Quasi-Periodic Oscillations and Timing Features

A quasi-periodic oscillation (QPO) is a localized excess in a power spectrum: more coherent than broadband noise, but broader and less stable than a clock. Its centroid, width, amplitude, harmonics, energy dependence, phase lag, and evolution through an accretion state together constrain the emitting flow. A frequency alone does not identify an orbit, precession mode, instability, radius, mass, or spin. The scientific task is therefore comparative: require each proposed mechanism to predict the same set of timing and spectral observables, then propagate selection and response effects before interpreting parameters.

Required background. Waves, Green Functions, and Geometric Optics supplies Fourier propagation and resonance concepts used to define power, phase, damping, and coherence.

Helpful background. Black-Hole Geometry and Mechanics supplies orbital and precession frequencies; Relativistic Matter and Compact Objects supplies disk and hot-flow dynamics; Black-Hole Astrophysics supplies accretion-state and inference context; and Computational Relativity supplies response-aware simulations and parameter-recovery tests.

Measuring a peak rather than naming it

For a light curve with mean count rate $\bar{x}$, choose a one-sided power density on $f\geq0$ whose integral is the component’s squared fractional rms, $r^2=\mathrm{Var}(x)/\bar{x}^2$. An exactly normalized Lorentzian on that domain is

$$L(f)=\frac{r^2\Delta} {\mathcal N[(f-f_0)^2+\Delta^2]}, \qquad \mathcal N=\frac{\pi}{2}+\arctan\frac{f_0}{\Delta}, \qquad f\geq0,$$ $$Q=\frac{f_0}{2\Delta}=\frac{f_0}{\mathrm{FWHM}},$$

where $f_0$ is the centroid and $\Delta$ the half-width at half-maximum of the Lorentzian profile. Direct integration gives $\int_0^\infty L(f)\,df=r^2$. Replacing $\mathcal N$ by $\pi$ normalizes the profile over the whole real line instead; its positive-frequency area would be only $r^2\mathcal N/\pi$. The difference is small for $f_0\gg\Delta$, but reaches a factor of two at $f_0=0$. A two-sided spectrum of a real light curve must also include its negative-frequency mirror. Alternative parameterizations can quote a characteristic frequency rather than $f_0$, so preserve both the normalization and frequency definition when comparing fits. A fitted $Q$ describes spectral concentration; it does not prove a continuously phase-coherent oscillator.

A credible detection records the time resolution, segment length, frequency binning, dead-time and Poisson-noise treatment, energy band, background, number of averaged spectra, broadband-noise model, searched frequency range, and trials correction. It also asks whether the feature persists under alternative segmentations and noise continua. Cross-spectra add phase lags and intrinsic coherence, while phase-resolved spectroscopy and harmonic amplitudes test the waveform. Uttley and collaborators give the relevant power, cross-spectral, covariance, and uncertainty machinery in Uttley et al. 2014, §§ 2.1.3, 2.2.1–2.2.3, and 2.3.

Relativistic frequencies are hypotheses, not labels

For a prograde equatorial circular geodesic around a Kerr black hole,

$$\nu_\phi=\frac{c^3}{2\pi GM}\frac{1}{r^{3/2}+a_*},$$

where $r$ is in units of $GM/c^2$ and $a_*$ is the dimensionless spin. Radial and vertical epicyclic frequencies give the periastron-precession and nodal-precession combinations $\nu_{\rm per}=\nu_\phi-\nu_r$ and $\nu_{\rm nod}=\nu_\phi-\nu_\theta$. These frequencies scale as $M^{-1}$ at fixed dimensionless radius and spin. The formulas are exact for test particles; an emitting disk or thick hot flow has pressure, magnetic stress, radial extent, emissivity weighting, and damping.

The relativistic-precession model maps selected QPOs to these combinations. A related geometric model assigns a type-C low-frequency QPO to global precession of an extended inner flow, so the observed frequency is a weighted radial average rather than a test-particle value. Reynolds reviews both the promise and the theoretical identification problem in Reynolds 2021, § 3.4.1. Ingram and Motta develop the full low-frequency phenomenology, candidate mechanisms, phase lags, harmonics, and tests in Ingram and Motta 2019, §§ 2.1–2.3, 4.1–4.4, 5.1–5.6, and 6.1–6.5.

Several alternatives remain physically live. Orbital resonances predict coupled frequency ratios but must explain excitation and finite lifetime. Diskoseismic modes predict eigenfrequencies and spatial eigenfunctions, but magnetic turbulence and disk thickness can shift or damp them. Accretion–ejection instabilities couple disk structure to a coronal or jet response. Propagating fluctuations can excite broad or narrow components whose centroid follows a changing flow timescale. Nonlinear or stochastic oscillators naturally produce frequency wandering, harmonics, and amplitude–phase coupling. A useful discriminator must therefore reproduce state evolution, energy dependence, waveform, lags, and simultaneous peaks—not merely one centroid.

A named RXTE detection and its analysis boundary

The concrete example is the RXTE Proportional Counter Array observation of GX 339−4, ObsID 92085-01-02-03, in the high-soft state. Franchini, Motta, and Lodato report a type-C candidate with $f_0=10.59\pm0.18$ Hz, $Q=3.46\pm0.50$, and fractional rms $5.05\pm0.03$ per cent in Franchini et al. 2017, Table 1 and Appendix A. Their sample selection, reduction, peak search, and relativistic-precession interpretation are described in Franchini et al. 2017, §§ 2–3.

