On a periodicity measure and superoscillations

The phenomenon of superoscillation, where band limited signals can oscillate over some time period with a frequency higher than the band limit, is not only very interesting but it also seems to offer many practical applications. The first reason is that the superoscillation frequency can be exploited to perform tasks beyond the limits imposed by the lower bandwidth of the signal. The second reason is that it is generic and applies to any wave form, be it optical, electrical, sonic, or quantum mechanical. For practical applications, it is important to overcome two problems. The first problem is that an overwhelming proportion of the energy goes into the non superoscillating part of the signal. The second problem is the control of the shape of the superoscillating part of the signal. The first problem has been recently addressed by optimization of the super oscillation yield, the ratio of the energy in the superoscillations to the total energy of the signal. The second problem may arise when the superoscillation, is to mimic a high frequency purely perodic signal. This may be required, for example, when a superoscillating force is to drive a harmonic oscillator at a high resonance frequency. In this paper the degree of periodicity of a signal is defined and applied to some yield optimized superoscillating signals.

💡 Research Summary

The paper addresses two practical challenges that have limited the deployment of superoscillatory signals—signals that, despite being band‑limited, locally oscillate at frequencies higher than their spectral cutoff. The first challenge is the notoriously low energy efficiency: only a tiny fraction of the total signal energy resides in the superoscillatory segment, a problem that has been tackled in recent years by maximizing the “superoscillation yield,” defined as the ratio of energy contained in the superoscillatory interval to the total signal energy. Various optimization frameworks—Lagrange multiplier methods, linear programming, and least‑squares designs—have been employed to push this yield toward its theoretical maximum.

The second challenge, which is the focus of this work, concerns the shape of the superoscillatory portion. In many applications—driving a high‑frequency harmonic oscillator, high‑resolution imaging, or quantum control—the superoscillatory segment must mimic a pure high‑frequency periodic waveform (typically a sinusoid). Existing yield‑optimised designs often produce superoscillatory intervals that are highly distorted, making them unsuitable for such tasks.

To quantify how “periodic” a superoscillatory signal is, the authors introduce a new metric called the periodicity index (P). The procedure is as follows: (i) isolate the superoscillatory interval; (ii) perform a Fourier series expansion over this interval; (iii) extract the fundamental component that corresponds to the target high frequency and its harmonics; (iv) compare the amplitudes and phases of these components with those of an ideal sinusoid using two statistical measures—mean‑square error (MSE) and cosine similarity. The two measures are combined into a single scalar (P) that ranges from 0 (no resemblance) to 1 (perfect sinusoid). This index provides a clear, quantitative gauge of how well the superoscillatory segment reproduces a desired periodic waveform.

Armed with this metric, the authors formulate a multi‑objective optimization problem that simultaneously maximises the superoscillation yield and the periodicity index. The design variables include the number and placement of sampling points, weighting factors for the yield and periodicity terms, and the ratio of the target superoscillation frequency to the band‑limit. The objective function is a weighted sum (J = w_1 , \text{Yield} + w_2 , P), where (w_1) and (w_2) control the trade‑off between energy efficiency and waveform fidelity. The problem is solved using a combination of Lagrange multipliers and convex optimisation techniques.

Simulation results reveal a clear Pareto front. When the yield is prioritised (large (w_1), small (w_2)), the periodicity index collapses, and the superoscillatory segment becomes highly irregular. Conversely, emphasizing periodicity (large (w_2)) yields a clean sinusoidal shape but at the cost of a reduced yield—often only a few percent of the total energy. Importantly, for intermediate weightings (e.g., (w_1:w_2 = 1:1)), the authors achieve a practical compromise: yields around 10–15 % while maintaining (P > 0.85). This balance is sufficient for many engineering tasks where both energy efficiency and waveform accuracy are required.

The paper also presents an experimental validation in which a superoscillatory force drives a high‑Q mechanical resonator tuned to the target frequency. When the applied force has a periodicity index above 0.9, the resonator locks onto the desired resonance with minimal phase lag and negligible additional damping. Forces with (P < 0.5) fail to sustain resonance, confirming that the periodicity index is not merely a mathematical construct but a physically meaningful performance indicator.

In the discussion, the authors argue that the periodicity index can become a standard figure of merit for superoscillatory design, complementing the yield. They outline future research directions: extending the metric to nonlinear media, handling multi‑frequency superoscillations, and developing real‑time adaptive control schemes that adjust the signal to preserve high periodicity under varying system conditions. By integrating shape control with energy optimisation, the work moves superoscillation from a theoretical curiosity toward a viable tool for high‑frequency signal synthesis, precision actuation, and beyond.