Time-dependent Schwinger Boson Mean-Field Theory

• Time-dependent Schwinger boson mean-field theory is a dynamical framework that extends static SBMFT by promoting bond and constraint fields to evolve in real time.
• It employs a quadratic bosonic Hamiltonian and paraunitary transformations to model spinon excitations and compute observables like the dynamical spin structure factor.
• Self-consistent nonequilibrium formulations in this theory enable the study of magnetization switching, supermagnonic propagation, and thermal Hall effects in quantum antiferromagnets.

Time-dependent Schwinger boson mean-field theory is the dynamical extension of Schwinger boson mean-field theory in which spins are represented by bosonic spinons subject to a local constraint, quartic exchange terms are decoupled into quadratic bond fields, and real-time evolution is generated either by a static quadratic spinon Hamiltonian or by a self-consistent Hamiltonian whose bond fields and Lagrange multipliers are themselves time-dependent. In the literature, the framework appears in two closely related forms: a dynamical-response construction used to compute quantities such as the dynamical spin structure factor from a static saddle point, and an explicitly nonequilibrium construction in which $A_{ij}(t)$, $B_{ij}(t)$, and $\mu_i(t)$ evolve self-consistently under external driving. It has been applied to kagome, triangular, honeycomb, and square-lattice antiferromagnets, and to problems including neutron-scattering spectra, thermal Hall transport, supermagnonic propagation, and Néel-vector switching (Halimeh et al., 2018, Mukherjee et al., 2021, Bouman et al., 2023, Johannesmann et al., 8 Jan 2026).

1. Schwinger bosons, bond fields, and gauge structure

The formal starting point is the standard Schwinger-boson representation

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$

with bosonic commutators and the local constraint

$$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$

For spin-$\tfrac12$ systems this becomes one boson per site. In principle the constraint is sitewise, but in many SBMFT implementations it is enforced only on average through a global or sublattice-dependent Lagrange multiplier, a simplification that is technically decisive and physically consequential (Halimeh et al., 2018, Mukherjee et al., 2021).

Nearest-neighbor exchange is rewritten in terms of $SU(2)$-invariant bond operators. In the standard notation,

$$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$

and the Heisenberg interaction takes the form

$$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$

Static mean-field theory introduces complex bond parameters $A_{ij}=\langle \hat A_{ij}\rangle$ and $B_{ij}(t)$0, producing a quadratic bosonic Hamiltonian. This bond-field description is also the natural entry point for the time-dependent theory, because the same $B_{ij}(t)$1, $B_{ij}(t)$2, and constraint fields become the collective variables of the dynamical problem (Mukherjee et al., 2021, Yang et al., 2015).

The representation has a local $B_{ij}(t)$3 gauge redundancy $B_{ij}(t)$4, under which

$B_{ij}(t)$5

Physical observables are gauge invariant. When both pairing and hopping channels are nonzero, the invariant gauge group is typically $B_{ij}(t)$6, so the mean-field states are $B_{ij}(t)$7 spin liquids in Wen’s sense (Mukherjee et al., 2021). On the square lattice, projective symmetry group analysis yields six symmetry-preserving $B_{ij}(t)$8 Schwinger-boson states relevant to the $B_{ij}(t)$9–$\mu_i(t)$0 Heisenberg model; on kagome, fully symmetric $\mu_i(t)$1 and $\mu_i(t)$2 states and chiral Ansätze such as cuboc1 are organized by fluxes and unit-cell structure following Wang and Vishwanath and Messio et al. (Yang et al., 2015, Mondal et al., 2020). This suggests that any time-dependent ansatz that remains within a fixed spin-liquid phase must preserve the corresponding PSG constraints and gauge-invariant flux pattern.

2. Quadratic bosonic dynamics and the meaning of “time dependent”

After Fourier transformation and Nambu doubling, the mean-field Hamiltonian becomes a bosonic Bogoliubov-de Gennes problem. A representative kagome construction uses

$\mu_i(t)$3

so that

$\mu_i(t)$4

A paraunitary transformation $\mu_i(t)$5 satisfying $\mu_i(t)$6 diagonalizes $\mu_i(t)$7, giving bosonic spinon branches $\mu_i(t)$8 and

$\mu_i(t)$9

The Heisenberg-picture evolution is then trivial in the quasiparticle basis,

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$0

In the kagome study of low-temperature spin dynamics, this step is described as precisely what turns SBMFT into a time-dependent mean-field theory: one first solves for static mean fields and then uses the quadratic Hamiltonian to propagate quasiparticle operators in time and build dynamical correlation functions (Halimeh et al., 2018).