The published analysis used custom IDL software, 16-second intervals, a 2048-Hz Nyquist frequency, Leahy normalization converted to squared fractional rms, and broad plus narrow Lorentzians. Its approximate total and hard bands were PCA absolute channels 0–64 and 14–64; the integrated continuum rms used 0.1–64 Hz. Candidate peaks required single-trial significance above $3\sigma$, survival of the trials correction, and $Q\geq2$. Because PCA channel energy boundaries and detector response depend on gain epoch and active detector units, a present-day rerun must retrieve the exact standard-products and event modes, generate the matching response, and pin HEASoft and CALDB versions, screening, background and dead-time prescriptions. This page performs no new reduction or likelihood fit.

The paper’s conversion from the highest soft-state type-C frequency to spin assumes that the oscillation is nodal precession at, or extremely close to, the innermost stable circular orbit. That is a model-conditional inference, not a unique measurement: a missed higher-frequency peak, a radially extended flow, a different mode identification, or an inner edge displaced from the orbit changes the mapping. A reproduction should fit alternative broadband continua, report posterior or likelihood intervals for all Lorentzians, repeat the trials calculation, and inject synthetic peaks through the same sampling and count-rate distribution.

Results and limitations through 2026-08-18

QPOs and their correlations with source state are durable observations. The physical mode assignment remains unresolved. Competing positions are sensitive to source population, inclination, accretion state, frequency window, and the assumed parameter regime; rare high-frequency peaks are especially exposed to trials and selection effects.

Position or evidence	Result	Scope and limitation
Analytic geodesic combinations	Give precise $M^{-1}$ frequency scaling and possible simultaneous relations	A test-particle frequency does not supply an excitation, emissivity, width, or waveform
Global-precession calculations	Link type-C frequency evolution, inclination effects, phase-resolved spectra, and a changing hot-flow radius	Depend on radial surface density, warp communication, truncation, and illumination geometry
Diskoseismic, resonance, and instability simulations	Can generate preferred frequencies, harmonics, or coupled disk–corona variability	Frequency and survival depend on thickness, magnetic field, boundary conditions, and numerical regime
GX 339−4 ObsID 92085-01-02-03	Shows a 10.59-Hz, moderately narrow high-soft-state feature in a named individual object	One centroid and an assumed type do not uniquely determine radius or spin
Population and state correlations	Type A, B, and C phenomenology separates characteristic noise, frequency, rms, and transition behavior	Classification boundaries, inclination, sensitivity, and bandpass create selection limitations
Null searches and rare high-frequency peaks	Constrain duty cycle, amplitude, and simultaneous-frequency claims	Nondetection is exposure-, state-, response-, and trials-dependent rather than proof of absence

An independent check should demand a single forward model for centroid relations, widths, harmonics, energy-dependent rms and phase, waveform, and state evolution, with an explicit mass or radius scaling. It should recover injected alternatives through the actual response and sampling and show that the preferred mechanism survives noise-continuum and selection changes.

What could change the conclusion. The comparison could change if a new stable multi-frequency detection, phase-resolved spectrum, independent polarization or reverberation constraint, or population scaling discriminates among mechanisms; it could also change if a calibration revision or trials-aware reanalysis changes a named peak’s significance.

Solved diagnostic: width and coherence scale

Using the GX 339−4 central values, estimate the Lorentzian width and the exponential damping time associated with the adopted convention. How many centroid periods fit in that damping time?

Solution

From $Q=f_0/\mathrm{FWHM}$,

$$\mathrm{FWHM}=\frac{10.59\ \mathrm{Hz}}{3.46}=3.06\ \mathrm{Hz}.$$

For a damped sinusoid whose power Lorentzian has this full width, the amplitude damping time is $\tau_d=1/(\pi\,\mathrm{FWHM})$, so

$$\tau_d=\frac{1}{\pi(3.06\ \mathrm{Hz})}=0.104\ \mathrm{s}.$$

The centroid period is $P_0=1/f_0=0.0944$ s, hence $\tau_d/P_0=1.10$. This is only an equivalent coherence scale for a specified stochastic-damping model; frequency drift or a superposition of radii can broaden the peak without each emitter losing phase in one cycle. The calculation shows why $Q$ must be combined with segment-by-segment frequency evolution and phase information.

References

• Abramowicz, M. A., and Fragile, P. C. (2013). “Foundations of Black Hole Accretion Disk Theory.” Living Reviews in Relativity 16, 1. DOI.
• Franchini, A., Motta, S. E., and Lodato, G. (2017). “Constraining Black Hole Spins with Low-Frequency Quasi-Periodic Oscillations in Soft States.” Monthly Notices of the Royal Astronomical Society 467, 145–154. DOI.
• Ingram, A. R., and Motta, S. E. (2019). “A Review of Quasi-Periodic Oscillations from Black Hole X-Ray Binaries: Observation and Theory.” New Astronomy Reviews 85, 101524. DOI; Open PDF.
• Reynolds, C. S. (2021). “Observational Constraints on Black Hole Spin.” Annual Review of Astronomy and Astrophysics 59, 117–154. DOI; Open PDF.
• Uttley, P., Cackett, E. M., Fabian, A. C., Kara, E., and Wilkins, D. R. (2014). “X-Ray Reverberation around Accreting Black Holes.” Astronomy and Astrophysics Review 22, 72. DOI; Open PDF.