Skip to content
Quantum Mechanics

Emergent Gauge Fields

An emergent gauge field is a collective low-energy degree of freedom whose gauge structure arises from the constrained Hilbert space and dynamics of a many-body system. It is not inserted as a fundamental interaction. Its electric charges, flux defects, photons, and confinement scales are encoded in gauge-invariant properties of the original spins, electrons, atoms, or qubits.

The phrase contains two logically separate claims:

  1. A useful low-energy description has a local redundancy. Different auxiliary-field configurations label the same physical state.
  2. Fluctuations associated with that redundancy have dynamics and survive over a controlled energy and length window. Depending on dimension and matter content, they may form a deconfined Coulomb phase, a gapped discrete gauge phase, a critical state, or a confined regime.

A parton decomposition establishes the first claim, not the second. Conversely, a Berry connection, an applied electromagnetic potential, or a smoothly varying magnetic texture can be called a gauge field in other contexts without constituting an emergent deconfined gauge sector. The decisive questions are what enforces the Gauss law, what carries gauge charge and flux, and what gauge-invariant measurement tests the resulting theory.

Required background. Fractionalization supplies the distinction between an auxiliary parton representation and deconfined excitation sectors. Quantum Spin Liquids supplies the phase taxonomy used below. From Phase Symmetry to Gauge Theory distinguishes global symmetry, local redundancy, and a dynamical gauge field.

From a constraint to Gauss’s law

Section titled “From a constraint to Gauss’s law”

A spin has fewer states than the auxiliary particles often used to describe it. The resulting constraint is the starting point for the gauge theory; it is also what prevents the auxiliary particles from being mistaken for independent microscopic degrees of freedom.

Redundant variables are not extra physical states

Section titled “Redundant variables are not extra physical states”

For a spin-1/21/2 moment, introduce fermionic partons fiαf_{i\alpha} through

Si=12fiα†σαβfiβ,nif=∑αfiα†fiα=1.\mathbf S_i = \frac{1}{2} f_{i\alpha}^{\dagger} \boldsymbol\sigma_{\alpha\beta} f_{i\beta}, \qquad n_i^f = \sum_\alpha f_{i\alpha}^{\dagger}f_{i\alpha} =1.

Every spin operator is unchanged by the site-dependent transformation

fiα⟼eiθifiα.f_{i\alpha}\longmapsto e^{i\theta_i}f_{i\alpha}.

This local U(1) is a subgroup of the full SU(2) redundancy available in the fermionic spin representation. It is a redundancy introduced by enlarging the two-dimensional spin Hilbert space to a fermionic Fock space. The constraint nif=1n_i^f=1 projects back to physical states. A mean-field expectation value such as ⟨fi†fj⟩\langle f_i^\dagger f_j\rangle is gauge dependent and therefore cannot itself be an observable order parameter.

In this Euclidean path-integral representation, set ℏ=kB=1\hbar=k_{\mathrm B}=1 and use the ordinary kinetic term ∑iαfiα†∂τfiα\sum_{i\alpha}f_{i\alpha}^\dagger\partial_\tau f_{i\alpha}. A Lagrange multiplier enforces the constraint:

Scon=∫dτ ∑i−ia0,i(nif−1).S_{\mathrm{con}} = \int d\tau\, \sum_i -i a_{0,i}(n_i^f-1).

A bond decoupling can be parameterized by the coefficient multiplying fi†fjf_i^\dagger f_j in χijfi†fj+h.c.\chi_{ij}f_i^\dagger f_j+\mathrm{h.c.}:

χij∝⟨fj†fi⟩,χij=∣χij∣eiaij.\chi_{ij}\propto\langle f_j^\dagger f_i\rangle, \qquad \chi_{ij}=|\chi_{ij}|e^{ia_{ij}}.

