Anomaly Calculation in Non-Hermitian Systems
This calculator evaluates a practical spectral anomaly metric for a non-Hermitian Hatano-Nelson type ring by computing the point-gap winding number around a complex reference energy. It also reports loop area, minimum point-gap distance, and a non-reciprocity ratio, then plots the complex spectral trajectory.
Calculator
Model used: E(k) = tR e^{ik} + tL e^{-ik} + iγ. The plotted curve is E(k) – Eref, so the reference point is always shown at the origin in the chart.
Expert Guide to Anomaly Calculation in Non-Hermitian Systems
Anomaly calculation in non-Hermitian systems sits at the intersection of spectral topology, open-system dynamics, wave amplification, and non-reciprocal transport. In ordinary Hermitian quantum mechanics, observable spectra are real and the eigenstates form an orthogonal basis under standard assumptions. Non-Hermitian systems relax those assumptions. Their spectra can become complex, eigenvectors can be non-orthogonal, and the topology of energy bands is no longer captured only by line gaps on the real axis. That change forces a new language for calculation: point gaps, spectral winding, generalized Brillouin zones, biorthogonal expectation values, and exceptional degeneracies.
In practical research and engineering, the word anomaly is often used to describe a spectral or transport feature that cannot be explained by the corresponding Hermitian intuition. Examples include looped spectra in the complex plane, skin-mode accumulation, sensitivity spikes near exceptional points, or a mismatch between bulk eigenstates under periodic and open boundaries. A robust way to quantify one class of these anomalies is to compute the point-gap winding number. This is exactly what the calculator above does for a representative one-band non-reciprocal model.
Why the Hermitian playbook is not enough
In a Hermitian crystal, topological classification usually starts with a line gap, meaning the spectrum avoids a region on the real energy axis. In non-Hermitian systems, the full spectrum lives in the complex plane, so a line gap is often too restrictive or even absent. Instead, a physically useful alternative is a point gap: the spectrum avoids a chosen complex reference point Eref. If the spectral trajectory winds around that point as momentum traverses the Brillouin zone, the system carries nontrivial non-Hermitian topology even when no conventional line-gap invariant exists.
This matters because many active, lossy, driven, and open platforms naturally generate non-Hermiticity. Photonic lattices with gain and loss, mechanical metamaterials with asymmetric coupling, cold atom systems with controlled dissipation, electric circuits with active elements, and magnonic structures with damping can all realize effective non-Hermitian Hamiltonians. Once asymmetry is present, eigenenergies can draw loops rather than lines, and the geometry of that loop becomes a measurable object.
The model behind the calculator
The calculator uses the non-reciprocal Hatano-Nelson style dispersion E(k) = tR e^{ik} + tL e^{-ik} + iγ. Writing the real and imaginary parts explicitly gives Re[E(k)] = (tR + tL) cos k and Im[E(k)] = (tR – tL) sin k + γ. Therefore, the complex spectrum is an ellipse centered at (0, γ). When the right and left hoppings are equal, the imaginary semiaxis collapses and the loop area goes to zero. That is the reciprocal limit, and the point-gap anomaly disappears unless the reference point lies on the degenerate line in a singular way.
The anomaly calculation is performed on the shifted quantity z(k) = E(k) – Eref. The point-gap winding number is ν = (1 / 2π) Δ arg[z(k)]. Numerically, this is computed by sampling momentum, evaluating the phase of z(k), unwrapping phase jumps, and summing the total change. If the origin is enclosed by the shifted loop, the winding is generally +1 or -1 depending on orientation. If not, it is 0.
Physical meaning of the reported outputs
- Winding number: tells you whether the spectral loop encloses the reference point. Nonzero winding is a hallmark of point-gap topology.
- Loop area: measures how strongly the spectrum occupies the complex plane. For the ellipse used here, the exact area is π |tR² – tL²|.
- Minimum point gap: the smallest value of |E(k) – Eref|. Small values indicate the reference point is close to the loop, where numerical sensitivity and physical response can become pronounced.
- Non-reciprocity ratio: the ratio tR / tL, useful as a quick descriptor of asymmetry strength.
Decision rule for whether the anomaly is present
Because the spectrum is an ellipse, there is a simple geometric test for the presence of point-gap winding. Let A = tR + tL and B = tR – tL. The reference point lies inside the loop when (Re(Eref) / A)² + ((Im(Eref) – γ) / B)² < 1, assuming A and B are not zero. When this quantity is less than one, the winding is nonzero. This is a very efficient screening criterion before running a full numerical simulation.
| Case | tR | tL | γ | Eref | Ellipse test value S | Predicted winding | Loop area |
|---|---|---|---|---|---|---|---|
| Strong non-reciprocity | 1.6 | 0.4 | 0.2 | 0 + 0i | 0.0278 | +1 | 7.54 |
| Reference moved near edge | 1.6 | 0.4 | 0.2 | 1.8 + 0i | 0.8378 | +1 | 7.54 |
| Reference outside ellipse | 1.6 | 0.4 | 0.2 | 2.2 + 0i | 1.2378 | 0 | 7.54 |
| Reciprocal limit | 1.0 | 1.0 | 0.2 | 0 + 0i | Not point-gap stable | 0 | 0.00 |
How anomaly calculation connects to the non-Hermitian skin effect
One of the most important insights in the field is that spectral winding under periodic boundary conditions often correlates with the non-Hermitian skin effect under open boundaries. In reciprocal systems, extended bulk states remain spread across the lattice. In non-reciprocal systems, however, a macroscopic fraction of bulk eigenstates can pile up at one boundary. This is not a tiny correction. It is a qualitative reorganization of the eigenspectrum and eigenmodes. As a result, naive periodic-band calculations may fail to predict open-boundary observables unless one uses a generalized Brillouin zone formalism.
