We present three-dimensional general relativistic magnetohydrodynamic (GRMHD) simulations of supermassive binary black hole mergers embedded in gaseous environments, with a focus on the late inspiral phase. Our models self-consistently evolve the plasma dynamics, enabling a direct connection between the gravitational wave signal and the electromagnetic response. In particular, we investigate the impact of key physical parameters, such as black hole spin precession, on the structure of the accretion flow and the evolution of magnetic fields.

We find that spin precession induces strong variability in the Poynting-dominated outflows of accreting binary black hole systems compared to non-precessing configurations. This variability imprints distinct, time-dependent electromagnetic signatures that are correlated with the binary’s orbital dynamics. Our results highlight the importance of GRMHD simulations in establishing a robust connection between gravitational wave signals detectable by LISA and their electromagnetic counterparts.

Supermassive binary black holes are expected to be important evolutionary stages of galactic nuclei, however, they have remained elusive. While there are hints of these systems in the gravitational wave background as measured by pulsar timing arrays (PTAs), confirmation of individual sources are still years away with either PTAs or the ESA/NASA LISA mission. Through 3-d general relativistic magnetohydrodynamics (GRMHD) simulations, our group aims to learn how accretion may produce electromagnetic signatures with which we may search for candidates and how it may affect the orbital evolution of the binaries. In this talk, we will present recent results exploring these two aspects. We will share results of merging binaries spanning the post-Newtonian (inspiral) to dynamic GR (merger to post-merger) regimes for the first time with relaxed initial conditions for the gas. We find that the Poynting flux and thermal luminosity change over the course of merger process for different spin configurations to varying degrees.

Next, we will communicate new results on long-term GRMHD simulations of the circumbinary disk using a range of radiative loss rates. We find that hotter disks lead to greater rates of increasing binary separation but no significant affect on the rate of eccentricity growth. By using a different, more realistic model of radiative losses, using the so-called leakage method, we will show how one models radiative losses impacts the structure of the circumbinary disk and its accretion variability.

We present the first 3D general relativistic magnetohydrodynamic simulation that self-consistently evolves both the spacetime and magnetized plasma of a misaligned supermassive binary black hole system embedded in an equilibrated circumbinary disk, from late inspiral through merger and recoil. After 165 orbits of disk relaxation, we follow the final 40 orbits to merger. We find that the minidisks around each black hole become strongly warped, with jets aligned with the individual black hole spins near the holes and with the binary angular momentum on larger scales. Following merger, the remnant receives a recoil kick exceeding 1000 km/s while retaining its bound circumbinary disk. Because most of the luminosity originates near the black hole, the recoiling remnant disk remains the dominant post-merger emission source and may be detectable for hours in LISA sources.

The recent detection of a gravitational wave (GW) background by Pulsar Timing Arrays (PTAs) provides the first evidence for a population of inspiraling massive black hole binaries (MBHBs). Combining individual GW signals from PTAs and the forthcoming Laser Interferometer Space Antenna (LISA) with their electromagnetic (EM) counterparts is crucial for testing theories of gravity and constraining the universe’s expansion rate. Therefore, theoretical models predicting these EM signatures are needed to pinpoint the source locations, overcoming the poor sky localization of gravitational wave detectors. Calculating these EM signatures from first principles is challenging, requiring a detailed understanding of the gas, radiation, and magnetic fields within the accretion flows around MBHBs. This research directly addresses this challenge by performing global three-dimensional radiation magnetohydrodynamic simulations of the accretion flows around MBHBs to calculate their EM signatures.

Supermassive black hole binaries embedded in gas are among the most compelling multimessenger targets for LISA. In the late inspiral, the binary can leave an imprint not only on the gravitational-wave signal, but also on the surrounding plasma through shocks, accretion-flow disruptions, and large-scale jet variability. Small-mass-ratio systems may be especially important in this regard, since the secondary can act as a strong perturber of the gas while the primary continues to power relativistic outflows.

