WorksPhysics and Astronomy
Lasting mEMRIs: Signatures of Multiple Black Hole Extreme Mass Ratio Inspirals from AGN discs
Abstract
Encounters of stellar-mass black holes (sBH) in active galactic nuclei discs naturally form mean-motion resonances of sBH pairs. These mean-motion resonances can be broken close to the central supermassive black hole when gravitational-wave emission dominates over gas torques. We derive a simple condition that depends only on the sBH mass ratio and the resonance index for which a pair of sBH in a mean-motion resonance will remain intact and can be detectable as a multiple Extreme-Mass-Ratio-Inspirals (mEMRIs) in the LISA band. We present for the first time the gravitational-wave strain signatures of a pair of co-migrating sBH in mean-motion resonance, including additional gravitational-wave harmonics associated with orbital conjunctions of the individual EMRIs. The frequency evolution and the associated chirp mass is degenerate with isolated EMRIs of different masses. However, the mEMRI mass ratio is quantised by the resonance order and is larger than the true mass ratio. The degeneracy can be broken if the true sBH mass ratio can be inferred from individual strain amplitudes. We argue that mEMRIs should be common if disc driven EMRIs are to be detected by LISA, with mEMRIs per year in the local universe (), highlighting the need to model mEMRI systems in future LISA data analysis.
INTRODUCTION
The accretion discs of active galactic nuclei (AGN) are among the most dynamically rich environments in the Universe for compact objects. These include the central supermassive black hole (SMBH) and stellar-mass black holes (sBHs), which are either captured from the surrounding nuclear star cluster (Bartos et al. 2017; Tagawa et al. 2020) or formed in situ in the disc’s self-gravitating outer regions (Stone et al. 2017; Gilbaum & Stone 2022). Gas drag drives sBHs onto nearly circular, coplanar orbits (Generozov & Perets 2023; Wang et al. 2024), after which disc torques cause them to migrate inward (Bellovary et al. 2016; Grishin et al. 2024). The dense gas also leads to sBH–sBH binary formation through dynamical encounters at low relative velocities ([46] (Rowan et al. 2023)), making AGN discs a prolific channel for gravitational-wave (GW) mergers detectable by LIGO–Virgo–KAGRA (LVK) (McKernan et al. 2020; Tagawa et al. 2020).
In the absence of migration traps, whose locations depend on the AGN disc profile and sBH masses, sBHs may migrate close enough to the central SMBH for GW emission to drive a merger within the AGN lifetime (Pan & Yang 2021; Gilbaum et al. 2025; Sun et al. 2025). The final inspiral produces an extreme mass-ratio inspiral (EMRI), potentially detectable by the Laser Interferometer Space Antenna (LISA) (Amaro-Seoane et al. 2023) or Tianqin (Lian et al. 2025). The imprint of the AGN environment on the EMRI waveform, through gas drag, nearby perturbers, and relativistic precession, offers a unique probe of disc physics and the SMBH spacetime (Barausse et al. 2007; Kocsis et al. 2011). Moreover, combining the GW-inferred luminosity distance with an electromagnetic redshift enables a standard-siren measurement of the Hubble constant (Laghi et al. 2021).
Most studies of BH dynamics in AGN discs incorporate simplified 1D prescriptions for the interactions of sBHs during encounters (e.g., Tagawa et al. 2020; Vaccaro et al. 2026, although see Rowan & Pessah 2026 for a recent hybrid 3D N-body simulations). However, N-body simulations coupled with semi-analytical prescriptions for gas torques show that differential migration naturally leads to the formation of mean motion resonances (MMRs, [46] (Secunda et al. 2019; Lott et al. 2025; Gonglewksi et al. 2026)). MMRs maintain the period ratio of two orbiting objects at a simple integer ratio, preventing further differential migration, close encounters, and binary formation. These MMRs have been shown to be important in determining the expected population properties of sBH mergers in AGN discs ([46] (Epstein-Martin et al. 2025; Gonglewksi et al. 2026)), and can remain stable until very close to the central SMBH (Peng & Chen 2023; Peng et al. 2024). A detection of a sBH pair in MMR with LISA and/or Tianqin could lead to unique observational signatures.
In this paper, we study the processes of MMR formation, co-migration in MMR and multiple EMRIs (mEMRI), sketched in Figure 1. Initially, two sBHs embedded in an AGN disc experience convergent migration, until they are captured into MMR. They then migrate in lock-step through gas torques towards the central SMBH. Eventually, the torques from GW emission overtake those from interactions with the accretion disc, and the system remains in MMR only if is below some critical value (see § 2). In the relativistic regime, differential apsidal precession drives eccentricity increase of the outer mass, leading to non-negligible eccentricity. If , the two masses will enter the LISA frequency band, being detectable as a multiple EMRI.

Figure 1. Cartoon illustrating the process of mEMRI formation in AGN disc. Left panel: convergent migration from gas torques brings the two BHs together, with the outer (more massive) massive BH migrating faster under type II migration, until they are captured into MMR. Middle panel: the MMR prevents further convergence, and the BH migrate in lock-step from the gas dominated region closer to the SMBH where GW dissipation is more dominant. Right panel: If (see Eq. (5)), the MMR remains stable in this GW dominated regime, migrating to the highly relativistic regime through GW emission.