Under the local rephasing, the phases transform as

aij⟼aij+θi−θj,a0,i⟼a0,i+∂τθi.\begin{aligned} a_{ij}&\longmapsto a_{ij}+\theta_i-\theta_j,\\ a_{0,i}&\longmapsto a_{0,i}+\partial_\tau\theta_i. \end{aligned}

The kinetic variation is +i(∂τθi)nif+i(\partial_\tau\theta_i)n_i^f, while the constraint variation is −i(∂τθi)(nif−1)-i(\partial_\tau\theta_i)(n_i^f-1). Their sum is the background-charge boundary term i∂τθii\partial_\tau\theta_i; an allowed integer winding changes the action by 2πi2\pi i times an integer and leaves its Boltzmann weight unchanged. The temporal multiplier and fluctuating bond phase therefore assemble into a lattice gauge connection. Whether these fluctuations are deconfined, massive, or confining is a dynamical question that must be answered after the constraint has been imposed.

Constrained configurations give a more direct route

Section titled “Constrained configurations give a more direct route”

Quantum dimer models and spin ice expose the same structure without beginning from an arbitrary factorization. Orient the links of a lattice and assign an integer or half-integer electric field EℓE_\ell to each link. A local rule on allowed dimers or spins becomes

(∇⋅E)r≡∑ℓ∋rηrℓEℓ=ρr,(\nabla\cdot E)_r \equiv \sum_{\ell\ni r}\eta_{r\ell}E_\ell = \rho_r,

where ηrℓ=+1\eta_{r\ell}=+1 for an outgoing link and −1-1 for an incoming link. The background or defect charge ρr\rho_r is fixed by the microscopic rule. In spin ice, the two-in/two-out condition is a divergence-free law after a conventional background subtraction; a violated tetrahedron carries emergent gauge charge.

The logic is structural:

local constraint⟹Gauss-law sector⟹gauge-compatible dynamics.\text{local constraint} \quad\Longrightarrow\quad \text{Gauss-law sector} \quad\Longrightarrow\quad \text{gauge-compatible dynamics}.

Not every constraint produces a useful long-wavelength gauge field. An extensive constrained manifold and coherent local dynamics are needed. Local loop moves can connect configurations within a sector while leaving distinct global flux sectors disconnected; that separation is part of the physics, not a failure of the description.

Compact U(1) Dynamics

Section titled “Compact U(1) Dynamics”
Section titled “Link variables and the lattice Hamiltonian”

Place a periodic phase aℓ≡aℓ+2πa_\ell\equiv a_\ell+2\pi and an electric field EℓE_\ell on each oriented link. The well-defined rotor operator is Uℓ=eiaℓU_\ell=e^{ia_\ell}, with

[Eℓ,Uℓ′]=δℓℓ′Uℓ.[E_\ell,U_{\ell'}]=\delta_{\ell\ell'}U_\ell.

Thus UℓU_\ell raises EℓE_\ell by one. The usual shorthand [aℓ,Eℓ′]=iδℓℓ′[a_\ell,E_{\ell'}]=i\delta_{\ell\ell'} uses a local angular coordinate; a globally defined periodic angle is not an ordinary Cartesian position operator. A microscopic quantum-link model can instead have a finite link Hilbert space. Its long-distance rotor description is an effective approximation.

The minimal compact U(1) Hamiltonian is

Hgauge=U2∑ℓEℓ2−K∑pcos⁡ ⁣[(∇×a)p],H_{\mathrm{gauge}} = \frac{U}{2}\sum_\ell E_\ell^2 -K\sum_p \cos\!\left[(\nabla\times a)_p\right],

restricted by (∇⋅E)r=ρr(\nabla\cdot E)_r=\rho_r. The first term penalizes electric flux. The second resonates flux around a plaquette pp. Compactness matters: the plaquette flux is defined modulo 2π2\pi, and the theory can admit magnetic monopoles. In two spatial dimensions these appear as spacetime tunneling events; in three they are particle excitations.

The unitary operator generated by the Gauss constraint,

G[θ]=exp⁡ ⁣[i∑rθr((∇⋅E)r−ρr)],G[\theta] = \exp\!\left[ i\sum_r\theta_r \bigl((\nabla\cdot E)_r-\rho_r\bigr) \right],

acts trivially on physical states. Here ρr\rho_r includes any dynamical matter charge and fixed background. With the convention O↦GOG†O\mapsto GOG^\dagger, it shifts arr′a_{rr'} by θr−θr′\theta_r-\theta_{r'} and rephases matter of gauge charge one. Gauge invariance is thus the operator statement that the dynamics preserves the constrained physical subspace.

