Space-based gravitational wave detectors, such as the Laser Interferometer Space Antenna (LISA), will open a new frequency window in the millihertz band. One of the most promising targets are stellar-mass compact objects inspiralling into supermassive black holes in the centers of galaxies.
These systems, known as Extreme Mass Ratio Inspirals (EMRIs), are very effective in probing strong-gravity physics and are expected to perform hundreds of thousands of revolutions in the strong field regime. Furthermore, their orbital evolution is much slower with respect to symmetric binaries, therefore it is expected that previously imperceptible phenomena, such as transient orbital resonances, will appear in the data.
In this work, we investigate three different techniques to build the adiabatic inspiral of EMRIs. One is based on the Numerical Kludge (NK) model (first-order system), whereas the other two are second-order methods. We provide the first systematic comparison between the three techniques: the agreement over the whole evolution is consistent.
We review the Effective Resonance Model (ERM), a phenomenological model that includes the effects of orbital resonances within the NK framework. We implement this model for the 3:2 resonance, that has the strongest impact on the orbital motion.
We first analyze the evolution of the integrals of motion from the start to the end of the resonance, and investigate how the flux changes individually affect the adiabatic inspiral. We find that the ERM shows a potential degeneracy: since the individual effects coming from each integral of motion may counterbalance each other, it is not clear what the net effect would be.
We then evolve a wide variety of EMRI systems and compute the orbital dephasing between orbits evolved with the ERM and orbits evolved only with the non-augmented NK model, and find an overall agreement with known results.
We then compute the gravitational waveforms predicted by the ERM and by the NK model, and find that the mismatch caused by the 3:2 orbital resonance is likely to affect the detection of the EMRIs considered in this work.
The ERM provides an efficient tool to assess at which order of magnitude resonances affect EMRI evolution and detection, across a wide parameter space. This phenomenological model could be employed in LISA data analysis to match EMRI signals over resonances.