In this talk, I will discuss new general relativistic magnetohydrodynamic simulations of a gas-rich, small-mass-ratio (q=1:7) supermassive black hole binary during the final stages before merger. I will focus on how the inspiraling secondary drives shocks in the circumbinary flow and disrupts the accretion stream around the primary as the binary separation decreases. These effects produce strong small-scale asymmetries and rapidly varying inner-disk dynamics that may give rise to distinctive electromagnetic activity in the immediate pre-merger regime.

I will also describe how these same simulations reveal a clear large-scale jet precession on scales of thousands of gravitational radii. The jet is predominantly powered by the rapidly spinning primary black hole, whose precessing spin axis links the binary dynamics directly to an observable large-scale outflow signature. Together, these results suggest that late-inspiral, gas-rich supermassive black hole binaries can generate coupled small-scale variability and large-scale jet modulation, providing promising electromagnetic tracers of LISA-band sources.

Expectations for the detection of gravitational radiation from massive black hole binaries have been raised by the detection of the gravitational wave background by the Pulsar Timing Arrays consistent with massive binaries, and by the selection of the Laser Interferometer Space Antenna (LISA) for a large-class mission in the European Space Agency science program. In light of these developments, the rates at which massive binaries form and evolve to coalescence remain important open questions in black hole astrophysics. I will discuss how theoretical modeling helps us address these questions and prepare for the upcoming multi-messenger detections of merging binaries with LISA and contemporary electromagnetic observatories.

Gravitational waveform models which are both highly accurate and efficient are crucial for future gravitational-wave detectors such as LISA. Towards this necessity, the Spinning Effective-to-Backwards One Body (SEBOB) formalism combines the effective-one-body (EOB) formalism of the SEOBNRv5 inspiral model with the analytical Backwards-One-Body (BOB) merger-ringdown model to yield fast and accurate waveforms for the dominant (2,2) mode. We now extend the SEBOB framework by modelling additional, higher-order modes with BOB, and we test its accuracy against numerical relativity (NR) simulations from the SXS collaboration. We show that the extended BOB model is robust, providing an accurate description of the merger-ringdown across several modes. Lastly, we show that the extended SEBOB model can be applied to construct displacement memory in excellent agreement with the (2,0) mode memory found in SXS Ext-CCE catalog simulations.

While almost all current gravitational-wave observations involve comparable-mass binary black holes (mass ratio q 2), LISA will be sensitive to populations of massive black hole binaries, intermediate mass-ratio inspirals, and extreme mass-ratio inspirals, spanning more than six orders of magnitude in mass ratio. However, no current waveform model is calibrated to precessing numerical-relativity simulations beyond q ≈ 8, and the accuracy of approximate semi-analytic models at high mass ratios remains unknown. Here, we study the accuracy of current models against a new set of numerical-relativity simulations of highly precessing binaries with a mass ratio of 18. We find that, in all cases, current models incur significant biases, with biases in mass measurement reaching up to 100%. The prohibitive computational cost of numerical-relativity simulations at higher mass ratios precludes the construction of accurate models in the near future, presenting a major challenge for a range of LISA science goals.

LISA data analysis will not and should not fully rely on numerical relativity surrogate models and other models calibrated to numerical relativity catalogs, since that risks letting subtle systematic errors go unnoticed when extreme accuracy is required, for example when trying to test GR with subdominant modes. To this end, we will present the latest update on our Spinning Effective-to-Backwards One Body (SEBOB) waveform model, which minimizes reliance on numerical relativity, and now includes generically precessing spins and higher harmonics through l=10.