We begin in § 2 by overviewing sBH dynamics in AGN discs and our numerical setup. We derive a simple criterion for MMR stability against differential GW-induced migration and verity it with numerical examples of MMR formation, break-up and retention in the relativistic limit. In § 3 we study the relativistic effects on the evolution of the stable mEMRI on apsidal anti-alignment, eccentricity growth and offset from exact resonance. In § 4 we analyse the GW signal from a fiducial mEMRI system, showing unique signatures. In § 5 we discuss the impact of mEMRIs on LISA data analysis, degeneracies and characteristics. In § 6, we discuss the the astrophysical implications, rates, comparison to past work, caveats and future prospects. We can expect conservatively mEMRIs per year in the local universe () if AGN disc driven EMRI production is efficient, constituting a significant fraction of the observed EMRI population. Finally, we summarise our main results in § 7.
MMR formation and breakup in AGN discs
We follow our setup in Moncrieff et al. [46], where the AGN disc model is first created with pAGN (Gangardt et al. 2024). We then set up a three body system with the 3.5PN regularised few-body code TSUNAMI (Trani & Spera 2023; Trani et al. 2024), with a central SMBH and two sBH. Each sBH is assigned a local gas torque and eccentricity damping corresponding to the AGN model. The radial torques include type I (Ward 1997), type II (Kanagawa et al. 2018), and thermal (Masset 2017; Grishin et al. 2024; Gilbaum et al. 2025) torques. In particular, type II migration can alter the location (or even the existence) of migration traps, where the radial net torque changes sign (see Moncrieff et al. [46], for details).
MMR Formation: As a first example, we set up a Sirko-Goodman AGN disc model (Sirko & Goodman 2003) with , accretion rate , and -viscosity parameter . The smaller mass orbits at a stable orbit at a migration trap near (). The heavier sBH of migrates inward. The top left panel of Figure 2 shows the radial decay of both sBH and the capture a MMR.

Figure 2. Formation of MMR in the AGN disc. We initialise two sBH of mass at distance and at in an AGN disc model of SMBH mass , Eddington ratio and -viscosity . We see that is initially in a migration trap, until it is captured into a series of MMRs with the inward migrating (gap opening) mass , resulting in the two sBHs co-migrating. The bottom left panel shows a zoomed in view of the resonance over six years, including the resonance angle and the difference in arguments of pericentre .
MMR are generally quantified by the condition , where is the resonance index and is the resonance order. Usually the first order resonances are the strongest, which is evident in the middle left panel of Figure 2, where the period ratios move between several first order resonances (and one second order) until converging to the resonance (top right). Capture into MMR can be shown more carefully by considering the resonance angle , defined by:
where is the mean longitude, and is an effective two body longitude of pericentre (e.g., Batygin & Morbidelli 2013), given by approximately (Deck & Batygin 2015; Epstein-Martin et al. 2025):
where is the longitude of pericentre, and is the eccentricity of body .
We see the librating resonance angle in the bottom left panel of Figure 2. Both BH orbits are anti-aligned, where the difference in their longitude of pericentre is near , which is the most stable configuration.
Besides the orbital periods, the additional characteristic timescales for the MMR are the synodic period and the libration timescale. The synodic period is the time between conjunctions and is associated with rapid changes in the orbital elements due to the sBH pair interaction. The time for the resonance angle to librate is the libration timescale Batygin (2015); Moncrieff et al. [46]. The ratio of the two timescales
II. The time for the resonance angle to li- brate is the libration timescale Batygin (2015); Moncrieff et al. [46]. The ratio of the two timescales is , which is correct to order unity when compared to the simulation in Figure 2.
MMR Breakup: It was shown in Moncrieff et al. [46] that resonance capture of convergently migrating sBH in AGN discs can be analytically quantified by the adiabatic criteria (e.g., Batygin (2015)), however, less attention was devoted to MMR breaking. The proximity to the SMBH leads to a transition where the migration is governed by GW dissipation, and gas torques become negligible3. This leads to a semi-major axis evolution , where the GW timescale is given by Peters (1964):
The relative migration rate is then:
Migration will become divergent when or when , leading to . For a condition at exact resonance , divergent migration occurs if
For an illustrative example, we take a disc model of , and . We normalise the distances with units of the gravitational radius . We run two simulations where in both the lighter sBH of is initialised at . The more massive sBH is set just outside the resonance at . The mass of is either or . In this case, , so the transitional mass for divergent migration would be . The simulation with is expected to breakup once the pair enter the GW dominated regime, while the simulation with is expected to remain in resonance until very close to the SMBH.
