State the admissible maps, physical ancillas, reference or edge systems, regulator, and sector or detector resolution.
4. Only then choose a quantity. Accessible entanglement, asymmetry, charged moments, soft distinguishability, and recovery error answer different questions even when calculated from the same regulated state.

This ordering prevents a common category error: if a state already has the Abelian form $\rho_A=\bigoplus_q p_q\rho_{A,q}$, then it has no off-diagonal charge coherence to be removed by the same $U(1)$ twirl. Its sector entropy can still be nonzero, but its relative entropy of $U(1)$ asymmetry is zero.

Sector blocks, coherence, and edge data

For a normalized state that commutes with an Abelian regional charge,

$$\rho_A=\bigoplus_q p_q\rho_{A,q}, \qquad \operatorname{Tr}\rho_{A,q}=1,$$

and the direct-sum entropy identity is

$$S(\rho_A)=H(\{p_q\})+\sum_q p_qS(\rho_{A,q}).$$

This identity is kinematic. For the standard pure-state, fixed-total-charge task with local charge-preserving operations, the within-sector average gives the accessible entanglement, while the Shannon term is not itself a one-copy Bell-pair yield. Collective copies and a supplied reference can change the resource theory, so the task must accompany the formula. The particle-number construction is given in Wiseman and Vaccaro 2003, Eqs. (1)–(4).

If $\rho_A$ instead has coherence between charge sectors, the subsystem twirl $\mathcal G_A$ removes that coherence. For a compact-group twirl in a finite or suitably regulated system,

$$A_G(\rho_A) =D\!\left(\rho_A\middle\Vert\mathcal G_A(\rho_A)\right) =S\!\left(\mathcal G_A(\rho_A)\right)-S(\rho_A).$$

This asymmetry measures distinguishability from the twirled state, not entanglement. A finite reference can make some relational coherence usable, but its state, localization, and degradation belong in the accounting Bartlett, Rudolph, and Spekkens 2007, §§ II.B–III.

Charge-resolved entropies use yet another object,

$$Z_n(\alpha)=\operatorname{Tr}\!\left(\rho_A^n e^{i\alpha Q_A}\right),$$

whose Fourier coefficients give unnormalized sector moments. The sector probability and the normalization of the conditional density matrix must be restored before taking an entropy. The construction is introduced in Goldstein and Sela 2018, Eqs. (1)–(6).

Gauge theories add an algebraic choice rather than merely another global charge. Gauss law can place boundary flux in the center of a regional gauge-invariant algebra. Electric-center, magnetic-center, and extended-Hilbert-space prescriptions need not assign the same formal entropy; the correct comparison first matches the algebra, edge system, measure, and regulator. The finite-lattice algebra classification is Casini, Huerta, and Rosabal 2014, §§ 2–4, and the gauge-invariant distillation rate is Van Acoleyen et al. 2016, Eq. (7) and the direct/converse proof on pp. 3–4.

Fermion-parity grading changes tensor-product conventions; negativity calculations additionally require a specified fermionic partial-transpose or partial-time-reversal convention Shapourian, Shiozaki, and Ryu 2017, § II. Generalized symmetries instead use topological defects acting on extended operators rather than ordinary pointlike charge generators Gaiotto et al. 2015, § 2.

Comparison table

The table linearizes the chapter’s shared distinctions. “Allowed operations” means allowed for the named task, not dynamically costless, and “reference” means a physical system or side information whose preparation and reuse are part of the protocol.

Symmetry and gauge information frameworks compared by algebra, center or sectors, operations, reference and edge resources, resolved observable, accessible content, and recovery boundary.

Framework	Algebra or center	Sectors	Allowed operations	Reference resource	Edge extension	Resolved observable	Accessible or distillable content	Recovery or validity boundary
Global charge superselection	Invariant local algebra	Local charge q with weights pq	Locally charge-preserving instruments and classical communication	No phase reference unless budgeted	None	Charge projectors and full-counting statistics	Within-sector average for the standard pure-state task	Collective activation or a supplied reference changes the task
Fermion parity grading	ℤ₂-graded CAR field algebra; even observable subalgebra	Even and odd parity blocks	Parity-preserving physical maps and ancillas	No parity-violating reservoir in the declared theory	None	Restricted covariance spectrum and a specified fermionic partial transpose or partial time reversal	Ordering-independent results after the algebra map	A naive qubit cut can represent a different subsystem
Finite reference and asymmetry	Selected algebra with a group action	Coherence modes between representation sectors	Covariant joint operations on target and reference	Finite token state, localization, and reuse tracked	None	Twirl and relative entropy of asymmetry	Relational coherence for the stated protocol	A perfect or reset reference overstates performance
Charge-resolved entanglement	Symmetric reduced algebra and regional charge	Probabilities and conditional states for each charge	Charge resolution with a declared Fourier grid	Distinct from a phase reference	None	Charged and projected moments	Conditional Rényi or von Neumann entropy	Normalization, continuation, tails, and aliasing must be controlled
Local symmetry restoration	Fixed subsystem algebra and twirl	Off-diagonal charge blocks	Symmetric dynamics with a fixed order of limits	Initial asymmetry and any frame stated	None	Subsystem asymmetry, commutator, and charged observables	Local restoration diagnostic	Recurrences and conserved modes forbid a universal decay claim
Gauge subregion and edge data	Gauge-invariant regional algebra with a named center	Boundary irreducible representations or fluxes	Gauge-invariant local operations	Physical boundary charges counted if supplied	Electric, magnetic, or extended-space prescription named	Flux projectors and matched bulk-plus-edge determinants	Conditional multiplicity-space entanglement; other terms are task dependent	Gauss law, edge measure, zero modes, and regulator must match
Soft infrared information	Dressed or inclusive measured algebra	Soft-charge or radiation labels at finite resolution	Finite-time detector operations	Dressing and boundary data declared	None unless a physical boundary system is introduced	Soft-cloud overlaps and inclusive distinguishability	Only distinctions resolved by the detector	The dressing, bandwidth, regulator, and order of limits matter
Generalized symmetry	Extended-operator and defect algebra	Charge or fusion labels only when defined	Operations compatible with the declared defect data	Inserted topological defect is part of the observable	Defect support and contact prescription stated	Twisted moments and linked-region correlations	A geometry- and task-dependent diagnostic	Identity normalization, deformation, and explicit-breaking controls are required
Covariant channel and recovery	Input and output algebras with group actions	Sector populations, coherences, and mode errors	Covariant decoder plus any locality constraint	Finite reference and correlated side information counted	Gauge center respected when present	Entanglement fidelity and relative-entropy loss	Performance achievable under the restriction	Decoder twirling preserves performance only for a symmetric task; erased labels require correlated side information