A wide range of theoretical and astrophysical problems—including modeling gravitational waves from extreme mass ratio inspirals (EMRIs)—require the accurate numerical solution of the time-domain Teukolsky equation for spin weight s=−2 in Boyer–Lindquist coordinates. However, long-time evolution of this equation is notoriously hampered by unphysical reflections from the outer boundary and by slowly growing spurious modes that contaminate the physical waveform. We develop and implement exact radiation outer boundary conditions (ROBCs) for the Bardeen–Press equation (the a=0 limit of the Teukolsky equation), which make the outer boundary fully transparent without the need for coordinate compactification or auxiliary transformations. In addition, we construct near-field-to-far-field teleportation kernels that allow the waveform to be evaluated directly at future null infinity from data recorded at a finite radius. Our results demonstrate that these boundary treatments eliminate unphysical late-time growth, yield the correct asymptotic decay rates, and enable efficient, long-duration time-domain simulations relevant to EMRI waveform modeling and other black hole perturbation problems.

Gravitational memory is a universal prediction of general relativity arising from the infrared structure of gravity, linked to asymptotic symmetries and soft theorems. It corresponds to a permanent deformation of spacetime left behind after gravitational radiation has passed. Despite its theoretical significance, memory has not yet been experimentally verified, primarily because frequency-band-limited detectors are insensitive to the defining DC offset and measure only its time-dependent transition. A robust detection therefore requires precise theoretical modeling and dedicated data analysis.

LISA offers a unique opportunity to achieve this. For massive black hole mergers with signal-to-noise ratios of hundreds to thousands, the memory transition can reach SNRs of order 50 in favorable cases. Recently, a sub-group within the LISA Fundamental Physics Working Group has developed the theoretical and Bayesian analysis framework needed for a credible detection claim. I will present this effort and discuss prospects for the first empirical detection of gravitational memory with LISA, thereby opening new avenues in an era of memory-full GW data.

The past decade of gravitational wave astronomy has seen over 200 mergers of black holes and neutron stars. The detection and parameter estimation of such events is largely made possible using accurate waveform models. With ground-based detectors reaching their design sensitivities and future space-based detectors requiring waveforms of unprecedented accuracies, understanding and mitigating errors in existing waveform models has become increasingly relevant. Moreover, tests of general relativity require precision since any deviations from the theory could be attributed to errors in waveform models. In this work, we use the numerical relativity surrogate model NRSur3dq8, trained on waveforms of aligned-spin black hole binary mergers, to build a model that predicts errors in the phase and amplitude of their corresponding gravitational waveforms. The data-driven nature of surrogate modeling and the complexity of the underlying non-linear models introduce several different sources of error, and accordingly, we explore different approaches to building the error models. We also discuss the applications of surrogate waveform error quantification to parameter estimation studies.

The LISA global fit, a simultaneous fit of all sources in LISA's rich data, poses a challenging cataloging problem. This problem is most pronounced for the large number of continuous galactic binary signals which are expected to be numerous and often overlap in time-frequency. Inspired by astronomical catalogs, we discuss the purity and completeness for galactic binaries as computed from the results of the LDC2a (Sangria) LISA data challenges using the results of the Erebor global fit catalogs. We find that both purity and completeness are dependent on the part of the parameter space that we consider as well as the confidence that we require from the global fit clustering algorithm that produced the catalog.

Modeling the electromagnetic (EM) emission from supermassive binary black holes (SMBBH) during their late inspiral and merger is a key hurdle to maximizing the scientific return of multimessenger astronomy with LISA. To address this, we use general relativistic magnetohydrodynamic (GRMHD) simulations to model sub-Eddington accretion onto SMBBHs and gravitational wave emission. In this talk, I will discuss smoking-gun quasiperiodic emission that arises uniquely from eccentric SMBBHs that can explain observed SMBBH candidates. I will also describe how magnetically arrested accretion flows can form around binaries, lead to magnetic reconnection, and power novel X-ray flares. Finally, I will describe efforts for modeling Eddington accretion flows onto binary SMBBHs via next-generation GPU-accelerated radiation GRMHD. These efforts add a new dimension to SMBBH accretion and unique models for identifying SMBBH candidates on-sky and supporting multimessenger astronomy with LISA.