In this dynamical-response sense, “time dependent” does not initially mean that the mean fields themselves fluctuate. Rather, it means that the static saddle point defines a quadratic propagator from which one computes real-time or frequency-resolved observables. The same logic underlies dynamical structure-factor calculations for kagome antiferromagnets with Dzyaloshinskii–Moriya interaction, where the spin operators are expressed in terms of Bogoliubov bosons and the spectral weight is written as sums over two-spinon intermediate states with energies $$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$1 (Mondal et al., 2020). A closely related construction is used in the kagome comparison of Abrikosov fermion and Schwinger boson approaches, where the mean-field Hamiltonian is written as

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$2

diagonalized by a paraunitary transformation, and then used to evaluate

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$3

as a two-spinon continuum (Sasamoto et al., 23 Mar 2026).

A broader reading of the literature suggests two operational meanings of time-dependent SBMFT. The first is this dynamical-response construction around a static saddle point. The second, discussed explicitly below, promotes the saddle-point fields themselves to $$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$4-dependent collective coordinates (Mukherjee et al., 2021, Bouman et al., 2023).

3. Self-consistent nonequilibrium formulations

An explicitly nonequilibrium formulation is obtained by promoting the mean-field parameters to time-dependent fields,

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$5

so that

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$6

In momentum space this gives

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$7

The corresponding Heisenberg equation is

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$8

and the time-dependent self-consistency conditions are

$$\hat{\mathbf{S}}_i=\frac{1}{2}\,\hat b^\dagger_{i,\alpha}\,\boldsymbol{\sigma}^{\alpha\beta}\,\hat b_{i,\beta},$$9

The honeycomb $$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$0–$$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$1–$$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$2 study presents these equations as the natural time-dependent generalization of static SBMFT and notes that linearizing them around a static solution yields collective-mode dynamics, including gauge fluctuations and Higgs modes of $$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$3 and $$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$4 (Mukherjee et al., 2021).

A fully worked nonequilibrium realization appears in the time-dependent Schwinger boson mean-field theory of supermagnonic propagation in square and honeycomb antiferromagnets. There the time dependence is induced by an exchange quench, the bond parameter becomes $$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$5, and the Hamiltonian is expanded as

$$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$6

The time-dependent self-consistency equation is

$$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$7

and, because the boson-number operators commute with the full spin Hamiltonian for that class of perturbations, the dynamical constraint implies

$$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$8

The resulting time-dependent self-consistency equations are solved analytically in Laplace space, with the entire bond-field dynamics encoded in a single scalar function $$\hat b^\dagger_{i,\uparrow}\hat b_{i,\uparrow}+\hat b^\dagger_{i,\downarrow}\hat b_{i,\downarrow}=2S.$$9 (Bouman et al., 2023).

A second worked nonequilibrium implementation is the switching problem in anisotropic square-lattice antiferromagnets. There the static mean-field ground state, prepared with a small auxiliary field $\tfrac12$0, is used as the initial condition; at $\tfrac12$1 one turns off $\tfrac12$2, switches on a constant uniform field $\tfrac12$3, and evolves a quadratic bosonic Hamiltonian with time-dependent $\tfrac12$4, $\tfrac12$5, and $\tfrac12$6. The sublattice magnetization in the rotated frame is

$\tfrac12$7

and the constraint

$\tfrac12$8

determines $\tfrac12$9 at each time step (Johannesmann et al., 8 Jan 2026).

4. Dynamical observables: spin structure factor, thermal response, and finite-temperature effects

The central observable of dynamical SBMFT is the dynamical spin structure factor,

$SU(2)$0

which is directly probed by inelastic neutron scattering (Halimeh et al., 2018). Because the spin operator is bilinear in bosons, the mean-field response is a two-spinon convolution with coherence factors built from the Bogoliubov matrices. In the kagome $SU(2)$1 spin liquid, the full expression contains four classes of processes: neutron creation of two spinons, absorption from two thermally excited spinons, and two difference-frequency scattering channels off thermally occupied spinons (Halimeh et al., 2018). At $SU(2)$2, only the two-spinon creation term survives; at finite $SU(2)$3, the additional terms produce subgap intensity.

One of the sharpest finite-temperature predictions of the framework is the rapid filling of a nominal spin gap in a gapped kagome $SU(2)$4 spin liquid. For the spin-$SU(2)$5 antiferromagnetic Heisenberg kagome model, the spin gap is $SU(2)$6 while $SU(2)$7 is the spinon gap. The dynamical spin structure factor develops robust low-energy spectral weight already by a temperature of $SU(2)$8, before any appreciable rise in spinon density or change in zero-temperature mean-field parameters. The reason is that the difference-frequency terms are suppressed only by $SU(2)$9, not by $$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$0; the paper identifies this as a consequence of spinon deconfinement (Halimeh et al., 2018). The same work proposes that, if impurity-free spectral functions can be obtained through neutron scattering on kagome herbertsmithites, the temperature dependence of the subgap weight can provide distinct signatures of a $$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$1 quantum spin liquid (Halimeh et al., 2018).

