SW‑SSB: Strong–Weak Spontaneous Symmetry Breaking

• SW‑SSB is a mechanism in mixed states where a strong symmetry reduces to a weak symmetry, creating steady states with preserved diagonal actions.
• It is diagnosed by nonlinear correlators such as fidelity and Rényi‑1/2 measurements, which capture long‐range order invisible to linear observables.
• SW‑SSB has practical implications in open quantum systems, linking gapless diffusion, topological phases, and experimental signatures in decohered states.

Strong–weak spontaneous symmetry breaking (SW‑SSB) is a mixed‑state and open‑system symmetry‑breaking pattern in which a symmetry that is realized strongly—either as separate left/right actions in Liouville space or as a fixed symmetry sector of the density matrix—is spontaneously reduced to a weaker implementation that survives only as conjugation or as a diagonal subgroup in a doubled description. In contrast to conventional spontaneous symmetry breaking, SW‑SSB is generally invisible to observables linear in the density matrix and is instead diagnosed by nonlinear correlators such as fidelity, Rényi‑1, Rényi‑2, or equivalent doubled‑space correlators. Recent work has developed SW‑SSB in Markovian open quantum systems, in decohered mixed states, in hydrodynamic effective field theory, and in topological and higher‑form settings, while also clarifying its experimental signatures and computational limitations (Lessa et al., 2024, Gu et al., 2024).

1. Strong and weak symmetry in Liouville space and mixed states

For Markovian open quantum systems, the dynamics is generated by a Lindblad master equation

$$\dot{\rho}=\mathcal{L}[\rho] =-i[H,\rho]+\sum_\mu \gamma_\mu\Big(2L_\mu \rho L_\mu^\dagger-\{L_\mu^\dagger L_\mu,\rho\}\Big).$$

In this setting, a symmetry group $G$ can act on the density matrix in two inequivalent ways. The weak symmetry superoperator is

$$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$

and the Liouvillian has weak symmetry if

$$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$

For a $U(1)$ generated by $N$, weak symmetry means covariance under $\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$, but $\langle N\rangle$ need not be conserved. By contrast, strong symmetry is defined by left or right action alone,

$$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$

with

$$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$

For strong $G$0, this is equivalent to

$G$1

so the physical charge is exactly conserved (Gu et al., 2024).

The distinction is most transparent in doubled Hilbert space. Writing

$G$2

one defines

$G$3

Weak $G$4 symmetry corresponds to conservation of $G$5, whereas strong $G$6 implies separate conservation of $G$7 and $G$8, i.e. a full $G$9 structure generated by $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$0. In closed systems there is no analogous strong/weak distinction because there is only single‑space dynamics (Gu et al., 2024).

A complementary mixed‑state formulation starts from the density matrix itself. Strong symmetry means that every pure component of an ensemble decomposition carries the same symmetry charge; weak symmetry means only

$$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$1

This formulation underlies the fidelity‑ and purification‑based literature, where the canonical purification $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$2 carries a doubled symmetry $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$3, and weak symmetry is identified with the diagonal subgroup. This suggests that SW‑SSB is, structurally, a reduction from separate left/right symmetry actions to a diagonal action, although the precise nomenclature depends on the framework being used (Lessa et al., 2024, Weinstein, 2024).

2. The strong-to-weak mechanism

For strong $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$4 Liouvillians, SW‑SSB arises from the existence of steady states in every conserved charge sector. If an initial pure state with definite charge $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$5 evolves under a strongly symmetric Liouvillian, it converges to a steady state $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$6 within that same sector, so there exist steady states $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$7 in each doubled‑space sector with $$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$8. Any linear combination

$$\mathcal{U}_{\mathrm{w}(g)}[\rho]\equiv U(g)\rho U^\dagger(g),$$9

is also a steady state. Under weak symmetry,

$$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$0

but under strong left action,

$$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$1

which differs from $$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$2 unless only one coefficient is nonzero. In this sense, strong symmetry is spontaneously broken while weak symmetry remains intact. The paper states that strong symmetry always spontaneously breaks into the corresponding weak symmetry, for any values of parameters and in any dimension, provided the Liouvillian has strong $$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$3 symmetry and there exist at least two charge sectors with stationary states (Gu et al., 2024).

