Colored Hofstadter Butterfly

• Colored Hofstadter Butterfly is a fractal energy spectrum where each gap is assigned topological integers like Chern numbers, serving as a visual and mathematical phase diagram.
• It utilizes Diophantine equations and extended gap-labeling methods to map quantum Hall plateaus, with examples including honeycomb lattices and moiré superlattices.
• Recent developments expand the concept with Floquet driving, non-Hermitian deformations, and non-Abelian constructs, bridging theoretical models with experimental realizations.

The Hofstadter butterfly is the fractal energy spectrum of electrons on a two-dimensional lattice in a perpendicular magnetic field, plotted as a function of magnetic flux. For rational flux, the spectrum splits into subbands and gaps, and those gaps carry quantized Hall data. In the literature, the expression “colored Hofstadter butterfly” does not denote a single universally fixed object. It refers, depending on context, to a butterfly whose gaps are labeled by Chern numbers or Hall-conductivity integers, to a visually color-coded gap map, or to generalized butterflies enriched by Floquet sectors, non-Hermitian structure, spinor degrees of freedom, crystalline invariants, or interacting many-body Hall labels [(Agazzi et al., 2014); (Naumis et al., 2015); (Zhang et al., 2023); (Marcelli et al., 24 Jun 2026)].

1. From the Hofstadter spectrum to explicit color assignment

In the standard lattice problem, rational magnetic flux $\phi=p/q$ produces a finite magnetic periodicity and a hierarchy of subbands and gaps. A central organizing relation is the Diophantine equation. In the honeycomb-lattice construction of “The Colored Hofstadter Butterfly for the Honeycomb Lattice,” the $r$-th gap at flux $p/q$ is labeled by

$r=\sigma_H p+s q,$

where $\sigma_H$ is the Hall conductivity/Chern number in units of $e^2/h$ (Agazzi et al., 2014). That work computes $\sigma_H$ from bulk-edge correspondence: the Chern number is the winding number of an eigenvector of a $2\times 2$ transfer matrix as the quasi-momentum $k\in(0,2\pi)$ is varied. The resulting picture is literally a colored energy-flux diagram, with horizontal axis energy, vertical axis flux, and color equal to $\sigma_H$ (Agazzi et al., 2014).

The same paper shows that the honeycomb lattice obeys the same Diophantine gap-labeling equation as the square lattice, but the usual square-lattice window condition does not generally select the correct branch. The standard square-lattice choice $r$0, or shifted variants, fails in general for the honeycomb case, and extensive numerics led to the conjecture that the relevant relaxed condition is $r$1 (Agazzi et al., 2014). In that precise sense, the colored butterfly is not merely a visualization; it is a global assignment of topological integers across the full spectrum.

This construction established a durable template. Subsequent work repeatedly treated the butterfly not just as an energy plot, but as a phase diagram whose open gaps are equipped with integers determining Hall response, edge-state flow, or related topological data [(Agazzi et al., 2014); (Marcelli et al., 24 Jun 2026)].

2. Arithmetic skeleton, Chern hierarchies, and topological collapse

A major development was the reinterpretation of the butterfly as a topological fractal. In “Topological Map of the Hofstadter Butterfly and Van Hove Singularities,” each visible gap and sub-gap is assigned integer labels through

$r$2

where $r$3 is particle density, $r$4 is the Chern number, and $r$5 is a second integer gap label (Naumis et al., 2015). For rational flux $r$6, the family of solutions is

$r$7

so higher-order solutions appear naturally in the fine structure near rational fluxes (Naumis et al., 2015).

That paper gives a cut-and-projection or floor-function description:

$r$8

together with the inverse map

$r$9

The associated Hull function,

$p/q$0

is interpreted as a “skeleton butterfly” organizing branches with Chern numbers $p/q$1 without requiring a full numerical energy calculation (Naumis et al., 2015). In this usage, color is not a new observable; it is a graphical separation of topological sectors.