Worked example: open strings and closed loops

Section titled “Worked example: open strings and closed loops”

Start with Eℓ=0E_\ell=0 on a square lattice, with every reference link pointing right or up. Acting with UℓU_\ell on a right-pointing link adds +1+1 to the divergence at its left endpoint and −1-1 at its right endpoint. Extending the path cancels these changes at each interior vertex. An open path therefore creates two endpoints with opposite Gauss-law charge. In the charge-free physical subspace, the open link operator alone is not allowed: it must be accompanied by endpoint matter operators, or treated as an insertion of external charges.

A counterclockwise plaquette move instead raises the bottom and right links and lowers the top and left links. Every vertex gains one incoming and one outgoing unit, so its divergence is unchanged. The operator

Up=exp⁡ ⁣(i∑ℓ∈∂psℓaℓ),cos⁡Φp=Up+Up†2,U_p=\exp\!\left(i\sum_{\ell\in\partial p}s_\ell a_\ell\right), \qquad \cos\Phi_p=\frac{U_p+U_p^\dagger}{2},

uses sℓ=+1s_\ell=+1 when the path follows the reference orientation and −1-1 otherwise. This is why a plaquette term supplies quantum dynamics without violating Gauss’s law. It does not prove deconfinement: the energy of widely separated endpoints still has to be determined.

An open electric-flux string has opposite endpoint charges, while an oriented closed plaquette changes no vertex divergence

Arrows show the added electric flux, starting from zero. The open string changes the divergence only at its endpoints; the closed loop preserves it everywhere. A leftward or downward added flux lowers the electric field on a rightward or upward reference link. The square lattice illustrates the algebra; spin ice has a different microscopic lattice.

The Coulomb phase and emergent photon

Section titled “The Coulomb phase and emergent photon”

When flux fluctuations remain small, expand the cosine and coarse-grain. In three spatial dimensions the leading continuum Hamiltonian has Maxwell form,

HMaxwell=12∫d3x (ϵ−1E2+μ−1B2),B=∇×a.H_{\mathrm{Maxwell}} = \frac{1}{2} \int d^3x\, \left( \epsilon^{-1}\mathbf E^2 + \mu^{-1}\mathbf B^2 \right), \qquad \mathbf B=\nabla\times\mathbf a.

For an isotropic, reciprocal long-wavelength medium, the two transverse fluctuations then obey

ωq=ce∣q∣+O(q3),ce=1ϵμ.\omega_{\mathbf q}=c_{\mathrm e}|\mathbf q|+\mathcal O(q^3), \qquad c_{\mathrm e}=\frac{1}{\sqrt{\epsilon\mu}}.

This emergent photon is a neutral collective mode of the medium. Its velocity cec_{\mathrm e} is set by microscopic exchange scales and lattice spacing, not by the vacuum speed of light. Gauge charges and magnetic monopoles are separate excitations, generally gapped in a three-dimensional quantum spin ice.

For one linearly dispersing photon with two transverse polarizations in three dimensions, the low-temperature energy density and heat capacity scale as

UγV=π215(kBT)4ℏ3ce3,CγV=4π215kB4T3ℏ3ce3.\frac{U_\gamma}{V} = \frac{\pi^2}{15} \frac{(k_{\mathrm B}T)^4} {\hbar^3c_{\mathrm e}^3}, \qquad \frac{C_\gamma}{V} = \frac{4\pi^2}{15} \frac{k_{\mathrm B}^4T^3} {\hbar^3c_{\mathrm e}^3}.

The T3T^3 law is not distinctive by itself because acoustic phonons have the same power. A credible assignment needs the coefficient, momentum and polarization structure, field evolution, and a phonon baseline.

Compactness, monopoles, and dimension

Section titled “Compactness, monopoles, and dimension”

The noncompact Maxwell theory omits tunneling events between flux sectors. In a compact theory, monopole or instanton operators insert quantized flux. Their infrared relevance depends on spacetime dimension, matter content, and microscopic symmetries.