In mixed‑state language, the same phenomenon is encoded by nonlinear distinguishability between $$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$4 and a symmetry‑twisted state. For a charged operator $$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$5, the fidelity correlator is

$$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$6

and SW‑SSB is defined by long‑range order in this fidelity correlator while ordinary charged correlators decay. This definition is robust against symmetric low‑depth local quantum channels, and the symmetry breaking is “spontaneous” in the sense that the effect of a local symmetry‑breaking measurement cannot be recovered locally (Lessa et al., 2024).

The canonical purification sharpens the analogy to ordinary symmetry breaking. If $$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$7 can be prepared, then SW‑SSB of the mixed state is equivalent to ordinary spontaneous symmetry breaking of a doubled symmetry in the purification. This makes the mixed‑state order pattern an ordinary two‑point long‑range order problem in a larger Hilbert space (Weinstein, 2024).

3. Nonlinear diagnostics and order parameters

The diagnostic distinction between ordinary SSB and SW‑SSB is that ordinary linear correlators probe weak symmetry breaking, whereas SW‑SSB requires nonlinear functionals of $$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$8. The basic objects used across the literature are as follows.

Diagnostic	Expression	Role
Ordinary correlator	$$[\mathcal{L},\mathcal{U}_{\mathrm{w}(g)}]=0 \quad \forall g\in G.$$9	Conventional weak‑symmetry SSB
Rényi‑1 correlator	$U(1)$0	SW‑SSB diagnostic
Fidelity correlator	$U(1)$1	Equivalent SW‑SSB diagnostic
Rényi‑2 correlator	$U(1)$2	Useful but not universally stable

Here $U(1)$3. The Rényi‑1 correlator

$U(1)$4

is the “just‑as‑good” fidelity or quantum affinity, and in canonical purification it becomes an ordinary two‑point correlator. The paper on efficient detection via canonical purification emphasizes that if the purification can be reliably prepared, then SW‑SSB can be detected via ordinary two‑point functions in the purified state (Weinstein, 2024).

The Wightman correlator

$U(1)$5

provides an alternative formulation. It is proven to be equivalent to the fidelity correlator for defining SW‑SSB, with inequalities relating the two so that nonzero long‑distance Wightman order is equivalent to nonzero long‑distance fidelity order. The thermofield‑double construction makes this correlator a standard pure‑state symmetry‑breaking probe in doubled space (Liu et al., 2024).

In fermionic dephasing problems, a broad class of Rényi‑2 correlators of bilinears

$U(1)$6

is upper bounded by the Rényi‑2 correlator of

$U(1)$7

which is an interlayer Cooper‑pair operator before the particle–hole transformation on the right copy. The resulting inequality

$U(1)$8

holds for arbitrary decoherence strength, even flavor number, and general tight‑binding Hamiltonians, and identifies the interlayer pairing channel as a proximate diagnostic of fermionic $U(1)$9 SW‑SSB (Sarma et al., 25 Mar 2026).

Two practical developments refine this picture. First, local “marginal fidelity” correlators of radius $N$0 require tomography only on $N$1-size regions, and for symmetry‑projected Gibbs‑like states satisfying suitable local indistinguishability assumptions, their error relative to the global fidelity correlator decays exponentially in $N$2 (Zhang, 27 May 2026). Second, state‑agnostic efficient detection is impossible in general: pseudorandom strongly symmetric mixed states can be computationally indistinguishable from genuine SW‑SSB states for both $N$3 and $N$4, ruling out efficient black‑box protocols without additional structure (Feng et al., 16 Apr 2025).