Figure 3 shows the results of these simulations. The left panel shows the simulation for , or . We see that that around the MMR breaks, as evident from the deviation of the period ratio from the commensurability (middle panel). The resonant angle is plotted at two snapshots of time, at yrs with a small libration, and at yr with full circulation. On the other hand, the right panel shows the simulation for , or . The resonance pertains up to very close to the SMBH. The eventual breakup can occur due to differential apsidal precession near if is low enough, or due to MMR overlap near for high . We study these conditions in detail in an upcoming paper.

Figure 3. The evolution of two near identical MMR systems into the gravitational wave dominated regime. The left column system has , the right . The left system goes from in resonance (librating) to out of resonance (circulating) once GW torques become strong enough (well outside the EMRI band). Meanwhile, the right system remains librating well into the GW dominated regime (mEMRI).
Evolution in the relativistic regime
Consider the MMR system that remains stable in the GW dominated regime. As the system gets closer towards the SMBH, relativistic effects begin to dominate the evolution of the MMR. At first post-Newtonian order, the longitude of pericenter precesses at rate:
Between orbital conjunctions, the orbits therefore differentially precess by an amount:
where at exact resonance, and .
This differential apsidal precession induced by 1PN terms acts as a secular forcing on the MMR system, modifying the orbital parameters from a standard Newtonian MMR. We observe this numerically in Figure 4, which shows the evolution of the orbital angles and eccentricity of the MMR. The three columns show the resonance angles and eccentricities at decreasing distances from the central supermassive black hole, .

Figure 4. Evolution of the stable MMR system with , in the relativistic regime. Each column shows the resonance angle , the difference in argument of pericenter between bodies , and the eccentricities over a short timescale (five synodic periods) as the system migrates closer toward the SMBH. Notice that the GR induced differential apsidal precession increase between synodic periods. This leads to a repartition of eccentricity to the outer body , with as the system approaches the SMBH, while increases.
We see through the libration of the resonance angle that the system remains in resonance at all radii (including the highly relativistic region at ), and becoming dominated by the 1PN precession between conjunctions. The resonance is maintained by the orbital interaction between near conjunction, where gravitational forces leading up to conjunction anti-align the arguments of pericentre to , and is less than after conjunction. The balance of during conjunction and from differential apsidal precession between conjunctions cancel over a synodic period, leading to a stable MMR about .
The asymmetry between angular momentum exchange between before and after conjunction leads to a small net boost in the outer eccentricity , and a decrease in eccentricity , which becomes significant over the EMRI migration timescale. This can be observed in the bottom row of Figure 4, showing the bodies individual eccentricities, and combined effective two body eccentricity (Tamayo & Hadden 2025), at various radii.
A derivation of the outer body eccentricity growth via Hamiltonian perturbation theory will be presented in an upcoming paper [46 in prep]. Here we simply note that this leads to an observable EMRI eccentricity for the outer body (see the next section), which distinguishes mEMRIs from isolated disc driven “wet” EMRIs (Sun et al. 2025).
Stability: The resonance angle librates in a stable MMR. In the limit where (which is the case for the relativistic regime), , and we reduce the resonance to an exterior one. Setting gives a condition for a stable MMR. This condition becomes more transparent when viewing (in the low eccentricity limit) the periapsis advance as a difference of radial and azimuthal frequency (Yang et al. 2019; Sun et al. 2025):
where we have introduced the radial (or Keplerian) frequency and azimuthal frequency (to 1PN order, e.g., Damour & Deruelle 1985), and is the orbital distance to the central SMBH. The change in the mean longitude is given by the azimuthal frequency. is the usual Keplerian frequency.
MMR offset: Averaging over a synodic period, a stable MMR requires:
Writing this relation in terms of the synodic frequency, , and using Eq. (8) to write , we obtain the relation:
Similar reasoning for the resonance angle gives the relation:
This replaces the Newtonian notion of MMRs, where the orbital periods/frequencies librate about exact period ratios. Instead, at 1PN order, the MMR interaction forces the radial frequencies (which due to periapsis precession are not the same as orbital frequencies) to librate about an integer ratio . The orbital period ratio, due to 1PN precession. Defining the MMR offset , we observe this offset increases as the outer MMR semi-major axis decreases - as shown in Figure 5. The offset can be predicted analytically by assuming the MMR coupling leads to both and precessing at the angular frequency , i.e:

Figure 5. Evolution of MMR period ratio as the 1PN differential apsidal precession between and increases closer to the SMBH. We see that the system is the same as in Figure 4. We see that the average offset increases closer to the central SMBH, matching the analytic prediction very well.
Note that we have assumed that the masses – are no longer precessing freely, but is forced to precess at the same rate as the more massive . We can then compute:
where we have used and . This formula, as shown in Figure 5, agrees very well with the simulated MMR offset, . We will see in the next section that these relations manifest in the gravitational wave strain of EMRIs in MMR.
mEMRI gravitational wave signals
We consider the gravitational waveform for two EMRIs in the LISA band at the same time, migrating via GW emission while locked in a MMR. We obtain the gravitational wave strain and for each of the sBH and , utilising the quadrupole formula for GW radiation from a time-varying mass quadrupole (Maggiore 2007). The total acceleration entering the quadrupole formula is computed directly within TSUNAMI, following its recent implementation of the corresponding waveforms (Grishin et al. 2026). We then compute the spectral properties and detectability of the GW strain. For all gravitational wave strain computations we assume a face-on observer at a distance Mpc.
mEMRI signal observability
In order to determine the observability of the mEMRI signal, we take a representative simulation consisting of a inner BH4 in a MMR with a BH orbiting a supermassive black hole of mass . For this mass ratio, the MMR is expected to survive the GW-dominated migration and lead to a mEMRI. The additional AGN disc parameters are , , but since gas torques in this regime are negligible compared to the GW back-reaction, it will not influence our results.
