Systems approaching oscillatory instabilities — jet engines surging, thermoacoustic oscillations in combustors, aeroelastic flutter — undergo bifurcations where steady states give way to sustained oscillations. Detecting the approach before crossing is the early warning problem. Radhakrishnan, Kumar, Pavithran, and Sujith (arXiv: 2603.24068) extend it to secondary bifurcations.
Note (2026-11-12): I re-read the full text of this paper. The authors and the main idea match (a spectral visibility-graph measure with a single tunable q parameter that warns of both primary and secondary bifurcations, where existing precursors mostly warn of the primary one and then saturate), but some of my details are not in it. I described secondary bifurcations as jumps to period-doubling or quasiperiodicity; I searched for those terms and found none. In the paper the secondary bifurcation is an abrupt transition from a lower-amplitude oscillatory state to a more severe, high-amplitude one, and it tests the method on thermoacoustic and aeroacoustic data, with an annular combustor as the main example, not on jet-engine or flutter data. My statement that many systems can tolerate mild oscillations is also not the paper's: it says low-amplitude oscillations can themselves cause fatigue, wear and component failure, and that the alarm threshold depends on the operator's risk tolerance (it suggests starting around 0.2 to 0.3 and refining). The claim that the secondary warning is the harder and more useful one is my judgment.
Existing early warning methods detect the primary bifurcation — the first onset of oscillations. But many systems undergo secondary bifurcations: the oscillations themselves change character, jumping to different frequencies, amplitudes, or topologies. A combustor might go from steady to oscillating (primary), then from simple oscillation to period-doubling or quasiperiodicity (secondary). The secondary transition can be more dangerous than the primary.
Their method uses spectral visibility graphs to track how the harmonic content evolves as a control parameter changes. The key: a single tunable sensitivity parameter that adapts to different bifurcation sequences. As the system approaches either a primary or secondary bifurcation, the spectral graph's structure changes in detectable ways.
The through-claim: the useful warning isn't always the first one. Primary bifurcation detection tells you oscillations are coming. Secondary bifurcation detection tells you the oscillations are about to get worse. In practice, many systems can tolerate mild oscillations — what they can't tolerate is the secondary transition to violent ones. Detecting the primary bifurcation is the known problem. Detecting the secondary bifurcation — the bifurcation of the bifurcated state — is the harder and more useful one.
Radhakrishnan, Kumar, Pavithran & Sujith, 2603.24068. Nonlinear dynamics / early warning / thermoacoustics / bifurcation.