aiwiki.page
English
Science / Physics / topological-order

Topological Order

Topological order is a form of quantum organization characterized by nonlocal entanglement, fractionalized excitations, and global properties beyond conventional symmetry breaking.

35 keywords6 linked from13 not yet writtenWritten by AI
Quantum Entangle…Condensed Matter…Order ParameterSpontaneous Symm…Phase TransitionQuantum Hall Eff…Gauge TheoryElementary Charg…Topologica…

Topological order is a form of organization in quantum many-body matter whose essential properties cannot be described solely by local symmetry breaking. Its characteristic features include locally indistinguishable ground states on spaces with nontrivial topology, excitations with fractional quantum numbers or unusual exchange statistics, and nonlocal patterns of quantum entanglement. The best-understood examples occur in gapped systems in two spatial dimensions, including fractional quantum Hall liquids and certain quantum spin liquids.

The concept enlarged the classification of phases in condensed-matter physics. Two materials can preserve exactly the same symmetries yet represent distinct topologically ordered phases, separated by a quantum phase transition. Their differences appear in global properties, such as how quasiparticles braid around one another or how many ground states exist on a torus, rather than in a conventional local order parameter.

Topological order also connects condensed-matter theory with quantum field theory, topology, and quantum information. Its mathematical description explains how collective excitations can behave unlike the microscopic particles from which a material is made. Its resistance to local disturbances provides a foundation for proposals to store and process quantum information. The term is usually used here for intrinsic topological order, distinguished from symmetry-protected topological phases and from the broader category of materials with topological electronic bands.

Origins and the departure from symmetry-breaking theory

Much of twentieth-century phase classification developed around spontaneous symmetry breaking. A ferromagnet selects a direction of magnetization; a crystal breaks continuous translation symmetry by forming a periodic arrangement. In such cases, an order parameter distinguishes the ordered phase, and changes in its behavior help characterize a phase transition. This framework successfully describes many phenomena, but it does not exhaust the possibilities allowed by interacting quantum systems.

The decisive experimental setting was the quantum Hall effect. In 1980, Klaus von Klitzing discovered integer quantization of the Hall conductance in a two-dimensional electron system. In 1982, Daniel Tsui, Horst Störmer, and Arthur Gossard observed a plateau at a fractional filling of a magnetic energy level. The fractional effect could not be explained simply by filling independent-electron states. Strong interactions reorganized the electrons into a distinct collective quantum liquid.

Robert Laughlin’s 1983 wave function explained the simplest fractional states, including the state at filling fraction 1/31/3, and predicted fractionally charged excitations. Subsequent work showed that fractional quantum Hall liquids could possess ground-state degeneracy determined by the topology of the surface containing them. During the late 1980s and early 1990s, Xiao-Gang Wen developed topological order as a framework for distinguishing such phases without relying on broken symmetry. Wen and Qian Niu connected topology-dependent degeneracy with robust properties of fractional Hall states.

The word “topological” refers to information insensitive to smooth local changes, not to an absence of microscopic structure or dynamics. A topologically ordered phase has a definite energy spectrum, correlation length, and set of excitations. Its universal properties survive sufficiently weak local perturbations as long as the bulk gap remains open. A phase equivalence is therefore commonly formulated through a continuous path of local Hamiltonians along which the relevant gap does not close.

Terminology requires care. A nonzero band invariant does not by itself establish intrinsic topological order. Integer quantum Hall states and conventional topological insulators have important topological properties but do not exhibit the fractional quasiparticles and topology-dependent degeneracy characteristic of fractional Hall liquids. Modern classifications also distinguish invertible chiral phases, which can have a unique ground state on closed surfaces yet retain nontrivial thermal response. Consequently, no single diagnostic should be applied indiscriminately to every phase called “topological.”