4. Spectral, hydrodynamic, and fluctuation consequences

In translationally invariant strongly symmetric Liouvillians, SW‑SSB has direct spectral consequences. The extra $N$5 associated with $N$6 is broken, and the corresponding Goldstone mode in Liouville space appears as a diffusive hydrodynamic mode of the conserved charge. In such cases the Liouvillian spectrum contains gapless modes

$N$7

or equivalently

$N$8

so SW‑SSB guarantees a gapless diffusive Liouvillian even when there is no conventional order parameter expectation value. This leads to the “enhanced Lieb–Schultz–Mattis theorem” for open systems: strong continuous symmetry plus translational invariance enforces gaplessness for any filling, except trivial empty or full limits (Gu et al., 2024).

When weak symmetry is also broken, the full $N$9 can be completely broken, yielding two Goldstone sectors. In the strong‑symmetric model near half filling, the effective Keldysh action decomposes into an order‑parameter Goldstone mode,

$\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$0

and a hydrodynamic diffusive sector,

$\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$1

The first describes phase fluctuations of the broken weak $\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$2; the second is the diffusive Goldstone sector associated with SW‑SSB (Gu et al., 2024).

A more general effective‑field‑theory perspective identifies hydrodynamics itself as the EFT of SW‑SSB. In this formulation, the diffusive mode is the Goldstone boson of a broken strong $\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$3, the static susceptibility is the order parameter, and the unusual Schwinger–Keldysh reparameterization symmetry arises because the strong $\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$4 is broken while the weak $\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$5 remains unbroken. The leading effective Hamiltonian density takes the form

$\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$6

and reduces to the standard diffusion EFT after integrating out the appropriate variables (Huang et al., 2024).

Static fluctuation consequences are subtler. For continuous symmetries, long‑range Rényi‑1 order together with a sufficiently rapid approach to its nonzero asymptotic value forces extensive block‑charge variance, equivalently extensive curvature of the truncated symmetry expectation. The implication is conditional and non‑reversible: dephased superfluids can retain Rényi‑1 SW‑SSB with subextensive charge variance when the Rényi‑1 tail is too slow, while sparse fixed‑charge projectors can have extensive charge variance without local charge‑transfer Rényi‑1 order. The same work introduces a twist‑overlap correlator that decomposes local block‑charge fluctuations into strong‑ and weak‑symmetry channels, with the weak channel directly related to the Wigner–Yanase skew information (Lee, 6 May 2026).

5. Model realizations and phase structure

Concrete Liouvillian models clarify the generic scenarios. One weak‑symmetric but not strong‑symmetric model is a spin‑$\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$7 XXZ system with sublattice gain and loss,

$\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$8

which has weak $\rho\mapsto e^{-iN\theta}\rho e^{iN\theta}$9 symmetry and undergoes a weak $\langle N\rangle$0 SSB transition as $\langle N\rangle$1 increases. A strong‑symmetric model with jump operators $\langle N\rangle$2 and dephasing $\langle N\rangle$3 preserves

$\langle N\rangle$4

exactly and exhibits a symmetric phase, a low‑filling Bose surface phase, and a high‑filling weak‑SSB phase in which the strong symmetry is completely broken and two Goldstone modes appear. A third model, the XXZ chain with dephasing,

$\langle N\rangle$5

has strong $\langle N\rangle$6, no weak symmetry breaking, and nevertheless displays SW‑SSB through gapless diffusive modes. In the strong‑symmetric model near half filling, mean field also reveals a filling‑driven transition from a symmetric Bose surface phase to a symmetry‑broken phase, and long‑range $\langle N\rangle$7 order is possible for $\langle N\rangle$8 but not in lower dimensions (Gu et al., 2024).

A distinct one‑dimensional realization arises in the critical XXZ chain under local strong‑symmetry‑preserving decoherence. For two‑site XX decoherence, the mixed state exhibits a trivial Luttinger liquid phase for $\langle N\rangle$9 and a SW‑SSB phase for $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$0, separated by a straight critical line at $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$1 for all $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$2. Field theory maps the problem to a boundary sine‑Gordon or boundary Kondo setting, and the transition is in the boundary Berezinskii–Kosterlitz–Thouless universality class with correlation length

$$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$3

while the effective central charge varies continuously with decoherence strength along the critical line (Guo et al., 18 Mar 2025).