The same arithmetic viewpoint is sharpened by exact meeting and nesting rules. If two branches meet at $p/q$2, then

$p/q$3

Butterfly boundaries are generated by Farey neighbors, and the central, left, and right fluxes satisfy

$p/q$4

For butterflies centered on the horizontal axis with even $p/q$5, the diagonal Chern pair is

$p/q$6

while more generally

$p/q$7

(Naumis et al., 2015).

“Topology and Self-Similarity of the Hofstadter Butterfly” places these rules inside a dominant recursive hierarchy associated with the irrational

$p/q$8

for which flux intervals scale by $p/q$9 and topological integers scale by $r=\sigma_H p+s q,$0 (Satija, 2014). That work also emphasizes that, near a rational flux $r=\sigma_H p+s q,$1, the simplest topological changes satisfy

$r=\sigma_H p+s q,$2

which explains why large Chern numbers appear in the fine structure (Satija, 2014).

A complementary quasicrystalline formulation is given in “A higher-dimensional quasicrystalline approach to the Hofstadter and Fibonacci butterflies topological phase diagram,” where the Diophantine equation is written as

$r=\sigma_H p+s q,$3

and band conductances are encoded by two-letter Sturmian sequences generated by a circle-map threshold rule (Naumis, 2019). In that approach, “coloring” can be understood as symbolic or arithmetic coding of the topological phase diagram.

The most distinctive topological claim of the 2015 map is the identification of Van Hove singularities as topological collapse points: branches of positive and negative Chern number collide and annihilate at every band center, and a new nested structure then reappears at the next scale (Naumis et al., 2015). This makes the colored butterfly a map not only of stable gaps, but also of critical singular points where the topological order of the fractal reorganizes.

3. Visual color coding and experimental butterflies in moiré and porous lattices

In experimental and semi-experimental settings, “colored” often has a literal graphical meaning. In graphene on hexagonal boron nitride, the Hofstadter regime is organized by the Diophantine relation

$r=\sigma_H p+s q,$4

so each gap is labeled by an integer pair $r=\sigma_H p+s q,$5 (Dean et al., 2012). The Wannier diagram then consists of straight gap trajectories with slope $r=\sigma_H p+s q,$6 and intercept $r=\sigma_H p+s q,$7, and the paper explicitly uses color to distinguish families of states: white lines for conventional bilayer graphene quantum Hall states, yellow/red lines for anomalous Hofstadter states, and blue bars for conventional conductivity features (Dean et al., 2012). The bubble representation further uses circle radius proportional to the gap size (Dean et al., 2012).

In twisted bilayer graphene, “Moire Butterflies” presents a color-coded plot of energy gaps versus filling factor and rational flux ratio, with color scale

$r=\sigma_H p+s q,$8

so that small gaps remain visible (Bistritzer et al., 2011). The gap lines satisfy

$r=\sigma_H p+s q,$9

and the line intercept gives the Hall conductivity while the slope gives the second integer $\sigma_H$0 (Bistritzer et al., 2011). Here color represents gap magnitude, while the line geometry encodes topological labels.

These moiré works share the same physical mechanism: the moiré unit cell is large enough that the flux per superlattice cell reaches rational values at laboratory magnetic fields [(Dean et al., 2012); (Bistritzer et al., 2011)]. That is why color-coded Wannier and fan diagrams became central to the modern experimental visualization of the butterfly.

A different spatial route to layered or hierarchical butterflies appears in two-dimensional covalent-organic frameworks. “Hierarchies of Hofstadter butterflies in 2D covalent-organic frameworks” emphasizes that several distinct plaquette areas coexist in one lattice, each with its own characteristic period

$\sigma_H$1

The resulting spectrum contains superposed oscillations, nested butterflies, and distorted sub-butterflies associated with different pore sizes (Bodesheim et al., 2022). In that work, “colored” is not literal figure color but a hierarchy of layers generated by multiple loop scales. The most experimentally promising example is Starphene-COF, for which the large-plaquette characteristic field is reported as

$\sigma_H$2

(Bodesheim et al., 2022).

Across these realizations, visual color coding serves two distinct purposes. It can mark topological branch families, as in Wannier or Landau-fan diagrams, or it can emphasize gap magnitude and robustness. In both cases, the color scale externalizes otherwise abstract integer or energetic structure.