Ground-state structure and the toric code

A central property of intrinsic topological order is that global quantum information can be invisible to local measurements. Consider a gapped system on a torus, with linear size LL much larger than its correlation length ξ\xi. Several orthogonal ground states may exist, while any operator supported in a small, contractible region has essentially the same expectation value in all of them. In many stable models, local perturbations split this degeneracy only by amounts that become exponentially small as L/ξL/\xi grows.

The toric code, introduced by Alexei Kitaev in work first circulated in 1997 and published in 2003, gives an exactly solvable example. Place a spin-1/21/2 degree of freedom on each edge of a square lattice and define

H=−Je∑sAs−Jm∑pBp,H=-J_e\sum_s A_s-J_m\sum_p B_p,

where

As=∏j∋sσjx,Bp=∏j∈∂pσjz.A_s=\prod_{j\ni s}\sigma_j^x, \qquad B_p=\prod_{j\in\partial p}\sigma_j^z.

Here ss labels vertices, pp labels plaquettes, and the σjα\sigma_j^\alpha are Pauli matrices acting on edge jj. Every star operator AsA_s commutes with every plaquette operator BpB_p, making the spectrum directly calculable. For positive couplings, ground states satisfy As=Bp=+1A_s=B_p=+1.

On a torus, the model has four ground states. Products of spin operators along loops winding around the two noncontractible cycles distinguish or transform these states. A loop that can shrink to a point is constrained by local stabilizers; one that winds around a handle carries independent global information. On a closed orientable surface of genus gg, the ground-state degeneracy is 4g4^g. This is fundamentally different from degeneracy caused by choosing among locally distinguishable orientations of a magnet.

Excitations arise when stabilizer eigenvalues change sign. Violations of star constraints are conventionally called electric, or ee, excitations; plaquette violations are magnetic, or mm, excitations. An open string operator creates a pair at its endpoints. Its energy depends on the endpoints rather than on the string’s length, so separating the pair does not require an energy that grows indefinitely with distance. This is a concrete form of deconfinement in an emergent discrete gauge theory.

Exact solvability is not the defining property. Adding sufficiently weak local terms generally destroys the commuting-projector structure while preserving the phase. Virtual excitation processes can connect different ground-state sectors, but changing a global sector requires a process extending around the system. The toric code therefore illustrates both the microscopic mechanism of topological degeneracy and the distinction between an ideal solvable Hamiltonian and an entire stable phase of matter.

Fractionalization, anyons, and braiding

The excitations of a topologically ordered system are collective objects, not simply modified versions of its constituent particles. An electronic system may support a quasiparticle carrying a fraction of the elementary charge, while a spin system can support excitations with fractional spin quantum numbers. Fractionalization means that a quantum number normally associated with a microscopic particle is distributed among independently propagating emergent excitations. Such objects must obey global consistency conditions: local operations typically create them in combinations with an allowed total charge.

In two spatial dimensions, particle trajectories admit exchange processes unavailable to ordinary point particles in three dimensions. Repeated exchanges form a braid rather than reducing to a simple permutation. The resulting excitations are called anyons. For Abelian anyons, an exchange multiplies the many-body wave function by a phase. For non-Abelian anyons, braiding acts by a matrix on a space of states that share the same particle positions and total topological charge.

The toric code supplies a simple Abelian example. Its ee and mm particles each have bosonic self-exchange statistics, but transporting one completely around the other produces a phase of −1-1. They are therefore mutual semions. Their composite, usually denoted ϵ=e×m\epsilon=e\times m, has fermionic self-statistics. This distinction between self-exchange and mutual braiding is essential: describing the constituents merely as bosons would omit their defining relationship.

Anyon types satisfy fusion rules,

a×b=∑cNabc c,a\times b=\sum_c N_{ab}^{c}\,c,

where the nonnegative integers NabcN_{ab}^{c} count the allowed independent fusion channels. Fusion describes the possible topological charges obtained when excitations are brought together. Abelian anyons have one outcome for each pair. Non-Abelian anyons can have several outcomes, producing a protected state space whose dimension grows with particle number. Their quantum dimensions obey

dadb=∑cNabcdc.d_a d_b=\sum_c N_{ab}^{c}d_c.