We collect the gravitational wave strain over an observing time years, starting with . The polarisation strain from each mass is shown in a 500 second time segment on the left hand panel of Figure 6. To determine the observability of this mEMRI system, we take the one-sided 1D discrete Fourier transform of the strains (as implemented in numpy.fft.rfft):

Figure 6. The gravitational wave strain of a mEMRI system with in a MMR situated from a central SMBH of mass . The left hand figure shows the polarisation gravitational wave strain from the orbit of and over a 500 second period. The right plot shows the characteristic strain for the total gravitational wave strain from the mEMRI system integrated over a time frame of years. Note that both the strain from and (shown in blue and red respectively) are well above the LISA sensitivity curve.
where are the uniformly sampled times at which the strain is computed, and is the time-step between samples. is a Hann window implemented to reduce spectral leakage and prevent unphysical strain-frequency peaks of the Discrete Fourier transform. The detectability of a given signal is determined by comparing the resulting characteristic strain:
with , where is the sky and polarisation averaged one-sided effective noise spectral density of the LISA detector - which we approximate from the analytic expression given by Robson et al. (2019). This strain is shown for our system on the right hand panel of Figure 6, where we see that this mEMRI is clearly observable by LISA. To characterise this further, we can compute the signal to noise ratio:
Using only the strain from we get a SNR of 116, while the strain has a SNR of 320. This makes each individual EMRI detectable given a correct waveform for the EMRI systems. The total SNR if both and strains are used is 341.
We have made several arbitrary choices in our representative setup that can be varied to increase/decrease the SNR of a mEMRI system. SNR scales strongly with observation window , observation distance , and inclination angle , with (Robson et al. 2019):
Further analysis of mEMRI detectability is beyond the scope of this paper, and from here on we will focus on the qualitative aspects of mEMRI waveforms that differ from isolated EMRIs.
Fundamental frequencies of a mEMRI signal
Consider two stellar mass BH orbiting a SMBH. Several fundamental frequencies are expected in the strain-frequency diagram. For isolated systems with low eccentricity, the fundamental azimuthal frequencies with the strongest peaks are for , corresponding to varying quadrupole moment of over one orbital period. For isolated eccentric orbits, additional peaks will appear at where (e.g., Drasco & Hughes 2006). These additional eccentric lines are a result of the epicyclic motion of about the guiding orbital centre, and are suppressed by an order of .
For a system in MMR with index , we expect additional frequency peaks for , where (see Eqs. (10) and (11)) is the synodic frequency. These result from the small change in the orbital parameters of and during an orbital conjunction.
We see that these fundamental frequencies associated with EMRIs via the GW strain in frequency space. Consider again the MMR from Figure 6, but reduce the observation window from years to days. We choose this short integration time such that there is negligable frequency evolution of the mEMRI system. We compare the total strain in frequency space for the mEMRI system with the strain of isolated EMRIs with identical conditions. The resulting line spectra are compared in Figure 7, where the top panel shows the total strain from the mEMRI simulation, while the bottom panel shows in blue (red) the spectra of an isolated EMRI system of mass () initialised around with the same orbital parameters of () in the mEMRI simulation.

Figure 7. A comparison of the line profile (i.e. short time interval) of the EMRI strain spectra from a mEMRI system (top panel, and in MMR) and reloaded individual EMRI systems with identical initial conditions (bottom, no interaction between and ). Vertical dashed lines are shown relating the peaks in the spectra relative to the dominant harmonic , which, in the mEMRI case, are spaced by integer multiples of the synodic frequency .
For the isolated EMRI simulations, we compare the GW strain of the EMRI simulations with the characteristic orbital frequencies, , computed from the N-body orbits (see appendix section A3 for details). As expected, the dominant peaks in coincide with the eccentric frequencies. Note that in this frequency range only the lines are present, and the lines are a factor weaker than the corresponding peak.
The mEMRI strain contains additional, smaller peaks at integer multiples of the synodic frequency, , where , as expected. Note that additional fundamental frequencies in the mEMRI figure are spaced by integer multiples of the synodic frequency.
Eccentric and synodic offset: Interestingly, the lines of the isolated strain are offset slightly from the synodic lines of the mEMRI of Figure 7, while their amplitudes match almost exactly. This offset can be predicted analytically by computing:
where we’ve assumed that the system is in MMR and used Eqs. (10)–(12) to relate the frequencies to the synodic frequency. The difference will be small when . For our MMR, and the -values are for . For , , plugging in gives , which is in excellent agreement with the value obtained from the numerical simulation in Figure 7.