  • Pure compact U(1) gauge theory in 2+12+1 dimensions confines at long distance through monopole proliferation.
  • Compact U(1) gauge theory can have a stable deconfined Coulomb phase in 3+13+1 dimensions, with a photon and gapped electric and magnetic charges.
  • Gapless matter can suppress monopoles in 2+12+1 dimensions. Large-flavor compact QED3_3 supplies a controlled deconfined limit, but stability at the small physical flavor number is model and symmetry dependent.
  • A lattice distortion or another symmetry-allowed perturbation can couple directly to monopole operators and destabilize a proposed U(1) Dirac spin liquid.

Writing a Maxwell action is therefore not enough. One must specify whether the gauge field is compact and list every symmetry-allowed monopole perturbation.

Discrete and Non-Abelian Gauge Structures

Section titled “Discrete and Non-Abelian Gauge Structures”

From U(1) to ℤ₂

Section titled “From U(1) to ℤ₂”

Let a charge-two field Φ\Phi couple to a U(1) gauge field:

LΦ=∣(∂μ−2iaμ)Φ∣2+V(∣Φ∣)+14e2fμνfμν.\mathcal L_\Phi = \left|\left(\partial_\mu-2ia_\mu\right)\Phi\right|^2 +V(|\Phi|) +\frac{1}{4e^2}f_{\mu\nu}f^{\mu\nu}.

In the Higgs regime where the amplitude of Φ\Phi is nonzero, transformations with e2iθ=1e^{2i\theta}=1 leave the condensate representative unchanged. The surviving low-energy gauge structure is Z2\mathbb Z_2. A fully gapped deconfined liquid in 2+12+1 dimensions has gapped gauge charge, gapped flux called a vison, and no gapless photon. Discrete gauge structure alone does not require the gauge-charged matter to be gapped; the zero-field gapless Kitaev phase is a counterexample. The charge and vison have nontrivial mutual statistics, as developed on Topological Order.

One should not say that a local gauge symmetry spontaneously breaks. Elitzur’s theorem forbids a gauge-noninvariant local expectation value from serving as a physical order parameter. “Higgs regime” refers to a gauge-invariant change in spectrum and long-distance response. With matter in a fundamental representation, nominal confinement and Higgs regions may even be analytically connected; the phase diagram must be established from the actual matter representation and global symmetries.

Loop diagnostics and their caveats

Section titled “Loop diagnostics and their caveats”

For a closed contour CC, the Wilson loop is

W(C)=⟨exp⁡ ⁣(i∑ℓ∈Caℓ)⟩.W(C) = \left\langle \exp\!\left(i\sum_{\ell\in C}a_\ell\right) \right\rangle.

In a pure gauge theory, an area law −ln⁡W(C)∝A(C)-\ln W(C)\propto A(C) indicates a string tension and confinement, while a perimeter law is compatible with deconfinement. Here CC is a large spacetime loop used to extract the static interaction of external charges; an equal-time spatial loop alone does not measure that potential.

These statements need a matter caveat. Dynamical fundamental charges can screen external probes and break a long string, converting an asymptotic Wilson-loop area law into a perimeter-like form even in a confining regime. Finite-size numerics should therefore combine loop behavior with flux-sector splittings, charge-pair energetics, string breaking, and the excitation spectrum.

Non-Abelian cases

Section titled “Non-Abelian cases”

A single spin representation may have a larger microscopic redundancy than its low-energy gauge structure. SU(2) parton formulations of a spin-1/21/2 system can be Higgsed to U(1) or Z2\mathbb Z_2 by a particular ansatz. The physically relevant gauge group is the subgroup that leaves the ansatz invariant, together with the fluctuations that remain dynamical after high-energy modes are integrated out.

Deconfined non-Abelian gauge fields in 2+12+1-dimensional magnets are more delicate than the familiar Z2\mathbb Z_2 case. Pure Yang–Mills dynamics is confining; gapless matter or topological terms can qualitatively change the infrared theory. A Chern–Simons gauge theory can encode non-Abelian anyons without supplying a Maxwell-like propagating photon, so “non-Abelian anyons” and “gapless non-Abelian gauge bosons” are not synonyms.