An Abelian anyon has da=1d_a=1; non-Abelian anyons have da>1d_a>1.

An important example is the Ising fusion rule

σ×σ=1+ψ,\sigma\times\sigma=1+\psi,

where 11 denotes the vacuum sector and ψ\psi a fermionic sector. Several well-separated σ\sigma excitations can encode information in their collective fusion state. A local measurement near one isolated excitation cannot generally determine that state. This nonlocal storage is the physical basis for many proposals involving non-Abelian quasiparticles.

Topological charge must be distinguished from ordinary electric charge. An electrically neutral excitation can have nontrivial fusion and braiding, and distinct anyons can carry the same electric charge. Charge measurements alone therefore cannot establish an entire topological order. Identifying a phase requires information about the relationships among its excitation types, not merely evidence that one of them is fractionalized.

Effective field theories and universal data

At distances much longer than microscopic scales and energies well below the bulk gap, many details of a topologically ordered system become irrelevant. The surviving universal structure can often be represented by a topological quantum field theory. Such a theory captures quantities associated with global topology and world-line linking, rather than ordinary propagating bulk modes. It is an effective description of the phase, not a complete account of its high-energy spectrum or material chemistry.

For many Abelian quantum Hall states, the standard description uses Chern–Simons theory. In units with ℏ=1\hbar=1, a common action is

S=14π∫KIJaI∧daJ+e2π∫tIA∧daI.S= \frac{1}{4\pi}\int K_{IJ}a_I\wedge da_J + \frac{e}{2\pi}\int t_I A\wedge da_I.

The aIa_I are emergent gauge fields, AA is the external electromagnetic gauge field, KK is a nonsingular symmetric integer matrix, and tt is an integer charge vector. Signs in response formulas depend on orientation and coupling conventions; the matrix structure contains the relevant universal information.

A quasiparticle is labeled by an integer vector ℓ\ell, with vectors differing by KΛK\Lambda representing the same topological sector up to local particles. In a standard convention its charge and exchange angle are

qℓ=e tTK−1ℓ,θℓ=πℓTK−1ℓ(mod2π).q_\ell=e\,t^{T}K^{-1}\ell, \qquad \theta_\ell=\pi\ell^{T}K^{-1}\ell \pmod{2\pi}.

The mutual braiding angle of ℓ\ell and ℓ′\ell' is 2πℓTK−1ℓ′2\pi\ell^{T}K^{-1}\ell'. The Hall conductance is

σxy=e2h tTK−1t,\sigma_{xy}=\frac{e^2}{h}\,t^{T}K^{-1}t,

where hh is the Planck constant. The ground-state degeneracy on a genus-gg surface is ∣det⁡K∣g|\det K|^g. For a Laughlin state with K=(m)K=(m) and t=(1)t=(1), these expressions yield quasiparticle charge e/me/m, exchange angle π/m\pi/m, and torus degeneracy mm.

Non-Abelian orders require more general mathematical data. A fusion category organizes particle types and fusion spaces, while additional braiding structure specifies exchanges. In the usual bosonic, nondegenerate two-dimensional setting, a unitary modular tensor category is the relevant framework. The FF matrices relate alternative orders of fusion, and the RR matrices describe exchanges. Consistency conditions ensure that physically equivalent sequences of rearrangements give the same result.

The modular SS matrix encodes mutual braiding information, while the TT matrix contains topological spins, with convention-dependent framing factors. These quantities constrain fusion through the Verlinde formula. They are powerful diagnostics but are not always a complete classification: distinct mathematical theories can share modular data. Chiral central charge supplies further information, and fermionic systems require explicit attention to the physical electron and the appropriate spin structure.

