Massive Black-Hole Binary Signals

A massive black-hole binary can remain in a space detector’s band through a long inspiral and then contribute a short, information-rich merger and ringdown. Advance warning comes from phase evolution before coalescence; sky localization comes from time-dependent response, finite-arm structure, and waveform harmonics; remnant inference comes from the merger–ringdown waveform. None is a mission-independent number. This page derives the controlling signal features and separates robust mechanisms from configuration-, population-, waveform-, and prior-dependent forecasts.

Helpful background. Probability and Statistical Inference supplies conditioning, uncertainty propagation, and selection. The gravitational-wave likelihood supplies network inference and covariance. Gravitational Radiation and Waveforms supplies inspiral–merger–ringdown models, while Black-Hole Astrophysics supplies massive-black-hole formation, environments, and host interpretation.

The redshifted chirp

For source-frame masses $m_1$ and $m_2$, define

$$M=m_1+m_2, \qquad \eta=\frac{m_1m_2}{M^2}, \qquad \mathcal M=\eta^{3/5}M.$$

Cosmological time dilation makes the detector-frame masses

$$M^\mathrm{det}=(1+z)M^\mathrm{src}, \qquad \mathcal M^\mathrm{det}=(1+z)\mathcal M^\mathrm{src}.$$

At leading quadrupole order for a quasicircular inspiral,

$$\dot f =\frac{96}{5}\pi^{8/3} (G\mathcal M^\mathrm{det})^{5/3}f^{11/3},$$

and the remaining detector time is

$$t_c-t =\frac{5}{256} (G\mathcal M^\mathrm{det})^{-5/3} (\pi f)^{-8/3}.$$

The time formula follows by integrating $dt=df/\dot f$: $\int_f^\infty f'^{-11/3}df'=(3/8)f^{-8/3}$. The upper endpoint is a formal Newtonian extrapolation; a physical waveform instead reaches merger at a finite frequency. For scale, $\mathcal M^{\rm det}=10^6M_\odot$ at $f=0.1\,\mathrm{mHz}$ gives about 35 days, using $GM_\odot/c^3=4.9255\,\mu\mathrm s$. This is remaining inspiral time, not a promised alert lead time. Spin, eccentricity, environmental torques, higher post-Newtonian orders, and the merger transition supply corrections; a forecast propagates the full posterior.

Where advance warning comes from

For declared TDI channels $\alpha$, a forecast models

$$d_\alpha =R_\alpha h_{\mathrm{MBHB}}(\vartheta) +\sum_bR_\alpha h_b(\lambda_b) +n_\alpha ,$$

where the sum contains overlapping sources and $n_\alpha$ has the instrument-plus-confusion covariance. Advance-warning and localization posteriors require this likelihood or a documented approximation to it; an isolated waveform against a sensitivity curve omits source overlap and covariance uncertainty.

Let $\widetilde{\boldsymbol h}$ collect the signal in the declared TDI channels and $\boldsymbol C(f)$ be their cross-spectral covariance. Define cumulative optimal signal-to-noise ratio through a cutoff $f_*$ by

$$\rho^2(f_*) =4\int_{f_{\min}}^{f_*} \widetilde{\boldsymbol h}^{\,\dagger}(f) \boldsymbol C^{-1}(f) \widetilde{\boldsymbol h}(f)\,df,$$

where $\boldsymbol C$ is the one-sided instrument-plus-foreground PSD matrix. Only after a verified transformation diagonalizes it does this reduce to $4\sum_\alpha\int|\widetilde h_\alpha|^2/S_\alpha\,df$. A frequency cutoff approximates a time cutoff for a slowly evolving monotonic harmonic under the stationary-phase approximation; higher harmonics, merger, gaps, and window leakage require a time-truncated response and likelihood. A warning rule also needs a detection statistic, threshold, trials model, duty, latency, and a distribution for $t_c$. “Visible before merger” and “localized well enough for an alert” are distinct selections.

Long inspiral supplies cycles for masses and aligned spin combinations. Higher harmonics extend to different frequencies and break inclination, distance, mass-ratio, and sky degeneracies. Precession adds modulation when spins and mass ratio make it appreciable. The constellation’s orbital motion helps only to the extent sampled before the event; much of a high-mass system’s signal-to-noise ratio may arrive too late for a long modulation baseline.

Bayesian injection studies show explicitly that full time- and frequency-dependent response and higher harmonics can remove degeneracies that survive a frozen, quadrupole-only analysis Marsat et al. 2021, §§II, IV–VI. This is evidence about the injected systems and response used there, not a universal localization guarantee.

Localization is a posterior, not one angle

A sky result is a probability density over direction, usually correlated with distance, inclination, polarization, phase, masses, and spins. A useful area summary must state its credible content and sky measure. Host association additionally needs a three-dimensional localization, galaxy model, catalog completeness, and any counterpart likelihood.

Three response features carry sky information:

• annual or orbital Doppler phase depends on $\widehat{\boldsymbol k}\cdot\boldsymbol x(t)$;
• rotating link projections modulate amplitude and polarization; and
• finite-arm transfer functions break low-frequency response symmetries.