4. Floquet dressing, complexification, and the cocoon

A second major use of the colored-butterfly idea concerns spectra that acquire additional sectors under periodic driving or non-Hermitian deformation. In the non-Hermitian continuation of Harper’s equation,

$\sigma_H$3

the real parts of the eigenvalues still resemble a Hofstadter butterfly, while the imaginary parts form a separate fractal-like structure called the Hofstadter cocoon (Jones-Smith et al., 2014). As the non-Hermiticity parameter $\sigma_H$4 increases, pairs of real eigenvalues collide and become complex conjugate pairs in “double-pitchfork bifurcations,” reflecting spontaneous breaking of an anti-linear conjugation symmetry (Jones-Smith et al., 2014). In this setting, one may think of the butterfly as “colored” by $\sigma_H$5, but the paper’s own terminology distinguishes the real-part butterfly from the imaginary-part cocoon (Jones-Smith et al., 2014).

Floquet studies produce a different spectral multiplication. In honeycomb-lattice driving, monochromatic light converts the static energy spectrum into a Floquet quasienergy spectrum with replicas indexed by photon sector, and the resulting deformation depends on frequency, intensity, and polarization (Wackerl et al., 2018). Circular polarization twists the spectrum and breaks particle-hole symmetry, whereas linear polarization preserves particle-hole symmetry but deforms the bands anisotropically (Wackerl et al., 2018). The appropriate topological invariant of the driven system is not only the band Chern number but also the Rudner $\sigma_H$6 invariant, which counts chiral edge modes in quasienergy gaps (Wackerl et al., 2018).

On kagome and triangular lattices, periodic light reshapes the butterfly in a symmetry-selective way. Circularly polarized light breaks the reflection symmetry about $\sigma_H$7, while linearly polarized light preserves it; inversion symmetry about $\sigma_H$8 is preserved in both cases (Du et al., 2018). In the off-resonant regime, low circular intensity does not change the lowest-band Chern number, but beyond a characteristic intensity it does, consistent with the Bessel-function zero at

$\sigma_H$9

(Du et al., 2018). For linearly polarized light, both the spectrum and the Chern number depend on the polarization direction (Du et al., 2018).

The magnonic analogue extends the same logic to bosonic spin excitations. In the honeycomb ferromagnet, a spatially varying electric field induces an Aharonov–Casher phase and produces a magnonic Hofstadter spectrum; when the field is also time dependent, a Floquet Hofstadter butterfly emerges entirely from the oscillating electric field (Owerre, 2018). In that system the Chern numbers are odd integers,

$e^2/h$0

and the magnon Hall conductance is quantized by odd integers (Owerre, 2018).

These driven and complexified butterflies broaden the meaning of “colored.” The added “colors” are not conventional hues but extra spectral coordinates: imaginary parts, Floquet harmonics, quasienergy-zone copies, or photoinduced topological sectors.

5. Multicomponent butterflies: spin, non-Abelian gauge fields, and replicated copies

Internal degrees of freedom generate another family of colored butterflies. In two-dimensional non-Abelian generalizations of the Hofstadter model, scalar U(1) phases are replaced by SU(2) matrix-valued link phases, so the spectrum becomes a pair of spin-orbit-coupled butterflies rather than a single scalar one (Yang et al., 2020). The genuine non-Abelian condition is expressed through loop-operator noncommutativity, compactly written as

$e^2/h$1

and when that condition fails the system reduces to two decoupled Hofstadter butterflies (Yang et al., 2020). In the genuinely non-Abelian regime, the butterfly becomes a spin-resolved, coupled fractal with Weyl points, Dirac points, higher-fold Dirac points, and $e^2/h$2 topological-insulator gaps (Yang et al., 2020).

The three-dimensional extension replaces a single pair by a trio. In the cubic SU(2) model with gauge potentials

$e^2/h$3

each Cartesian surface reduces to a two-dimensional non-Abelian Hofstadter problem, so the full spectrum can be viewed as spin-orbit coupling among three butterflies living on the $e^2/h$4, $e^2/h$5, and $e^2/h$6 planes (Liu et al., 2020). The necessary and sufficient condition for genuine non-Abelian behavior is that at least two of $e^2/h$7 are neither $e^2/h$8 nor $e^2/h$9 (Liu et al., 2020). In that regime, new gaps appear and strong or weak three-dimensional topological insulating phases can occur (Liu et al., 2020).