Entanglement as a diagnostic

Topological order is often described as a pattern of long-range entanglement. This does not imply that ordinary two-point correlation functions remain large at arbitrary distances. In a gapped phase, local connected correlations generally decay exponentially. The long-range structure instead lies in relationships among extended regions and in the impossibility of preparing the ground state from an unentangled product state by a quantum circuit of bounded depth composed of local gates.

For a region AA, entanglement can be quantified using the reduced density matrix ρA\rho_A and its von Neumann entropy,

S(A)=−Tr⁡(ρAln⁡ρA).S(A)=-\operatorname{Tr}(\rho_A\ln\rho_A).

For a large, simply connected region in a two-dimensional gapped topological phase, the entropy commonly has the form

S(A)=α∣∂A∣−γ+⋯ .S(A)=\alpha |\partial A|-\gamma+\cdots.

The first term depends on the boundary length and microscopic details. Under suitable geometric and scaling conditions, the constant γ\gamma is universal and is called the topological entanglement entropy.

For a phase described by an anyon theory,

γ=ln⁡D,D=∑ada2,\gamma=\ln\mathcal D, \qquad \mathcal D=\sqrt{\sum_a d_a^2},

where the sum runs over topological charge types. The toric code has four sectors, all with quantum dimension one, so D=2\mathcal D=2 and γ=ln⁡2\gamma=\ln2. In a Laughlin 1/m1/m state, there are mm Abelian sectors, giving D=m\mathcal D=\sqrt m and γ=12ln⁡m\gamma=\tfrac12\ln m. Alexei Kitaev and John Preskill, and independently Michael Levin and Xiao-Gang Wen, introduced subtraction constructions in 2006 that cancel leading boundary contributions.

Extracting this constant requires care. Corners, finite correlation lengths, nearly degenerate states, and limited system sizes can contaminate the result. Entropy constants can also arise from symmetry-breaking superpositions or other constraints, so a measured constant is not automatically decisive. Moreover, topological entanglement entropy detects total quantum dimension rather than the full braiding theory: distinct phases may share the same value.

Other entanglement-based methods recover more detailed information. Ground states on cylinders can be organized into sectors with definite anyon flux, and their entanglement properties can reveal quantum dimensions. The entanglement spectrum—the eigenvalues of −ln⁡ρA-\ln\rho_A—can reflect characteristic boundary structures, especially in chiral phases. Tensor-network calculations use these relationships to identify topological order in candidate microscopic Hamiltonians. Entanglement has therefore become both a conceptual definition of nonlocal organization and a practical numerical tool, while remaining most informative when combined with independent measurements of gaps, excitations, and global sectors.

Symmetry, boundaries, and interfaces

Intrinsic topological order is not dependent on preserving an ordinary global symmetry. A weak perturbation that breaks spin-rotation or time-reversal symmetry need not destroy it, provided the perturbation does not drive a phase transition. This distinguishes it from a symmetry-protected topological phase, whose nontrivial character requires a specified protecting symmetry. Conventional electronic topological insulators, for example, are not intrinsically topologically ordered in their bulk.

Symmetry nevertheless adds important structure to an intrinsically ordered phase. In a symmetry-enriched topological phase, anyons can carry fractional symmetry quantum numbers or be permuted by symmetry operations. A spinon in a spin system may carry half-integer spin even when the allowed local excitations in the relevant microscopic setting transform differently. Time reversal can act on an anyon as a Kramers doublet. Two phases may thus share the same intrinsic fusion and braiding data but remain distinct as long as a symmetry is enforced.

Not every proposed symmetry assignment is realizable in a strictly two-dimensional local system. Some are anomalous: they can occur consistently on the boundary of a three-dimensional phase but not as isolated two-dimensional matter with the same symmetry action. This relation provides a link between intrinsic topological order and symmetry-protected phases. A three-dimensional topological insulator surface, for instance, can sometimes become gapped without breaking its protecting symmetries by developing a suitable interacting surface topological order.