LISA will detect numerous merging SMBHBs, including some at high redshift, with electromagnetic counterparts for localisation and redshift determination. Gravitational-wave observations provide the luminosity distance, and combining these allows probing the Universe’s expansion at previously unexplored redshifts. The dominant uncertainty in SMBHB distances arises from weak lensing. Delensing - utilising external data to remove the lensing effect on gravitational waves - with galaxy surveys has been proposed, but low galaxy densities make this impractical. We investigate an alternative using CMB lensing, which traces the same large-scale structure as high-redshift SMBHBs. The CMB’s full-sky, deep measurements enable precise magnification estimates and mean-field subtraction, with the lensing field inferred directly, unlike galaxy-based delensing relying on second-order shear. I will discuss the promising potential of CMB delensing to enhance cosmological inference with LISA SMBHBs.

Nearly all massive galaxies host supermassive black holes (SBHs) at their centers, and according to hierarchical structure formation major mergers are an important growth channel for galaxies. It is expected that SBH binaries (SBHBs) are a natural consequence of galaxy evolution, though their demographics and coalescence timescales are largely unknown or poorly constrained. These uncertainties translate into uncertainties in the expected detection rates of LISA sources, which evolve under the same mechanisms. Here I discuss my multi-messenger investigation of the efficiency of stellar scattering in tightening SBHBs by jointly comparing models to the observed galaxy stellar core population and to results of nanohertz gravitational wave observations. I find that stellar hardening rates capable of explaining mass deficits in core-galaxies are incapable of explaining the low frequency turnover seen in the nanohertz background. Mechanisms resolving this tension are discussed, including recoiling SBHs, which constitute promising LISA observables.

Compact double white dwarfs are among the most important sources for the Laser Interferometer Space Antenna (LISA) and represent key targets for multi-messenger astrophysics. Several of these systems are already well characterized through time-domain photometry and spectroscopy, providing precise measurements of their orbital properties. The combination of electromagnetic observations with gravitational-wave data from LISA offers the potential to improve constraints on the physical parameters of these binaries. In this work, we develop a joint Bayesian framework to perform a multi-messenger analysis of eclipsing double white dwarf systems. Our approach jointly analyzes electromagnetic light curves and gravitational-wave signals within a unified parameter estimation framework. This strategy enables a direct assessment of how complementary observables contribute to constraining the same underlying binary configuration. We focus on validating the methodology using synthetic data inspired by well-studied verification binaries. This work represents a step toward future multi-messenger analyses of double white dwarf systems.

The Zwicky Transient Facility (ZTF) has been the most prolific discovery channel for LISA-detectable Galactic binaries, yet the survey's sensitivity to these systems across ranges of physical properties remains unquantified. We present the first empirical selection function for LISA Galactic binaries in ZTF, constructed by injecting synthetic light curves from population synthesis models into real photometry and measuring recovery in a GPU-accelerated periodicity search. Our forward models use full Roche-geometry light curve calculations including tidal distortion, eclipses, irradiation, and accretion disks, effects usually neglected in population forecasts but which substantially shape detectability. By evaluating recovery across a wide range of population synthesis prescriptions, we enable the first observationally-calibrated constraints on the Galactic binary population, with implications for modeling the LISA gravitational-wave foreground and the binary yield of Rubin and UVEX.

(This presentation is based on our recent work: arXiv:2506.01898)

The synergetic multiband observation of stellar-mass binary black holes (sBBHs) by LISA and next-generation ground-based detectors (ET and CE) promises unprecedented insights into astrophysics and fundamental physics. However, most sBBHs will have a signal-to-noise ratio (SNR) below 5 in the LISA band. Current multiband parameter estimation methods -- which typically use ground-based posteriors as priors for LISA -- struggle to analyze these low-SNR sources due to complex likelihood surfaces and numerous local maxima.

In this talk, we present a novel coherent multiband analysis framework that directly calculates a joint likelihood for LISA and 3G detectors. This high efficiency is achieved by employing rotation matrices to unify coordinate systems and utilizing importance sampling to marginalize over the extrinsic parameter space. Our method works robustly even when the LISA SNR is as low as 3.