Impact on EMRI evolution
The difference between the gravitational waveforms between isolated EMRIs and mEMRIs becomes stark when we consider the evolution of the orbital frequency peaks. When isolated, the frequency evolution of a (nearly circular) EMRI follows from Peters (1964):
This formula is derived by setting , where is the Keplerian orbital energy, and is the orbit averaged rate of energy loss from gravitational wave emission in a Keplerian orbit:
where we have approximated and defined for small eccentricity.
When locked in MMR, the ratio of the orbital frequencies must remain constant (up to a small correction from GR, discussed in § 3), meaning the frequency evolution of the masses and instead evolve under the combined gravitational wave energy loss of both masses. Setting constant, and
solving , we get a frequency evolution of the inner EMRI:
When , the frequency evolution of the orbital gravitational wave of the inner body will increase by a factor compared to its evolution in isolation.
We can see this difference in the frequency evolution in the isolated EMRI vs mEMRI case in Figure 8. In this figure, we compare the GW spectrogram of the mEMRI system (integrated for 5 years, with ) with the spectrogram for the reloaded individual EMRI systems. This is similar to the setup considered for Figure 6, but we choose a longer integration time frame and a larger initial radius to observe a greater frequency evolution. This spectrogram is computed using scipy.signal.spectrogram5

Figure 8. mEMRI frequency evolution (right panel) vs frequency evolution for two individual EMRIs with identical initialised conditions (but not interacting). This system is identical to that shown in Figure 6, except it is initialised with and migrates to over years. Notice the vastly different chirp of as a lone source in the first panel, with a frequency change much less than the isolated simulation (second panel). In contrast, in the mEMRI migrates in lockstep with the larger . Notice also that the combined mEMRI system has faint GW peaks at multiples of the synodic frequency, which are not present in the non-interacting single EMRI systems.
The right panel shows the spectrogram of the total GW strain from the mEMRI integration, showing a red dashed line over the dominant harmonic of the body , and the line in dashed blue. Notice that the two dominant frequency evolutions of and evolve in lock-step over the five year time-scale, mirroring the joint migration of the MMR pair. Additional frequency lines corresponding to the synodic period modes are also visible. The two left panels show the frequency evolution of the isolated EMRI systems. Note that while the dominant trajectory has a relatively similar frequency evolution as its corresponding mEMRI track, the system dramatically differs from the isolated case (as expected from Eq. (21), since ). In addition, the synodic lines are absent from the isolated simulations, as expected.
IMPLICATIONS FOR LISA DATA ANALYSIS
Degeneracies and environmental effects
Environmental effects associated with orbital dephasing from gas torques (Barausse et al. 2007; Kocsis et al. 2011; Cole et al. 2023), transient resonances near Kerr SMBHs (Flanagan & Hinderer 2012; Bonga et al. 2019; Gupta et al. 2021) and/or stochastic migration (Garg et al. 2026) can impact EMRI waveforms. Understanding dephasing effects from an external environment is crucial due to the extreme precision required for accurate tests of GR with EMRIs (LISA Consortium Waveform Working Group et al. 2025; Arun et al. 2022). Here we estimate for the first time the environmental effects on EMRI waveforms associated with the presence of two bodies in MMR. Rewriting Eq. (21) in terms of the gravitational wave frequency , we get:
where we have defined the effective mass:
and the effective chirp mass . The dimensionless constant is . Similarly, , giving:
where and .
We see that, to first order, there is a degeneracy between an mEMRI pair with effective masses , and isolated EMRIs with the same effective masses. Given Eq. (5), we know that is required for the MMR to remain stable in the GW dominated regime. Using , and dividing Eq. (23) by , we get the inequality:
showing that the effective mass of the inner sBH when in MMR. A similar estimate will lead to the inequality . Thus, if the phase evolution alone is used to determine the mass of an inner (outer) EMRI in MMR, the inferred mass () will be systematically less than (greater than) the true mass (). Interestingly, the observed mass ratio will have a discrete set of values for a mEMRI event (up to a small shift from relativistic effects, Eq. (13)).
Identifying and characterising mEMRIs
We have shown that in a MMR pair, the source frame effective mass of the individual EMRIs will not match the true mass. The detector frame mass changes according to the redshift, with , but both EMRIs will have the same redshift. This gives a very simple method of identifying mEMRIs, namely, search for EMRIs with matching frequency evolutions , and sky localisation posteriors. Statistical association tests would compare the best fit models of isolated EMRIs with independent extrinsic parameters, to a joint mEMRI evolution model with identical parameters (orientation, distance, redshift, SMBH mass, right ascension, declination, etc).
Joint mEMRI strain evolution models could break the degeneracy between isolated EMRIs with chirp masses by incorporating the gravitational wave strain amplitude of the individual EMRI amplitudes. To leading quadrupole order, the gravitational wave strain of an EMRI is (Peters & Mathews 1963; Maggiore 2007):
which does not depend on the effective mass when two EMRIs are in MMR. This quantity can then be used to determine the mass ratio of the mEMRI system, via:
where () are the amplitudes of the peak harmonic of EMRI (). With this mass ratio, we can break the mass degeneracy, allowing the determination of the component masses assuming the two EMRIs are in MMR.