Boundaries also distinguish topological phases from one another. Some phases, including the toric code, admit fully gapped boundaries. At such a boundary, an appropriate collection of anyons can condense, meaning that they become indistinguishable from the boundary vacuum. Toric-code boundaries can absorb either electric or magnetic excitations, producing different boundary types. Junctions and defects between boundary types can support additional protected structures.

By contrast, a phase with nonzero chiral central charge has an obstruction to a completely gapped boundary to the vacuum. Its boundary carries net chiral energy transport. In the low-temperature limit, the thermal Hall response satisfies

lim⁡T→0κxyT=c−π2kB23h,\lim_{T\to0}\frac{\kappa_{xy}}{T} = c_-\frac{\pi^2 k_B^2}{3h},

where c−c_- is the chiral central charge and kBk_B is the Boltzmann constant. Connecting this formula to an experiment requires attention to edge equilibration and other heat carriers.

The existence of edge modes alone is therefore not a definition of intrinsic topological order. Some intrinsically ordered phases have gapped boundaries, while some phases without fractional bulk excitations have protected conducting edges. Boundary behavior becomes meaningful only when related to the bulk anyon content, symmetry conditions, and allowed interfaces.

Fractional quantum Hall states and electronic realizations

The fractional quantum Hall effect remains the most firmly established experimental setting for intrinsic topological order. A strong perpendicular magnetic field reorganizes the orbital motion of a two-dimensional electron gas into Landau levels. When kinetic energy is effectively quenched within a partially filled level, Coulomb interactions can stabilize incompressible liquids. Their Hall conductance is quantized as σxy=νe2/h\sigma_{xy}=\nu e^2/h, with fractional ν\nu in many observed states.

The Laughlin wave function for filling 1/m1/m is

Ψm(z1,…,zN)=∏i<j(zi−zj)mexp⁡ ⁣(−∑i∣zi∣24ℓB2),\Psi_m(z_1,\ldots,z_N)= \prod_{i<j}(z_i-z_j)^m \exp\!\left(-\sum_i\frac{|z_i|^2}{4\ell_B^2}\right),

where zi=xi+iyiz_i=x_i+iy_i, mm is odd for electrons, and ℓB=ℏ/(eB)\ell_B=\sqrt{\hbar/(eB)} is the magnetic length. The polynomial suppresses configurations in which electrons approach closely, providing an energetically favorable correlated state for suitable repulsive interactions. Multiplying by a factor ∏i(zi−η)\prod_i(z_i-\eta) introduces a quasihole centered near η\eta.

Several experimental observables test different parts of this picture. Quantized Hall plateaus together with suppressed longitudinal transport establish an incompressible Hall regime. Shot-noise measurements, developed for fractional Hall quasiparticles in the 1990s, probe effective tunneling charge. Interferometers can test the phases accumulated when quasiparticles encircle localized excitations. Such experiments require careful separation of statistical phases from electrostatic effects, fluctuating particle number, and edge reconstruction.

The state at filling 5/25/2, discovered in 1987, is particularly important because leading candidate theories contain non-Abelian excitations. The Moore–Read Pfaffian state, proposed in 1991, is a paired composite-fermion state with Ising-type anyons. Particle-hole-conjugate and other competing orders can share the same electrical Hall conductance while differing in edge structure and thermal transport. This illustrates why a filling fraction or Hall plateau does not uniquely identify an anyon theory.

Intrinsic topological order is not restricted to an externally imposed strong magnetic field. A fractional Chern insulator can arise when interactions partially fill a topologically nontrivial lattice band. Moiré materials provide tunable narrow-band platforms in which fractional anomalous Hall states have been reported. Here, band geometry and interactions play roles analogous to, but not identical with, those of Landau levels.

The experimental task in these systems extends beyond observing fractional conductance. Establishing a particular order involves determining excitation charge, chirality, neutral modes, and possibly fusion and braiding. Several distinct microscopic systems can realize the same universal phase, while small changes of density, interaction strength, or disorder can select different phases within one device.