A different replication mechanism is provided by $\sigma_H$0-root topology in decorated square lattices. There, increasing root degree produces $\sigma_H$1 flux-insensitive flat bands with energies

$\sigma_H$2

and the Hofstadter diagram becomes a “kaleidoscope of $\sigma_H$3 butterflies” separated by those flat bands (Marques et al., 2022). For $\sigma_H$4 there are two butterflies separated by one flat band; for $\sigma_H$5, four butterflies separated by three flat bands; in general, each butterfly replica is topologically equivalent to the original square-lattice Hofstadter butterfly (Marques et al., 2022). In this precise sense, the index $\sigma_H$6 acts as an additional fractal dimension of the diagram (Marques et al., 2022).

These multicomponent constructions make explicit what is only implicit in visual color schemes: the butterfly may be “colored” by spin, SU(2) structure, replicated parent-child band topology, or symmetry-enforced internal multiplicity.

6. Crystalline, quantum-geometric, and many-body colorings

Recent work has extended the butterfly’s color labels beyond one-particle Chern numbers. “Complete crystalline topological invariants from partial rotations in (2+1)D invertible fermionic states and Hofstadter’s butterfly” introduces additional colorings by many-body crystalline invariants $\sigma_H$7, together with the usual $\sigma_H$8, $\sigma_H$9, and $2\times 2$0 (Zhang et al., 2023). These invariants are extracted from partial-rotation expectation values on a disk,

$2\times 2$1

with

$2\times 2$2

and

$2\times 2$3

The paper states that these provide “additional colorings of Hofstadter’s butterfly,” extending earlier colorings by the discrete shift and quantized charge polarization (Zhang et al., 2023). Here color means a discrete topological label that distinguishes phases which can share the same Hall response.

A different non-visual reinterpretation comes from quantum geometry. “Hofstadter’s Butterfly in Quantum Geometry” identifies the butterfly with the branch-cut structure of the quantum A-period of local $2\times 2$4 at root of unity, and shows that the imaginary part of the A-period counts the number of states of the Hofstadter Hamiltonian (Hatsuda et al., 2016). In that framework, the arithmetic data $2\times 2$5 of rational $2\times 2$6 organize the spectral structure through the polynomial $2\times 2$7 (Hatsuda et al., 2016). Although the paper does not use “colored” as a formal term, it supports an arithmetic interpretation in which different rational sectors play the role of distinct layers.

The strongest recent statement is the many-body theorem of “The Colored Hofstadter Butterfly as a Many-Body Quantum Hall Phase Diagram.” Starting from a non-interacting spectral gap $2\times 2$8, the paper constructs an open neighborhood

$2\times 2$9

in the three-parameter space $k\in(0,2\pi)$0 on which the infinite-volume interacting system has locally unique gapped ground states (Marcelli et al., 24 Jun 2026). On connected uniformly gapped regions meeting the non-interacting plane $k\in(0,2\pi)$1, the Hall conductivity is constant and quantized,

$k\in(0,2\pi)$2

so the integer colors of the non-interacting butterfly persist as Hall-conductivity labels of interacting quantum Hall phases (Marcelli et al., 24 Jun 2026). This elevates the colored butterfly from a one-particle spectral map to a genuine many-body phase diagram.

A recurring source of ambiguity is therefore also the central conceptual fact. In some works, color is only a plotting device used to distinguish branches, circles, or baby butterflies, as emphasized in the 2025 overview that discusses color-coded visualizations rather than a new physical variable (Satija, 17 Jul 2025). In others, color is a topological label: Chern number, Hall conductivity, crystalline invariant, or interacting phase index [(Agazzi et al., 2014); (Zhang et al., 2023); (Marcelli et al., 24 Jun 2026)]. The term is best understood not as a unique object, but as a family of Hofstadter constructions in which the fractal spectrum is supplemented by a second layer of structure that can be tracked globally across flux and energy.