The same quadratic formalism also supports transport calculations. In the honeycomb $$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$2–$$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$3–$$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$4 model, the scalar spin-chirality term breaks time reversal, produces nontrivial Chern numbers of the bosonic excitation bands, and leads to a thermal Hall coefficient computed following Shindou et al. The thermal Hall conductivity is

$$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$5

Within that mean-field theory, $$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$6 in time-reversal-symmetric gapped spin-liquid and valence-bond-crystal phases, while it becomes appreciable in chiral $$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$7 spin liquids and peaks at temperatures comparable to the spinon gap of the lower band (Mukherjee et al., 2021).

Kagome calculations with out-of-plane Dzyaloshinskii–Moriya interaction show how symmetry breaking reshapes the dynamical response already at mean-field level. In spin-liquid phases, $$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$8 is a broad two-spinon continuum; in the $$\hat A_{ij}=\frac12\big(b_{i\uparrow}b_{j\downarrow}-b_{i\downarrow}b_{j\uparrow}\big),\qquad \hat B_{ij}=\frac12\big(b^\dagger_{i\uparrow}b_{j\uparrow}+b^\dagger_{i\downarrow}b_{j\downarrow}\big),$$9 ordered phase, $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$0 shows sharp magnon branches inside the continuum, whereas $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$1 is strongly suppressed because the DM term reduces $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$2 to $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$3 and favors an $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$4-plane spin texture (Mondal et al., 2020).

The recent kagome comparison between Abrikosov fermions and Schwinger bosons adds a sharper contrast between parton statistics. In that work, Abrikosov-fermion mean-field spectra exhibit dome-shaped features, while Schwinger-boson mean-field spectra show a concave-down low-energy structure. Incorporating many-body effects beyond mean field substantially reduces the low-energy gap and enhances the low-energy spectral weight, consistent with experimental observations on herbertsmithite-type systems (Sasamoto et al., 23 Mar 2026).

5. Ordered phases, collective modes, and beyond-mean-field corrections

A central issue in time-dependent SBMFT is the status of collective excitations in magnetically ordered phases. A Gaussian-fluctuation analysis of the triangular-lattice Heisenberg antiferromagnet by Ghioldi et al. treats the saddle point as the leading term of an $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$5 expansion and keeps the $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$6 corrections generated by fluctuations of the Hubbard–Stratonovich fields and the constraint. The fluctuation propagator

$$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$7

produces poles in the dynamical susceptibility, and those poles are interpreted as magnons: collective modes, or two-spinon bound states, of a spinon condensate in the Higgs phase. In that treatment the spectrum contains sharp low-energy magnon branches together with a high-energy spinon continuum extending up to about three times the single-magnon bandwidth, and the ordered moment $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$8 agrees very well with numerical results (Ghioldi et al., 2018).

A contrary assessment appears in a later analysis of the triangular lattice, which argues that SBMFT fails even qualitatively for spin dynamics in the ordered phase. The stated failures are that SBMFT fails to predict the Goldstone mode, that the coherent peak in the two-spinon continuum in the presence of spinon condensate should not be interpreted as a magnon mode, and that SBMFT incorrectly predicts a gapless longitudinal spin fluctuation mode. The paper attributes these problems to three structural issues: spinon condensation fails to provide a consistent description of the order-parameter manifold of the $$\vec S_i\cdot \vec S_j=:\hat B_{ij}^\dagger \hat B_{ij}:-\hat A_{ij}^\dagger\hat A_{ij}.$$9 phase, the formalism lacks the coupling between the uncondensed spinon and the spinon condensate that breaks spin-rotational and translational symmetry, and it lacks the rigidity related to the no-double-occupancy constraint on the spinon system (Zhang et al., 2021). This disagreement marks one of the principal controversies surrounding dynamical SBMFT in ordered phases.

The contrast between these two triangular-lattice analyses clarifies the role of “beyond mean field.” In the first, Gaussian fluctuations of bond and constraint fields are essential because they generate collective poles absent in the bare two-spinon bubble. In the second, even a generalized SBMFT with an explicit magnetic order parameter does not fully cure the dynamical problems because local-constraint rigidity is fundamentally missing at the quadratic level (Ghioldi et al., 2018, Zhang et al., 2021). A plausible implication is that the physical content of time-dependent SBMFT depends strongly on whether the calculation is interpreted as a bare parton continuum, as a self-consistent collective theory, or as the starting point of a systematic fluctuation expansion.