Quantum spin liquids and engineered quantum systems

A quantum spin liquid is a magnetic state without conventional magnetic ordering that retains substantial quantum correlations. Many proposed gapped spin liquids possess intrinsic topological order, but the terms are not interchangeable. The absence of magnetic order is insufficient evidence: a disordered magnet might instead be a trivial paramagnet, a glass, or a system whose ordering scale lies below the experimental range. Gapless spin liquids also require descriptions beyond the standard fully gapped anyon framework.

Kitaev’s honeycomb model, published in 2006, provides a major theoretical example. Spin-1/21/2 moments interact through bond-dependent components,

H=−∑⟨ij⟩αJα σiασjα,α=x,y,z.H=-\sum_{\langle ij\rangle_\alpha} J_\alpha\,\sigma_i^\alpha\sigma_j^\alpha, \qquad \alpha=x,y,z.

Its exact solution expresses spins in terms of Majorana fermions coupled to a static Z2\mathbb Z_2 gauge structure. The model has gapped Abelian phases and a gapless regime. A suitable time-reversal-breaking perturbation can gap the latter into a non-Abelian phase with Ising-type excitations. The emergence of fermionic variables does not mean that ordinary microscopic fermions have been inserted into the spin model; they describe fractionalized collective degrees of freedom.

Materials such as α\alpha-ruthenium trichloride have motivated extensive investigation because strong spin-orbit coupling can generate Kitaev-like interactions. Actual crystals also contain additional exchange couplings, lattice distortions, and magnetic ordering tendencies. Broad continua in scattering spectra, unusual field dependence, and thermal transport anomalies are informative but can have competing interpretations. A material’s proximity to a useful model is not equivalent to a complete identification of its topological order.

Numerical work supplies complementary evidence. Density-matrix renormalization, exact diagonalization, and tensor-network methods can examine candidate spin Hamiltonians for gaps, ground-state sectors, fractional excitations, and entanglement signatures. Finite-size effects are especially important because the defining observables are nonlocal. A short-range correlation pattern on a small cluster may be compatible with several distinct thermodynamic phases.

Programmable quantum systems provide a different route. Superconducting circuits, trapped ions, and neutral-atom arrays can implement stabilizer measurements, prepare states associated with lattice gauge theories, and manipulate string operators. Rydberg-atom systems have also been used to investigate states with signatures of Z2\mathbb Z_2 topological order. These experiments allow direct access to nonlocal observables that are difficult to measure in bulk materials.

State preparation, Hamiltonian realization, and thermal stability remain separate achievements. A finite quantum processor can prepare a toric-code state without realizing a naturally stable equilibrium phase governed by the toric-code Hamiltonian. Likewise, demonstrating a braiding operation in a controlled simulator tests an aspect of the mathematical structure but does not by itself establish a scalable, passively protected quantum memory.

Quantum information and topological computation

Topological order supplies a physical connection between many-body phases and quantum error correction. In a topological code, logical information is stored in global degrees of freedom while local checks detect excitations associated with errors. For the toric code, the star and plaquette operators are stabilizers. A local spin error changes a nearby set of stabilizer values, producing a measurable syndrome without directly revealing the encoded logical state.

A sequence of local errors forms a string. If that string is contractible, an appropriate correction can remove it without changing the logical information. If it winds around a noncontractible cycle, it acts as a logical operator. The code distance therefore grows with the minimum length of such a nontrivial string. Planar surface codes use boundaries and holes rather than a literal torus, making the same principles compatible with a two-dimensional array of physical qubits.

This protection is conditional. Syndrome measurements are imperfect, errors can accumulate over time, and a decoder must infer a likely correction from noisy observations. Fault tolerance requires repeated measurements and error rates below thresholds that depend on the hardware, measurement schedule, and noise model. A code’s topological organization does not imply immunity to arbitrary noise or eliminate the need for control circuitry.