The Kitaev honeycomb model illustrates another useful distinction. Its exact solution contains a Z2\mathbb Z_2 gauge field. In a time-reversal-breaking gapped regime, the mobile Majorana sector supports non-Abelian Ising anyons. The gauge field itself has not become an SU(2) field merely because the anyon fusion rules are non-Abelian.

Coupling to Fractionalized Matter

Section titled “Coupling to Fractionalized Matter”

Dirac spinons and compact QED₃

Section titled “Dirac spinons and compact QED₃”

A common continuum theory for a two-dimensional Dirac spin liquid is

LQED3=∑a=1Nfψˉaγμ(∂μ−iaμ)ψa+14e2fμνfμν+Lmon.\mathcal L_{\mathrm{QED}_3} = \sum_{a=1}^{N_f} \bar\psi_a\gamma^\mu (\partial_\mu-ia_\mu)\psi_a + \frac{1}{4e^2}f_{\mu\nu}f^{\mu\nu} + \mathcal L_{\mathrm{mon}}.

The fermions ψa\psi_a carry emergent gauge charge but need not carry electrical charge. Physical spin operators are gauge-neutral fermion bilinears or monopole operators. Their lattice momentum, spin, and point-group quantum numbers depend on the parton ansatz and its projective symmetry implementation.

The term Lmon\mathcal L_{\mathrm{mon}} contains monopole insertions allowed by microscopic symmetries. If every allowed monopole is irrelevant, an algebraic spin liquid can persist. If an allowed monopole is relevant, its proliferation can confine spinons and often select Néel or valence-bond order. A continuum scaling analysis must therefore be matched to the microscopic symmetry quantum numbers.

A spinon Fermi surface

Section titled “A spinon Fermi surface”

A finite density of gauge-charged fermions Landau damps the transverse U(1) gauge field. In a two-dimensional patch description with ℏ=kB=1\hbar=k_{\mathrm B}=1, the Matsubara propagator has the form

DT−1(q,iωn)∼χq2+γ∣ωn∣∣q∥∣,D_T^{-1}(\mathbf q,i\omega_n) \sim \chi q^2 + \gamma\frac{|\omega_n|}{|q_\parallel|},

where q∥q_\parallel is tangent to the local Fermi-surface patch. With the Dyson convention G−1=iωn−ξ−ΣG^{-1}=i\omega_n-\xi-\Sigma, the corresponding fermion self-energy has the patch-scaling form

Σ(iωn)∼−iC sgn⁡(ωn)∣ωn∣2/3,C>0.\Sigma(i\omega_n) \sim -i C\,\operatorname{sgn}(\omega_n) |\omega_n|^{2/3}, \qquad C>0.

Because ∣Σ∣/∣ω∣|\Sigma|/|\omega| grows at low frequency, the Landau quasiparticle is not asymptotically sharp. This is a canonical mechanism for a Non-Fermi Liquid, but numerical coefficients, transport exponents, and even controlled extrapolation can be subtle. Momentum relaxation, disorder, gauge-neutral sectors, and possible pairing of spinons must be treated separately.

Gauge charge is not laboratory charge

Section titled “Gauge charge is not laboratory charge”

An emergent gauge charge labels the response to aμa_\mu; electric charge labels the response to the physical electromagnetic field AμA_\mu. A spinon may be electrically neutral and gauge charged, while a chargon may carry both. Only gauge-neutral combinations are locally creatable in the bulk.

This distinction is why response rules for parton sectors are constrained rather than simply additive. For the Ioffe–Larkin composition rule and its assumptions, see Fractionalization.

What can experiments measure?

Section titled “What can experiments measure?”

Neutrons and thermodynamic probes couple to gauge-invariant microscopic operators, not directly to the redundant potential aμa_\mu. The spin-ice calculation of Hermele, Fisher, and Balents provides a microscopic realization of emergent electrodynamics. Benton, Sikora, and Shannon calculate its scattering signatures, including the difference between classical pinch points and the quantum ground state.