Three gates before an operational claim

A formal entropy, overlap, or recovery fidelity becomes an operational statement only after three independent gates pass. The second diagram is the chapter’s stop rule.

Validity and failure map. Passing all three gates licenses a resource, channel, or diagnostic statement only for the declared task. Changing the subsystem prescription, supplying an undeclared phase or edge ancilla, or leaving ultraviolet, infrared, or sector data unresolved changes the question rather than merely its numerical answer. Schematic and not to scale.

The gates are logically independent. A perfectly normalized charge distribution does not repair an unphysical tensor product; a valid gauge algebra does not make an undeclared edge reference free; and a covariant channel does not become local merely because the group average exists. Independent controls should therefore test sector normalization, the even-algebra map, reference degradation, regulator refinement, and detector resolution separately.

The unrestricted recovery theorem also has a precise scope. For a reference state $\sigma$, a channel $\mathcal N$, and states supported in $\sigma$, a universal averaged rotated-Petz map obeys the one-map fidelity bound in Junge et al. 2018, Remark 2.2, Eqs. (20)–(21). Symmetry covariance, gauge compatibility, locality, and reference cost are additional properties that must be proved for the QFT task; the theorem does not supply them automatically.

Check your route

1. A pure fixed-charge state has $p_0=p_1=1/2$ and one-dimensional conditional states. Find its reduced entropy and its accessible entanglement for the standard one-copy local charge-superselection task.

Solution

The direct-sum entropy is $S=H_2(1/2)=1$ bit because both conditional entropies vanish. The accessible entanglement is $\sum_qp_qS(\rho_{A,q})=0$. The missing bit is charge-label uncertainty for this task; collective copies or a supplied reference define a different task.

2. A state $\rho_A$ already commutes with $Q_A$. What are $\mathcal G_A(\rho_A)$ and $A_{U(1)}(\rho_A)$?

Solution

The twirl leaves the state unchanged: $\mathcal G_A(\rho_A)=\rho_A$. Therefore $A_{U(1)}=D(\rho_A\Vert\rho_A)=0$. Sector uncertainty can remain nonzero, showing explicitly that asymmetry and the Shannon entropy of sector weights are different quantities.

3. A decoder reconstructs charge coherence only when an undeclared phase-locked ancilla is attached. Which validity gate fails, and what is the strongest surviving statement?

Solution

The operation-and-reference gate fails. The surviving statement is conditional: the protocol is reference-assisted, and its performance must be reported together with the ancilla state, localization, correlations, and degradation. It is not a no-reference covariant recovery protocol.

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.
• Casini, Horacio, Marina Huerta, and José Alejandro Rosabal. “Remarks on Entanglement Entropy for Gauge Fields.” Physical Review D 89 (2014): 085012. DOI.
• Gaiotto, Davide, Anton Kapustin, Nathan Seiberg, and Brian Willett. “Generalized Global Symmetries.” Journal of High Energy Physics 2015, no. 2 (2015): 172. DOI.
• Goldstein, Moshe, and Eran Sela. “Symmetry-Resolved Entanglement in Many-Body Systems.” Physical Review Letters 120 (2018): 200602. 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.
• Shapourian, Hassan, Ken Shiozaki, and Shinsei Ryu. “Partial Time-Reversal Transformation and Entanglement Negativity in Fermionic Systems.” Physical Review B 95 (2017): 165101. DOI.
• Van Acoleyen, Karel, Nick Bultinck, Jutho Haegeman, Michael Mariën, Volkher B. Scholz, and Frank Verstraete. “The Entanglement of Distillation for Gauge Theories.” Physical Review Letters 117 (2016): 131602. DOI.
• Wiseman, Howard M., and John A. Vaccaro. “Entanglement of Indistinguishable Particles Shared between Two Parties.” Physical Review Letters 91 (2003): 097902. DOI.