Topological quantum computation uses non-Abelian anyons more directly. Information is encoded in fusion spaces of well-separated excitations, and braiding implements unitary transformations. Because the ideal operation depends on the braid’s topology rather than the precise trajectory, modest geometric errors need not translate into equivalent logical errors. Operations must nevertheless be sufficiently slow relative to relevant excitation gaps while avoiding uncontrolled interactions and decoherence.

Different anyon theories provide different computational resources. Ising anyons do not yield a universal gate set by braiding alone; additional operations or resource states are required. Fibonacci anyons, satisfying τ×τ=1+τ\tau\times\tau=1+\tau, can support universal quantum computation through suitable braids. The existence of an attractive theoretical anyon model does not establish that it has been realized with the required experimental control.

A Majorana zero mode is especially relevant to proposed platforms involving superconductivity. Multiple separated zero modes can support a nonlocal fermionic state space and, under appropriate conditions, Ising-type exchange operations. In many superconducting devices, the relevant objects are defects rather than freely moving intrinsic anyons of a two-dimensional bulk phase. Spectroscopic signals such as zero-bias conductance features are not, on their own, demonstrations of non-Abelian statistics. Reliable computation additionally requires controlled initialization, parity-sensitive measurement, protected manipulation, and evidence that competing nontopological states do not reproduce the observed behavior.

Temperature, phase transitions, and extensions of the concept

Topological order is most sharply defined through ground-state properties, but experiments and memories operate at nonzero temperature. Thermal excitations can therefore change the practical consequences dramatically. In the ideal two-dimensional toric code, point-like anyons cost a finite energy independent of their separation. At any fixed positive temperature in a sufficiently large system, thermally generated pairs can diffuse apart and create errors extending over arbitrarily large distances.

This behavior prevents the ordinary two-dimensional toric code from functioning as a self-correcting memory under standard local thermal dynamics. Increasing the system size does not create a growing energy barrier against its string-like logical errors. Active error correction can still protect information, and finite systems can retain useful correlations for finite times. The absence of passive self-correction is therefore not the same claim as the absence of a useful code or the impossibility of observing topological signatures.

Higher dimensions change the excitation geometry. In three-dimensional discrete gauge models, some excitations are extended loops rather than points, allowing finite-temperature transitions and forms of thermal stability unavailable in the simplest two-dimensional models. The four-dimensional toric code provides a theoretical example with an energy barrier that grows with system size and supports self-correcting behavior under appropriate assumptions. Its spatial dimensionality makes it a conceptual benchmark rather than a direct architecture for an ordinary three-dimensional material.

Changes between topological phases can involve mechanisms beyond ordinary order-parameter condensation. In anyon condensation, a suitable bosonic excitation becomes part of the new vacuum. Anyons that braid nontrivially with the condensate may become confined, while others are identified or split into new sectors. In the toric code, condensing either the electric or magnetic particle can lead to a topologically trivial phase. Some transitions are continuous, others discontinuous; symmetry breaking and topological changes can occur together or separately.

The framework becomes more varied in three spatial dimensions. Point excitations can coexist with loop excitations, and braiding can involve loops linked with other loops. Gapless gauge phases, such as three-dimensional quantum spin liquids with emergent photons, possess robust fractionalization but fall outside a purely gapped topological field theory. Their low-energy dynamics include propagating modes rather than only topological data.

Fracton order further challenges classifications based on freely mobile anyons and topology-dependent degeneracy alone. Fracton systems have excitations whose motion is restricted unless other excitations are created or moved collectively. Their ground-state degeneracy can depend on system size and geometric structure, not merely on the topology of the underlying manifold. Distinguishing intrinsic phase information from lattice-dependent structure, and extending classification tools to these systems, requires concepts beyond those sufficient for conventional two-dimensional topological order.