In $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$4 dimensions, the transverse‑field Ising model with a strongly $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$5-symmetric bond decoherence channel realizes four mixed‑state phases characterized by three nonlinear correlators $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$6: a strongly symmetric phase, an $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$7-SWSSB phase with $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$8 but $$\mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv U(g)\rho \quad \text{or} \quad \mathcal{U}_{\mathrm{s}(g)}[\rho]\equiv \rho\,U^\dagger(g),$$9, an $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$0-SSB phase with $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$1 but $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$2, and an ordinary SSB phase with all three nonzero. Away from the bulk quantum critical point, the defect field theory reduces to a two‑dimensional Ashkin–Teller model, with Ising critical lines, a merged self‑dual line with continuously varying exponents, and a tricritical point identified with the 4‑state Potts CFT (Ding et al., 25 Mar 2026).

Local Markov and Lindblad dynamics furnish another route. In $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$3 dimensions, an absorbing‑state construction produces a transition between strong‑paramagnetic behavior and SW‑SSB with scaling consistent with the parity‑conserving branching‑annihilating random‑walk universality class. In $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$4 dimensions, a pair‑flip variant of Toom’s rule gives evidence for a stable strong‑paramagnetic/weakly symmetry‑broken regime and a transition into an active SW‑SSB regime (Zhang, 27 May 2026).

6. Extensions, topological variants, experiment, and interpretation

SW‑SSB has been extended beyond ordinary on‑site continuous symmetries. In one dimension with $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$5 symmetry, combining SW‑SSB with average symmetry‑protected topological order yields the “double ASPT” phase. In this phase, a strong $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$6 symmetry undergoes SW‑SSB, while the remaining weak $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$7 together with strong $$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$8 protects ASPT order in each branch. The resulting mixed‑state phase diagram contains trivial, SWSSB, SNSSB, ASPT, and double ASPT phases, with two topologically distinct triple points related by domain‑wall decoration duality (Guo et al., 2024).

A higher‑form and non‑invertible generalization appears in non‑Abelian Kitaev quantum double models under decoherence. There, topological order is interpreted as spontaneous symmetry breaking of higher‑form symmetries, and in the non‑Abelian case the broken higher‑form symmetries are non‑invertible. Under decoherence, strong electric 1‑form symmetries remain strong while magnetic 1‑form symmetries become weak, yielding a strong‑to‑weak symmetry‑breaking structure. The resulting decohered mixed states form a locally indistinguishable information convex set whose dimension equals the ground‑state degeneracy of the corresponding pure state, so quantum information stored in the code space is degraded into classical information encoded in the convex set (Song et al., 29 Sep 2025).

The first experimental observation of SW‑SSB was reported in a dephased single‑component Fermi gas imaged by a quantum gas microscope. After full density dephasing,

$$[\mathcal{L},\mathcal{U}_{\mathrm{s}(g)}]=0 \quad \forall g.$$9

the Rényi‑1 and Rényi‑2 correlators reduce to overlaps of the occupation‑number distribution $G$00 with a displaced distribution $G$01. A machine‑learned Gaussian reference state and quantum–classical estimator give access to these nonlinear correlators and reveal long‑range Rényi order in the dephased Fermi liquid. Adding a commensurate superlattice drives the underlying fermions through a metal‑to‑insulator transition that, after full dephasing, appears as a sharp SW‑SSB phase transition (Wang et al., 17 Apr 2026).

These developments suggest a unifying interpretation. In Liouvillian language, SW‑SSB expresses the structural instability of strong symmetry toward its weak form and ties that reduction to gapless diffusion and Goldstone physics. In mixed‑state language, SW‑SSB identifies phases whose defining order is nonlinear in $G$02, robust under symmetric local channels, and naturally represented in doubled space. The resulting classification extends ordinary Landau symmetry breaking to settings where the physically relevant distinction is not symmetry versus no symmetry, but strong symmetry versus weak symmetry.