Observable Physical interpretation and limitation
Pinch-point angular structure A nearly divergence-free field; classical spin ice can produce it without a quantum photon
A low-energy linear mode A photon candidate only when its polarization, intensity, and microscopic spin coupling agree with the gauge theory
A three-dimensional T3T^3 heat capacity Compatible with a photon, but acoustic phonons give the same power
A broad spin continuum Compatible with fractional matter, but also with ordinary multiparticle states or disorder
String and loop correlations in a simulator Probe endpoint defects and flux structure, provided preparation, boundaries, and violations of the constraint are included

In quantum spin ice, coherent ring exchange can turn an ice-rule manifold into a U(1) quantum liquid. At zero temperature the long-wavelength equal-time field correlations need not retain the classical pinch-point intensity: quantum zero-point fluctuations change their momentum dependence. Finite temperature and finite resolution matter when comparing with neutron data. These are predictions about a specified microscopic operator, not features inferred from a Maxwell equation alone.

The photon, charge threshold, and thermodynamics should be compared using the same exchange parameters. An elastic pinch point and a T3T^3 term, without this connection, remain consistent with classical constraints plus phonons. Structure Factors explains the tensor information and sum rules used in that comparison. Quantum Spin Liquids discusses materials; Ultracold-Atom Quantum Simulation discusses controlled implementations.

A geometric connection need not be a dynamical gauge field. A Berry connection describes how states vary over parameter space. A spin texture may act as an effective magnetic field for electrons. Neither construction by itself supplies an independent Gauss law and a deconfined charge sector.

Exercises

Section titled “Exercises”

1. Derive the gauge transformation of a bond field

Section titled “1. Derive the gauge transformation of a bond field”

Let χij∝⟨fj†fi⟩\chi_{ij}\propto\langle f_j^\dagger f_i\rangle be the coefficient of fi†fjf_i^\dagger f_j, with χij=∣χij∣eiaij\chi_{ij}=|\chi_{ij}|e^{ia_{ij}}. Under fi↦eiθifif_i\mapsto e^{i\theta_i}f_i, find the transformation of aija_{ij} and show that the oriented plaquette flux is invariant.

Solution

The bilinear transforms as

fj†fi⟼e−iθjeiθifj†fi.f_j^\dagger f_i \longmapsto e^{-i\theta_j}e^{i\theta_i} f_j^\dagger f_i.

Thus aij↦aij+θi−θja_{ij}\mapsto a_{ij}+\theta_i-\theta_j, and χijfi†fj\chi_{ij}f_i^\dagger f_j is invariant. Around an oriented plaquette,

Φp=∑(ij)∈∂paij.\Phi_p = \sum_{(ij)\in\partial p}a_{ij}.

Every site phase appears once with each sign, so the shifts telescope and Φp↦Φp\Phi_p\mapsto\Phi_p modulo 2π2\pi. The individual link phase is redundant; the loop flux is gauge invariant.

2. Count the emergent-photon heat capacity

Section titled “2. Count the emergent-photon heat capacity”

Assume an isotropic three-dimensional U(1) Coulomb phase with dispersion ω=ceq\omega=c_{\mathrm e}q and two transverse polarizations. Derive the low-temperature heat-capacity density.

Solution

The energy density is

UV=2∫d3q(2π)3ℏceqeβℏceq−1.\frac{U}{V} = 2\int\frac{d^3q}{(2\pi)^3} \frac{\hbar c_{\mathrm e}q} {e^{\beta\hbar c_{\mathrm e}q}-1}.

Using x=βℏceqx=\beta\hbar c_{\mathrm e}q and ∫0∞dx x3/(ex−1)=π4/15\int_0^\infty dx\,x^3/(e^x-1)=\pi^4/15 gives

UV=π215(kBT)4ℏ3ce3.\frac{U}{V} = \frac{\pi^2}{15} \frac{(k_{\mathrm B}T)^4} {\hbar^3c_{\mathrm e}^3}.

Differentiation yields

CV=4π215kB4T3ℏ3ce3.\frac{C}{V} = \frac{4\pi^2}{15} \frac{k_{\mathrm B}^4T^3} {\hbar^3c_{\mathrm e}^3}.

The same power as an acoustic phonon makes the coefficient and independent spectroscopic checks essential.