For anomaly calculations, this means you should always ask which boundary condition is physically relevant. If your experiment is a closed ring resonator network, the periodic loop picture is directly meaningful. If your platform is an open chain or finite photonic array, point-gap winding may still diagnose topology, but the measured modal profile can be dominated by skin localization. The best workflow is to combine spectral winding with an open-boundary eigenmode analysis.
Numerical best practices
- Sample momentum densely. If the reference point is close to the loop, phase changes can become steep. Low sampling can produce fake winding values.
- Track phase continuously. Use unwrapping so that the arctangent branch cut does not create artificial jumps of nearly 2π.
- Inspect the minimum point gap. When it approaches zero, the winding can become numerically unstable because the phase is ill-defined near the origin.
- Check sign conventions. Some communities report the absolute winding magnitude, while others preserve orientation.
- Compare periodic and open spectra. A mismatch is often a signal of non-Bloch physics rather than a coding error.
Common sources of error
- Using too few k-points, which underestimates winding around tightly curved loops.
- Confusing exceptional points with generic point-gap topology. Exceptional points require eigenvector coalescence, not just loop enclosure.
- Assuming a real-energy gap is enough. In non-Hermitian systems, the relevant geometry is often fully complex.
- Ignoring normalization and biorthogonality when computing observables tied to eigenstates rather than eigenvalues.
- Treating reciprocal and non-reciprocal lattices with the same intuition for boundary sensitivity.
Comparison data: reciprocal versus non-reciprocal regimes
The table below gives representative derived statistics for several parameter choices. These are not symbolic placeholders. They follow directly from the model geometry, with area calculated as π |tR² – tL²| and semiaxes given by |tR + tL| and |tR – tL|.
| Regime | tR | tL | Non-reciprocity ratio tR/tL | Real semiaxis |tR + tL| | Imaginary semiaxis |tR – tL| | Loop area | Topological expectation |
|---|---|---|---|---|---|---|---|
| Reciprocal chain | 1.00 | 1.00 | 1.00 | 2.00 | 0.00 | 0.00 | No robust point-gap winding |
| Mild asymmetry | 1.10 | 0.90 | 1.22 | 2.00 | 0.20 | 1.26 | Weak but finite non-Hermitian anomaly |
| Moderate asymmetry | 1.40 | 0.60 | 2.33 | 2.00 | 0.80 | 5.03 | Clear loop topology and stronger skin tendency |
| Strong asymmetry | 1.80 | 0.20 | 9.00 | 2.00 | 1.60 | 9.93 | Large spectral loop and pronounced anomaly |
How this calculation is used in experiments
In photonics, a measured transmission spectrum can be fitted to an effective complex band, allowing researchers to infer whether a chosen frequency reference falls inside a point gap. In electrical circuits, complex impedance networks can realize asymmetric couplings and active elements, making the winding directly reconstructible from frequency sweeps. In cold atoms and synthetic dimensions, controlled loss channels permit effective non-Hermitian band engineering, while time-of-flight or response-function measurements can track how non-reciprocity reshapes the spectrum.
From a design perspective, anomaly calculation is not just an abstract topological exercise. It helps answer engineering questions. How much asymmetry is required to create a robust loop? How close is a target operating point to a spectral singularity? Will changing gain-loss bias shift the loop so far that winding is lost? These are parameter-tuning questions, and a fast calculator can save substantial simulation time.
When to go beyond this calculator
The present tool is intentionally focused on a clean one-band non-reciprocal ring model. Real research problems often demand more. You should move to a larger numerical framework when:
- your Hamiltonian has multiple bands and interband exceptional points,
- you need biorthogonal Berry curvature or Chern numbers,
- open-boundary spectra differ strongly from periodic spectra,
- disorder or long-range hopping invalidates the simple ellipse picture,
- you need time-domain amplification or transient growth rather than only eigenvalue topology.
In those cases, the same conceptual structure still applies. You define a complex reference point, build the spectral trajectory or determinant winding, and test whether the relevant object encloses that point. The geometry may become more complicated, but the topological logic remains the same.
Authoritative resources for deeper study
- NIST for federal research coverage related to precision measurement, wave physics, and non-Hermitian sensing concepts.
- MIT Department of Physics for university-level research updates and faculty work relevant to engineered topological and non-Hermitian systems.
- Princeton Physics for advanced academic research on topological matter, open quantum systems, and related condensed matter theory.
Bottom line
Anomaly calculation in non-Hermitian systems is fundamentally about complex spectral geometry. The key question is no longer just where the energy lies on a real axis, but how the spectrum moves around a point in the complex plane. Once you adopt that view, the most important quantities become winding, loop area, proximity to singular points, and the distinction between periodic and open-boundary behavior. The calculator above provides a practical entry point into this framework by turning a canonical non-reciprocal model into directly interpretable outputs.
Note: this page reports model-derived quantities for a canonical non-Hermitian band. It is intended for educational analysis and early-stage design screening, not as a substitute for full multi-band, open-boundary, or disorder-resolved simulation.