By lowering the detectable threshold, our approach nearly doubles the expected number of multiband sBBH sources. Furthermore, we demonstrate that even with merely one year of LISA observations, the multiband 90% credible interval for detector-frame chirp mass reaches sub-10^-4 solar mass precision, outperforming the most accurately measured events by the ET+2CE network (with 7.5 years of observation) by at least one order of magnitude. For the first time, we showcase efficient multiband Bayesian parameter estimation results on a population scale, paving the way for large-scale astrophysical tests using multiband gravitational-wave astronomy.

The main science goals of LISA focus on understanding ultra-compact objects such as white dwarfs, neutron stars, and black holes. However, more recent studies have shown that LISA may be able to detect exoplanets orbiting its gravitational wave (GW) sources. The most likely candidates are circumbinary exoplanets orbiting double white dwarfs (DWDs). These systems may also be observable electromagnetically, meaning they could become the first multi-messenger exoplanets. We present work for the exoplanet discovery space comparing estimated detection capabilities of GW radial velocity with LISA, combined with a relatively new electromagnetic technique – infrared (IR) excess with JWST. We note that this is the first exploration of IR excess around DWDs as a viable exoplanet detection method, extending prior work demonstrating its effectiveness for planets around solitary white dwarfs. We also explore the possibility of astrometry using either Gaia or the upcoming Habitable Worlds Observatory for additional validation. We find these techniques overlap in discovery space, which could therefore enable robust confirmation and characterization.

LISA is expected to observe tens of thousands of resolved Galactic Binaries (GBs), requiring data analysis pipelines to handle physically motivated populations and perform rigorous astrophysical inference. Due to the transdimensional nature of extracting an unknown number of sources, traditional post-processing population inference methods used in LVK data analysis are unsuited. Therefore, population inference must be intrinsic to the global fit.

In this work, we present a novel astrophysical inference framework that directly operates within the global fit and performs a Bayesian model selection algorithm using Reversible Jump MCMC. Driven by the Eryn sampler’s ensemble architecture, the algorithm dynamically jumps between competing population synthesis models. These jumps alter the overarching astrophysical priors and the Poisson distributions governing the expected number of resolved systems. By restricting competing models to differ by a single binary-evolutionary hyperparameter (e.g., common envelope efficiency), we map out a Bayes factor surface. This approach rigorously quantifies and improves our understanding of LISA’s sensitivity to specific astrophysical processes and explores physical degeneracies between astrophysical hyperparameters.

To achieve this integrated inference on future data streams, global fit pipelines must be capable of analyzing realistic and complex datasets. The most recent LISA Data Challenge, Mojito, provides an ideal testbed to meet this requirement, introducing realism through time-varying noise, realistic orbits, and sophisticated populations of interacting and detached GBs. To test our population inference framework against these complexities, we present recent advancements to the GPU-accelerated global fit pipeline, Erebor. Motivated by the Mojito data challenge, we implement computationally efficient waveform and likelihood evaluations via the recently developed TDI-on-the-fly approach (Littenberg & Cornish 2025). Furthermore, to make this simultaneous population inference and parameter estimation computationally tractable, we integrate normalizing flow-based priors and proposals. These updates significantly accelerate the sampling convergence of Erebor's resolved GB block. By making this complex joint analysis computationally feasible, our work ultimately demonstrates the necessity and capabilities of fully integrated population inference in the era of realistic LISA data.