The same point reappears in kagome systems. The 2026 kagome comparison performs an RPA resummation around the Schwinger-boson background,

$A_{ij}=\langle \hat A_{ij}\rangle$0

and finds that the RPA corrections strongly renormalize the low-energy sector while leaving the high-energy continuum comparatively intact (Sasamoto et al., 23 Mar 2026). In this sense, simple time-dependent SBMFT is not the endpoint of the theory but the quadratic kernel from which collective physics is built.

6. Applications, validity regime, and limitations

The framework has developed along several distinct application lines. In kagome antiferromagnets it has been used to calculate the temperature dependence of the dynamical spin structure factor in gapped $A_{ij}=\langle \hat A_{ij}\rangle$1 spin liquids, to study the effect of Dzyaloshinskii–Moriya interaction on static and dynamic spin structure factors, and to compare bosonic and fermionic parton descriptions beyond mean field (Halimeh et al., 2018, Mondal et al., 2020, Sasamoto et al., 23 Mar 2026). In honeycomb systems it has been used to track the transition from a gapped $A_{ij}=\langle \hat A_{ij}\rangle$2 spin liquid to a chiral $A_{ij}=\langle \hat A_{ij}\rangle$3 spin liquid with nontrivial Chern bands and a large thermal Hall coefficient (Mukherjee et al., 2021). On the square lattice, PSG analysis supplies the symmetry and gauge-theoretic foundation for candidate $A_{ij}=\langle \hat A_{ij}\rangle$4 spin liquids relevant to the $A_{ij}=\langle \hat A_{ij}\rangle$5–$A_{ij}=\langle \hat A_{ij}\rangle$6 model, even though time dependence is not treated explicitly there (Yang et al., 2015).

Two modern nonequilibrium applications demonstrate the range of explicitly time-dependent SBMFT. In the supermagnonic problem, the theory shows that the correlation front after an exchange quench is governed by the competition between propagating magnons and a localized quasi-bound state; the supermagnonic effect is determined by that competition and is tunable by lattice coordination and quantum spin value $A_{ij}=\langle \hat A_{ij}\rangle$7 (Bouman et al., 2023). In the switching problem, exact diagonalization on small clusters and time-dependent SBMFT give consistent short-time magnetization dynamics, with only about $A_{ij}=\langle \hat A_{ij}\rangle$8 deviations in the reported comparison, supporting the use of t-SB-MFT as a framework for sublattice-magnetization switching in quantum antiferromagnets (Johannesmann et al., 8 Jan 2026).

The validity regime is nonetheless restricted. In the kagome finite-temperature calculation, the low-temperature window $A_{ij}=\langle \hat A_{ij}\rangle$9 is identified as the regime where the spinon density is small and SBMFT is qualitatively reliable; for $B_{ij}(t)$00, the spinon density becomes sizable, spinon–spinon interactions and gauge fluctuations can no longer be neglected, and the mean-field picture breaks down. At still higher temperature, the self-consistent bond parameters collapse in an unphysical first-order-like way, which is explicitly identified as an artifact of the saddle-point treatment (Halimeh et al., 2018). The honeycomb thermal Hall study lists the same core approximations: mean-field factorization, the constraint treated on average, restriction to uniform or periodic ansätze, neglect of gauge fluctuations, and finite-size effects; for a time-dependent extension based only on $B_{ij}(t)$01, $B_{ij}(t)$02, and $B_{ij}(t)$03, non-Gaussian gauge configurations and tunneling between gauge sectors are missed, and linear-response treatments are expected to be reliable only for slow perturbations and energies below the spinon gap (Mukherjee et al., 2021).

These limitations are not merely technical. They are tied to the defining simplifications of the method: deconfined bosonic spinons in a quadratic background, average rather than exact constraint enforcement, and the omission of vison dynamics and gauge-field fluctuations unless they are restored by a separate approximation. The kagome finite-temperature study states explicitly that vison–spinon interactions, which can be included at a more sophisticated level as in Punk et al., are absent there (Halimeh et al., 2018). Time-dependent SBMFT is therefore best understood as a hierarchy of approximations rather than a single fixed scheme: a quadratic dynamical theory at the base, a self-consistent collective theory when mean fields are evolved in time, and a platform for Gaussian, RPA, or other beyond-mean-field corrections when magnons, gap renormalization, or stronger many-body effects are required.