Fisher matrices can diagnose high-signal-to-noise local curvature, but they miss separated sky modes and non-Gaussian tails. Time-truncated Bayesian injection studies are therefore more direct for warning forecasts. Mangiagli and collaborators quantify this evolution for a declared source grid and mission model in Mangiagli et al. 2020, §§II–IV; their numerical areas and lead times remain conditional on that grid.

Merger, ringdown, and the remnant

An inspiral–merger–ringdown likelihood can infer detector-frame final mass $M_f^\mathrm{det}$ and dimensionless spin $a_f$ through the full waveform or through quasinormal frequencies,

$$\omega_{\ell mn}^\mathrm{det} =\frac{1}{1+z} \omega_{\ell mn}^\mathrm{src}(M_f^\mathrm{src},a_f).$$

The source-frame mass requires redshift information or a cosmological model in addition to gravitational-wave luminosity distance. Ringdown mode amplitudes depend on mass ratio, spins, orientation, waveform frame, and start time. A precise full-signal remnant posterior is not automatically an independent spectroscopic resolution of multiple modes.

Consistency can be tested by comparing a remnant predicted from the inspiral with one inferred from later data, but the split, covariance, waveform family, and response must be fixed. Calibration, residual laser noise, overlapping sources, and inaccurate high-frequency transfer functions can mimic an inconsistency.

Results and interpretations at the stated cutoff

All entries below are forecasts or simulation results as of 2026-08-18; no LISA flight observation is being summarized.

Question	Strongly supported mechanism	Conditional interpretation and limiting test
Advance warning	Inspiral phase fixes a posterior for $t_c$ before merger	Lead time depends on threshold, source, noise, gaps, waveform, and alert latency; rerun time-truncated injections
Sky localization	Motion, link projection, finite-arm response, and higher harmonics break degeneracies	Fisher ellipses can miss multimodality; compare full Bayesian recovery with frozen-response and quadrupole-only controls
Remnant measurement	Merger and ringdown encode detector-frame final mass and spin	Source-frame values require redshift; vary waveform, start time, modes, calibration, and foreground
Host or counterpart	A shrinking 3D posterior can prioritize follow-up	Unique association is not guaranteed; inject incomplete galaxy catalogs and unrelated transients

The ESA 2024 definition study supplies the reference mission and source-performance framework ESA 2024, §§2–4. Earlier and current forecasts use different source populations, mission assumptions, waveforms, and selection rules; variation between them measures conditionality rather than a disagreement about the basic chirp physics.

Reproducing the result and testing its robustness

The shared public simulation anchor is LISA Data Challenge 2b “Spritz,” which supplies simulated second-generation $X,Y,Z$ fractional-frequency channels at 5 s cadence with a declared orbit, instrument noises, gaps, artifacts, and unresolved Galactic foreground LISA Data Challenge 2b, dataset description. A massive-binary study must identify the exact challenge file and injection truth used, or state that a new catalog was injected into a selected challenge noise interval. It remains a simulation.

A reproducible forecast records the ESA configuration, analytic orbit or orbit file, link and TDI conventions, observing duration and duty model, cadence and band, instrumental and confusion covariance, population and cosmology, source-frame drawing measure, waveform and response commits, calibration perturbations, threshold, alert rule, priors, sampler settings, seeds, injection count, failures, and credible-area definition. Report distributions over sources and noise, not only a favorable orientation.

Independent checks recover the Newtonian time-to-merger scaling, inject the same system with full and simplified response, remove higher harmonics, change foreground and waveform families, and compare repeated-noise Bayesian posteriors with Fisher predictions. The conclusion should be refreshed after a superseding ESA configuration, a new public challenge, or a validated waveform or global-fit result that changes localization or remnant recovery.

Check your understanding

At fixed observed frequency, how does the leading time to coalescence change if $\mathcal M^\mathrm{det}$ doubles? At fixed chirp mass, how does it change if $f$ doubles?

Solution

The leading expression scales as

$$t_c-t \propto(\mathcal M^\mathrm{det})^{-5/3}f^{-8/3}.$$

Doubling chirp mass multiplies the remaining time by

$$2^{-5/3}\simeq0.315.$$

Doubling frequency multiplies it by

$$2^{-8/3}\simeq0.157.$$

These are local leading-order comparisons. A real warning posterior also carries correlations, higher-order phasing, response, threshold, and data-gap uncertainty.

References

• European Space Agency. LISA Definition Study Report. ESA-SCI-DIR-RP-002 (2024). Official PDF.
• LISA Data Challenge. “Challenge 2b: Spritz.” Public simulated dataset and documentation. Dataset.
• Mangiagli, Alberto, et al. “Observing the Inspiral of Coalescing Massive Black Hole Binaries with LISA in the Era of Multimessenger Astrophysics.” Physical Review D 102 (2020): 084056. DOI. Open PDF.
• Marsat, Sylvain, John G. Baker, and Tito Dal Canton. “Exploring the Bayesian Parameter Estimation of Binary Black Holes with LISA.” Physical Review D 103 (2021): 083011. DOI. Open PDF.