The requirement that the MMR is stable (§ 3) leads to several possible consistency checks on the combined mEMRI waveform, particularly if the eccentricity of the outer (more easily detectable) EMRI partner is measurable (e.g., eq. (10)). This should allow for more efficient searches of mEMRI partners (if there is a low SNR detection) and strong statistical tests of EMRI association (if isolated EMRI searches, e.g., Speri et al. 2026, find candidate mEMRI pairs). Without performing such tests, the presence of mEMRI pairs could exacerbate the global fit problem of LISA (Katz et al. 2025), which is already highly challenging for EMRIs (Chua & Cutler [14]).
DISCUSSION
mEMRI formation and rates
Predicting the rate of mEMRIs is challenging, owing to the substantial uncertainties in EMRI formation. The fraction of disc-driven EMRIs residing in MMRs depends sensitively on the AGN disc structure and the embedded stellar population. We provide a conservative order of magnitude estimate below. A more robust estimate requires large-scale Monte Carlo simulations across a suite of AGN-disc models (e.g., McKernan et al. 2025; Vaccaro et al. 2026; Tagawa et al. 2026), combined with prescriptions for MMR capture and breaking (Moncrieff et al. [46]).
We limit ourselves to SMBH masses below for two reasons. First, the peak GW emission must lie in the LISA sensitivity band . Taking and we get a conditions on the SMBH mass:
Second, even a modest sBH can open a gap and overcome the migration trap potential for lower mass AGN discs. In particular, the gap mass is where , is the disc aspect ratio and is the aspect ratio (Kanagawa et al. 2018). Thus, even moderate sBH can capture into a MMR and continue migrating to form EMRIs (Gilbaum et al. 2025) and/or mEMRIs.
The total number of host AGNs is the product of the occupation-corrected BH mass function (BHMF) , the fraction of active galaxies , and the local comoving volume . The SMBH number density is estimated via the Schechter-like prescription of Gallo & Sesana (2019). Integrating their Figure 2, the local occupation-corrected SMBH number density is between is . The AGN fraction in this mass range is around (Cho & Woo 2024) based on type 1 AGNs, or for the dwarf galaxies estimate of Pacucci et al. (2021). We take as a conservative estimate, since the Pacucci et al. (2021) fraction increases with higher SMBH mass. For the local volume (), the distance is Mpc and . Thus, the total number of host AGNs is
A given AGN may host around sBH (a conservative value compared to other estimates for larger , e.g. McKernan et al. 2018). For a typical disc lifetime of Myr (and migration time ) the EMRI rate per AGN is , in agreement up to a factor of a few with the averaged rate of from more detailed Fokker-Planck models of Pan & Yang (2021). Thus, the total EMRI rate in the local universe is . A fraction will be a mEMRI in MMR at the LISA band. is the fraction of sBH captured to a MMR and is the fraction of MMR pairs that remain stable until they reach the LISA band.
Capture to MMR is expected to be common in the adiabatic limit (Batygin 2015; Batygin & Petit 2023), however, the efficiency may depend on the radial location and the slope of the torques near the migration trap (Batygin et al. 2026). Nevertheless, the adiabatic limit is generally satisfied closer to the SMBH, leading to efficient MMR capture. This can be seen in Figure 12 of Moncrieff et al. [46], where sBH encounters at the inner migration trap almost always result in MMR capture. We’ll take a conservative estimate of . The MMR index cannot exceed a critical threshold where Chirikov’s resonance overlap criterion (Chirikov 1979) will lead to chaotic motion and escape from the resonance (Wisdom 1980)
The fraction of MMR that remain stable up to the LISA band is essentially satisfied by Eq. (5) and will depend on the mass ratio and . For a uniform mass-ratio distribution , about of the MMRs remain stable for . Lower values will result in smaller . We’ll conservatively take which corresponds to . The overall number of mEMRIs is then
Comparison to previous studies
Epstein-Martin et al. (2025) analytically studied MMR breaking in AGN discs, attributing differential apsidal precession as the main breaking mechanism close to the SMBH, occurring when the differential precession rate is comparable to the libration rate . In contrast, we find that MMRs remains stable in this regime. As evident in Figure 4, the pair develops a phase shift during the free precession, which is balanced by apsidal anti-alignment near conjunctions, occuring every synodic time . Hence, the amount of differential precession during the free GR precession phase does not lead to circulation, and the system finds a new equilibrium, analogous to a pendulum in an accelerating frame.
