Quantum junctions produce a different butterfly spectrum

A quantum particle’s allowed energies depend not only on where it can travel, but also on what happens where its paths meet. In a new theoretical study, Pavel Exner and Jan Pekař examine an infinite square network in a magnetic field. Changing the junction rules produces a butterfly-like spectral pattern distinct from the familiar Hofstadter butterfly.

The result concerns an idealized mathematical system, not a measured material. The supplied record identifies an arXiv preprint and provides no information establishing journal peer review.

A square network with unusual junctions

The researchers use a quantum graph: a network whose edges are one-dimensional paths carrying quantum waves. Each junction connects four edges. Mathematical matching conditions specify how the wave values and their derivatives fit together there, defining the particle’s quantum dynamics.

The study restricts these conditions in two ways. They are circulant, meaning the junction rule is unchanged when the four attached edges are cyclically relabelled. They are also scale-invariant: the junction’s scattering rule does not depend on the particle’s energy. This does not remove the lattice spacing; it removes an energy-dependent scale from the junction itself.

These restrictions leave 14 nontrivial junction rules after excluding two cases in which every edge is completely disconnected from its neighbours. Eight of the 14 break time-reversal symmetry at the junction level. Such symmetry concerns whether reversing the quantum motion gives dynamics governed by the same rule.

A uniform magnetic field points perpendicular to the network. The main analysis assumes that the magnetic flux through each square is a rational fraction, p/q, of a magnetic flux quantum. That assumption makes a repeating magnetic cell available for calculation.

Turning continuous waves into a lattice calculation

Solving wave equations along infinitely many edges would be cumbersome. The authors instead derive an exact correspondence with a discrete lattice operator acting on quantities assigned to the vertices.

Away from certain exceptional momenta, an energy E = k² belongs to the graph’s spectrum precisely when 4 cos k belongs to the effective operator’s spectrum, in the paper’s units. Here, the spectrum is the set of allowed energies.

Depending on the junction rule, the effective description requires one, two or three components at each vertex. These components arise from the matching conditions; they do not introduce a separate physical spin into the original model.

The reduction, together with spectral symmetries, groups the 14 rules into three classes.

Three outcomes from the same geometry

Eight junction rules reproduce the spectrum of the conventional Kirchhoff-connected square graph, associated with the familiar Harper operator and Hofstadter butterfly. This class also has flat bands: energies supporting infinitely many independent states. The paper notes that flat bands outside the standard Kirchhoff case were checked numerically.

Four other rules produce the new pattern. Their effective description has two coupled components and can be rewritten as a one-dimensional chain with periodically varying connections and site terms. Numerical calculations reveal three butterfly-like clusters within each repeating interval of k, rather than the familiar pattern.

The detailed butterfly plot samples rational fluxes with denominators up to 30. The reported absence of flat bands in this class is also based on numerical calculations. These observations should not be confused with a proof of every fine-scale feature suggested by the image.

The remaining two rules separate horizontal and vertical motion completely. The network becomes independent straight lines, with every nonnegative energy allowed regardless of magnetic flux.

What the butterfly does—and does not—show

For the new class at zero magnetic field, the authors derive an exact dispersion relation connecting momentum and lattice wave phases. It predicts conical contacts between dispersion surfaces, often called Dirac contacts. They interpret these contacts as the starting points of the central butterfly cluster when flux is introduced.

The spectral pattern repeats in k, not in energy E = k². Equal intervals in k therefore do not represent equal energy intervals.

Nor is breaking time-reversal symmetry alone sufficient to create the new butterfly. Some junctions in the conventional spectral class also break that symmetry. The detailed structure of the junction rule matters.

Important questions remain open. The authors conjecture a Cantor-set spectrum for irrational flux, but do not establish it. Their study also does not demonstrate a physical implementation or measure electrical transport. Its contribution is more specific: an exact mathematical framework, supported by numerical spectra, showing how junction conditions can reorganize the allowed energies of an otherwise unchanged square quantum network.