We present precise parameters for two compact double white dwarf (DWD) binaries, SDSS J232230.20 +050942.0 (J2322+0509) and SDSS J063449.92+380352.2 (J0634+3803), with orbital periods of 20 and 26.5 minutes, respectively. These systems will serve as verification sources for the Laser Interferometer Space Antenna (LISA). To significantly improve the electromagnetic (EM) constraints on these two systems and the LISA detectability predictions, we conducted spectroscopic follow-up observations using the Hubble Space Telescope/Space Telescope Imaging Spectrograph, Keck I/LRIS, and Keck II/Echellette Spectrograph and Imager. Our analysis significantly improves the temperature, surface gravity, and mass constraints for both primary and secondary components in J2322+0509, as well as dynamical properties such as radial velocities and orbital periods in both systems. For J2322+0509, we derive an updated inclination of i=25+-4.5 deg, while for J0634+3803, we obtain i= 43+-7.0 deg. We assess the detectability of these sources using LDASOFT/GLASS. Incorporating EM priors on inclination significantly enhances the gravitational-wave signal recovery, reducing uncertainties in amplitude by a factor of 2–4 and shortening the detection time by up to a few months. Our results underscore the importance of multimessenger observations in characterizing DWD binaries and maximizing LISA’s early scientific capabilities.

The galactic population of cataclysmic variables (CV) is an until recently overlooked class of sources in the mHz gravitational wave (GW) band. Scaringi et al (2023) found that the incoherent superposition of GWs from unresolved CVs will create a unique spectral feature observable by the upcoming Laser Interferometer Space Antenna (LISA) mission. We supplement the simulated population in Scaringi et al. with detached CV systems present in the EM period gap and explore the GW signatures of the entire population. We show that this feature may not be as prominent as previously thought, and assess LISA’s ability to detect and characterize the mHz CV population with the Bayesian LISA Inference Package (BLIP).

Very short-period double white dwarfs (DWDs) in our Milky Way (MW) will be one of the most numerous classes of sources detectable by the upcoming Laser Interferometer Space Antenna (LISA), with its scientific return strongly enhanced by the availability of electromagnetic (EM) counterparts. This makes the present an ideal opportunity to chart the population of (potential) LISA sources across the MW. We developed a strategy to identify LISA source candidates in multiband photometric surveys, in order to select targets for further follow-up and characterisation. Starting from a theoretical population of Galactic WDs, combined with a consistent cooling model for the evolution, we constructed a synthetic EM catalogue of WD detections. We find that the resulting magnitude and colour distributions can help us distinguish LISA source candidates from the broader white dwarf population: the former populate a specific area in colour-colour diagram, with little contamination from single systems and wide binaries. This method is now guiding the design of a proposal for the Canada-France-Hawaii Telescope Community Survey, happening over a 6-year period (2027-2033). Expanding the EM catalogue of ultra-compact DWDs is only the first step: correctly matching LISA detections to their EM counterpart will not be straightforward, as LISA’s sky localisation remains less constrained. The accuracy of such identifications will depend on the precision and completeness of the EM characterisation of the system, which must drive the planning of follow-up observations.

With gravitational wave detector LISA scheduled for launch in mid 2030, accurate waveform modeling of key gravitational wave sources such as extreme-mass-ratio inspirals (EMRIs) is increasingly important. For analytical computation of dissipative and conservative quantities, high post-Newtonian (PN) order expansions are essential for accurately capturing strong-field dynamics and achieving the phase precision required for LISA observations. In this work, we utilize such high PN order fluxes derived previously from black hole perturbation theory. We consider spherical inclined EMRI orbits with spinning primary black holes and compute adiabatic inspiral waveforms. We validate our analytical waveforms by comparing to those computed from fluxes based on numerical solutions of Teukolsky equations. Using dephasing and waveform-mismatch, such comparisons assess the accuracy and applicability of PN approximation in regions relevant for future LISA observations.