3. Higgs U(1) to ℤ₂

Section titled “3. Higgs U(1) to ℤ₂”

A field of gauge charge qq has nonzero amplitude in a Higgs regime. Which transformations leave its representative unchanged? Specialize to q=2q=2 and identify the residual group.

Solution

The field transforms as Φ↦eiqθΦ\Phi\mapsto e^{iq\theta}\Phi. A fixed nonzero representative is unchanged when

qθ=2πn,n∈Z.q\theta=2\pi n, \qquad n\in\mathbb Z.

Thus the residual transformations are θ=2πn/q\theta=2\pi n/q, forming Zq\mathbb Z_q. For q=2q=2, the two elements are θ=0\theta=0 and π\pi, so the low-energy gauge structure is Z2\mathbb Z_2. This identifies the residual redundancy. Whether it describes a deconfined discrete gauge phase still depends on the dynamics; physical phase distinctions must still be stated through gauge-invariant quantities.

4. Why a Wilson loop can lose its area law

Section titled “4. Why a Wilson loop can lose its area law”

Suppose a confining string has tension σ\sigma but the theory contains a dynamical charge-anticharge pair with gap Δ\Delta per particle. Estimate the separation at which string breaking becomes favorable and explain the Wilson-loop caveat.

Solution

An unbroken string of length RR costs approximately σR\sigma R. Screening the external sources by creating a dynamical pair costs about 2Δ2\Delta. String breaking is favorable when

R≳Rbreak=2Δσ.R\gtrsim R_{\mathrm{break}} = \frac{2\Delta}{\sigma}.

Beyond this scale the static potential saturates even though isolated dynamical gauge charge need not be deconfined. Consequently an asymptotic Wilson loop may look perimeter-like in the presence of fundamental matter. Flux-sector spectroscopy and charge-pair energetics are needed in addition.

5. Distinguish a pinch point from a photon

Section titled “5. Distinguish a pinch point from a photon”

An elastic neutron experiment observes pinch points, and calorimetry finds C∝T3C\propto T^3. List a minimum follow-up program before assigning an emergent photon.

Solution

The pinch points should be fit to the tensor structure and correlation length of the proposed Gauss law, with instrumental resolution included. Inelastic scattering should then resolve a low-energy dispersing mode with the predicted polarization and velocity. The heat-capacity coefficient must be compared with independently measured acoustic-phonon velocities and nuclear terms. Finally, temperature and field should alter the elastic constraint, photon-like mode, and gauge-charge continuum according to one calibrated Hamiltonian. Classical spin ice already explains pinch points, and phonons already explain a T3T^3 term, so the conjunction without those cross-checks is not decisive.

6. Test a two-dimensional Dirac spin liquid

Section titled “6. Test a two-dimensional Dirac spin liquid”

A parton mean-field calculation gives four gapless Dirac fermions coupled to U(1). What additional information is required before calling the microscopic spin model a stable U(1) Dirac spin liquid?

Solution

One must restore compact gauge fluctuations and classify the allowed monopole operators under every microscopic symmetry. Their scaling dimensions determine whether they are relevant. Gauge-invariant fermion bilinears and four-fermion perturbations must also be tested. The mapping from continuum operators to lattice spin, momentum, and point-group channels should predict measurable correlators. Finally, finite-size numerics must distinguish algebraic behavior from nearby Néel or valence-bond order, and spin-lattice coupling must be checked because it can couple to monopoles. A stable large-NfN_f fixed point alone does not establish stability at Nf=4N_f=4.

Where the description can fail

Section titled “Where the description can fail”

Compact fluctuations can confine a mean-field spin liquid. In two dimensions, the symmetry quantum numbers and scaling dimensions of monopole operators must therefore be established for the actual microscopic model. Gapless matter can help, but a controlled large-flavor result does not settle a small-flavor material. Spin-lattice coupling can also destabilize a proposed Dirac state, as illustrated by Seifert and collaborators.

The spectrum separates several ideas often confused in words: a fully gapped discrete gauge liquid has no photon; a gapless discrete gauge liquid can have gapless matter; non-Abelian anyons need not imply non-Abelian propagating gauge bosons. For the corresponding excitation and braiding structures, continue with Fractionalization, Topological Order, and Anyons and Braiding.

References

Section titled “References”