Our condition in Eq. (5) is consistent with the recent studies on MMR formation and break-up in AGN discs. Yang et al. (2019) studied the effect of 1PN precession and radiation-reaction on the stability of MMRs, and found that MMRs tend to break once gravitational wave torques overtake gas torques. However, Yang et al. (2019) seem to only consider (nearly) equal mass sBH, which will be destabilised in the gravitational wave dominated regime for all values of according to Eq. (5). Peng & Chen (2023) and Peng et al. (2024) showed that an intermediate-mass-BH and sBH with small mass ratios can survive in MMR down to . Sun et al. (2025) studies a variety of mass ranges, including a case of mass ratio at (their Fig. 4) with a stable mEMRI, and a case of nearly unity mass ratio () at which was unstable (their Fig. 6). We find all of these results consistent with our criterion in Eq. (5) and conclude that gravitational wave driven MMR breakup depends on the mass ratio and , and not on individual masses, giving a unified picture of MMR breaking observed in previous numerical studies.
Caveats and future prospects
Accretion disc prescription: We use the Sirko-Goodman model of AGN accretion discs (Sirko & Goodman 2003), which is a modification of the standard Shakura Sunyeav disc model (Shakura & Sunyaev 1973). This model breaks down in the highly relativistic regime near the central SMBH, particularly in the regime where the GW emission dominates the gaseous torques. Considerations of relativistic effects in discs have pointed to significantly amplified gaseous torques (Duque et al. 2026). However, we still expect that once the EMRIs enter the LISA band, gaseous torques will still be subdominant to the migration from GW emission.
AGN disc feedback: sBH can have their own mini-discs that will increase the capture rate and affect merger timescale (e.g., Whitehead et al. 2024). This is most important once the sBH pair is separated by about their mutual Hill sphere. For our low MMR, this is not the case and can be safely neglected. The cavity formed by the more massive sBH and/or feedback from the accretion onto the sBH can affect the total torque of the smaller sBH (e.g., Masset & Snellgrove 2001), which is not currently modelled.
Post-Newtonian approximation: The 3.5PN radiation reaction we use starts becoming inaccurate relative to Peters formula for , and the PN formalism similarly becomes inaccurate for , even for optimal order expansions (Yunes & Berti 2008). To accurately model the evolution of mEMRIs beyond this regime, self-force models are required (LISA Consortium Waveform Working Group et al. 2025). However, we expect that this close to the SMBH, the adiabatic condition will cease to be satisfied as the inner sBH crosses the ISCO. More significant to our results would be the impact of SMBH spin, which we have ignored by assuming a non-spinning SMBH and coplanar sBH orbits. Considering inclined orbits would introduce the additional frequency related to the polar angle , which would enter into the MMR angle.
Detection: In this study we have only simulated the direct gravitational wave strain for such EMRI systems, assuming detectability by comparison to the LISA sensitivity curve. Future work should test the detectability of component mEMRIs in injected LISA noise compared to individual EMRI systems.
MMR chains: Long chains of sBH in MMR would likely be less stable than two-body MMRs. The resulting dynamics could be complicated and, to our knowledge, has not been studied. Future mEMRI rate estimates should account for this possibility, as well as resonance breaking by stochastic disc torques (Secunda et al. 2019; Trani & Di Cintio 2025; Garg et al. 2026) or stellar flyby’s from the Nuclear star cluster (Epstein-Martin et al. 2025).
Summary
In this paper, we have used the few body code TSUNAMI (Trani & Spera 2023), coupled with an AGN disc model generated with pAGN (Gangardt et al. 2024) to study the capture, evolution and stability of MMR from the outer AGN disc to the relativistic regime near the central SMBH. We derived a simple analytic criterion for the stability of these MMRs and analysed the GW signals that arise from such multiple-EMRIs (mEMRIs). Our conclusions can be summarised as follows:
MMRs naturally form in AGN discs due to convergent migration. If the sBH pair overcomes the migration trap, it migrates towards the SMBH (see sketch in Figure 1). As GW emission dominates over gas torques, differential migration breaks the resonance if the inner-to-outer mass ratio exceeds the critical value that depends only on the MMR index , (Eqn. (5)). This yields a simple, unified criterion for MMR survival, encompassing previously studied cases.
For smaller mass ratios, the MMR is stable and the pair migrates as a mEMRI. The orbits will librate around the apsidal anti-alignment equilibrium point , even for induced 1PN differential apsidal precession. The outer eccentricity will increase while the inner eccentricity will decrease (see Figure 4).
The relativistic MMR enforces the epicyclic frequencies (Eqn. (11)), while the period ratio drifts from commensurability , where the offset grows closer toward the SMBH (see Eq. (13) and Figure 5).
EMRIs (and mEMRIs) can be observable in the LISA band for from . The joint EMRI GW strain spectral signal contains additional harmonics at multiples of the synodic MMR frequency , due to periodic interactions of the sBH pair during conjunctions.
The frequency evolution of a mEMRI leads to incorrect effective sBH masses and . Interestingly, the inferred mass ratio exactly matches and depends only on the resonance index (while the true mass ratio is ). This degeneracy can be broken if both EMRIs are identified and associated (sec. 5.2), and the true masses can be inferred from the strain amplitudes (Eq. (27)).
Future work should provide detailed template banks for mEMRI evolution, devise more thorough statistical tests for the isolated vs mEMRI association, give further consideration to MMR breaking in the strongly relativistic regime, and perform population studies to estimate how the fraction of disc driven EMRIs that are in a MMR scales as a function of the AGN disc properties.
Acknowledgements
We thank Marguerite Epstein-Martin, Juliette Becker, and Joseph Livesey for helpful discussions. We thank Nicholas Stone for helpful discussions, and suggesting the name “mEMRIs”. JWNM, EG and FHP acknowledge support from the OzGrav Research and Innovation Grant. JWNM is supported by funding from the Australian Government Research Training Program. EG acknowledges support from the ARC Discovery Early Career Research Award (DECRA) DE260101802. FHP is supported by a Forrest Research Foundation fellowship. AAT acknowledges support from Fondecyt Iniciación grant 11260895. The authors are grateful for computational resources provided by the OzSTAR Australian national facility at Swinburne University of Technology. Claude was used to generate Figure 1, based on initial sketches by the authors, and the output SVG files edited. Details of prompts used can be shared on request.
Impact of general relativistic effects on resonance parameters
Measuring eccentricities
There is no unique, gauge-independent, definition of orbital eccentricity in general relativity, which can cause confusion in gravitational wave data analysis (Shaikh et al. 2023). For nearly circular EMRIs, we find that eccentricity measures evaluated from instantaneous positions and velocities become dominated by relativistic corrections. This leads to resonance angles that circulate, even when the system is in MMR. At 1PN order, the Newtonian Runge-Lenz vector, (where is the orbital angular momentum vector), acquires an additional correction (Will 2019):
where is the true longitude, and
is the “geometric eccentricity”, computed from the orbital periapsis and apoapsis radii of the orbit. For , this Runge-Lenz definition therefore tracks the orbital phase rather than the pericentre.
The 2PN quasi-Keplerian radial eccentricity (Memmesheimer et al. 2004), which TSUNAMI reports, matches the geometric eccentricity to lower eccentricities. However, this measure vanishes once . Following Yang et al. (2019), we instead measure the geometric eccentricity , directly inferred from the position and velocity outputs (osculating elements will lead to large errors) as described below.
We compute the longitude as , where the and components are found from the 3.5PN corrected positions of the particle at a given time-step. We interpolate with a cubic Hermite spline built from the radial velocity , locating pericenters and apocenters as roots of the derivative. The eccentricities and semi-major axis are then reported by the interpolated periapsis and apoapsis radii, with given by Eq. (33) and .
The true anomaly then gives the mean anomaly via and Kepler’s equation (Murray & Dermott 1999). The resonance angles are built from the mean longitude , as in Eq. (1).
Eccentricity definition comparison
This geometric definition of eccentricity and orbital parameters leads to a well defined resonance angle throughout the EMRI evolution, and reduces to the Newtonian value far from the SMBH. Figure A1 compares the three measures for a MMR that migrates all the way from to . The AGN disc has parameters , with a pair of masses . For the inner sBH, all three definitions of eccentricity initially agree down to , below which becomes dominated by the 1PN correction . follows until , though it abruptly drops to zero around , even though the geometrically defined eccentricity remains finite for the whole simulation. The outer sBH starts with , so is unreliable throughout, and fails at . Both osculating element measures miss the eccentricity growth of the outer body at late times, though in certain simulations (not shown here) this outer eccentricity growth can exceed , in which case the again becomes an accurate eccentricity measure. We compare the different eccentricity measures across a broader range of orbital configurations and relativistic regimes in a forthcoming paper (Trani et al., in preparation), where we systematically investigate their domains of applicability, with particular emphasis on the conditions under which osculating-element eccentricities cease to provide a faithful description of the orbital geometry.

Figure A1. Figure showing the difference between definitions of eccentricity for a simulation with in a MMR. We see that the Runge-Lenz eccentricity becomes dominated by the 1PN correction when the geometrical eccentricity becomes comparable to , while the 1+2PN corrected quasi Keplerian radial eccentricity goes to zero much later, though it vanishes eventually. Meanwhile, the geometric eccentricity remains finite all the way to , and also agrees with both other definitions of eccentricity in the relevant range of validity. Note that the physical eccentricity is only sampled at select times (shown with circular dots) due to the dense output required to accurately compute the physical eccentricity.
Orbital frequencies
Because MMR forcing shifts the orbital frequencies from their isolated values, we measure these frequencies directly from the simulation. The azimuthal frequency is measured as the slope of a linear fit of the unwrapped true longitude (divided by ). This fit is taken over at least one libration period. The radial frequency is computed from the frequency of the peak Fourier transform amplitude of . We limit the possible frequencies to between and to ignore possible spurious radial frequencies related to gravitational interactions near conjunction. We use a Hann window function for the Fourier transform, as implemented in numpy.fft.
Data availability
The data underlying this article will be provided by the corresponding author upon reasonable request.
References
- [14]Chua A. J. K., Cutler C. J., 2022, Phys. Rev. D, 106, 124046
- [46]Moncrieff J. W. N., Grishin E., Trani A. A., Panther F. H., Pietrosanti O., 2026, MNRAS, 545, stafx2217