Extreme mass-ratio inspirals (EMRIs) will enable LISA to map strong-field gravity through long, information-rich signals whose phase and amplitude encode the background spacetime. Their multi-harmonic content, secular evolution, and characteristic modulations make accurate inference computationally demanding: end-to-end analyses require generating large waveform ensembles across broad priors for searches, injection campaigns, and validation. This motivates kludge waveform families: fast, systematically improvable approximations that support high-throughput studies even before full self-force accuracy is available. In this talk I will present an updated Chimera non-adiabatic kludge, which couples Kerr geodesic motion to a local radiation-reaction prescription to capture non-adiabatic features beyond orbit-averaged evolution while remaining efficient. In this sense, Chimera provides a computationally affordable complexity jump for stress-testing inference pipelines, complementing rather than replacing precision self-force modeling. The new julia-based implementation improves the computation of high-order time derivatives of Kerr functionals, upgrades multipolar waveform construction and post-Newtonian–based approximations. I will benchmark Chimera against other kludges and Teukolsky-based waveforms, and highlight applications to resonance studies.

Extreme-Mass-Ratio Inspirals (EMRIs) are one of the main expected sources of gravitational waves in the low-frequency band where space-based detectors such as the Laser Interferometer Space Antenna (LISA) will operate. The large number of gravitational-wave cycles accumulated in the strong-field regime makes EMRIs extremely precise probes of the local spacetime geometry and highly sensitive to deviations from the Kerr black hole paradigm. In this work, we investigate EMRIs around generic non-Kerr compact objects characterized by a rich and arbitrary multipolar structure. At leading post-Newtonian and linear mass-ratio orders, we incorporate in the waveform model both axisymmetric and non-axisymmetric components of the mass quadrupole and octupole moments, thereby parameterizing the breaking of two fundamental symmetries of the Kerr metric. Following the philosophy of the EMRI Analytic Kludge models, we study the impact of these modifications on the gravitational-wave signal and use a Fisher-matrix analysis to evaluate LISA’s ability to constrain deviations of the multipole moments from their Kerr values, including the detection of symmetry-breaking effects. This allows us to assess how effectively LISA could probe models beyond General Relativity that predict horizon-scale modifications, such as the fuzzball model proposed in string theory. Our results show that future LISA observations of EMRIs will provide powerful and unprecedented tests of black hole structure and the underlying theory of gravity. In particular, using one year of LISA data from the inspiral of a 10 solar-mass compact object into a rotating supermassive black hole of 10^6 solar masses with a signal-to-noise ratio of 30, it will be possible to place tight bounds on deviations from the two fundamental symmetries of the Kerr metric, constraining equatorial symmetry breaking to the 10^-2 level and axial symmetry breaking to the 10^-3 level.

Fast EMRI Waveforms (FEW) is a state-of-the-art software package for generating fast and accurate waveforms for extreme mass ratio inspirals (EMRI). It is a development space for advanced physics models that feed directly into usable waveforms for real gravitational-wave data analysis. Last year, the FEW development community released the new state-of-the-art EMRI analysis waveform: “FastKerrEccentricEquatorialFlux.” This model covers sources in an eccentric orbit around a spinning central massive black hole (MBH) that is limited to the equatorial plane. It currently includes computations to adiabatic order in the gravitational self-force, and prograde and retrograde orbits. The European Distributed Data Processing Center (DDPC) used the FastKerrEccentricEquatorialFlux waveform model in its recent release of the Mojito-light dataset. In this talk, I will discuss the important updates provided in FEW’s FastKerrEccentricEquatorialFlux model and the role it is playing in the modern EMRI analysis landscape.

Modeling transient orbital resonances is a crucial yet challenging factor for constructing accurate gravitational-wave (GW) templates of extreme-mass-ratio inspirals (EMRIs). Resonance effects are significant because they induce drastic alterations in the orbital dynamics, leading to an overall dephasing in the emitted GW signal and potentially affecting the detection and parameter estimation of EMRI systems. In this talk, I will discuss the bias induced in EMRI parameter estimation by transient orbital resonances, using a Fisher matrix approach. We focus in particular on the most dynamically significant, low-order resonances - such as the 3:2 and the 2:1 - as well as on other high-order, subdominant resonances, including the 3:1 and the 4:3. For most of the generic orbits considered, neglecting resonance effects results in significant losses in signal-to-noise ratio and induces bias in parameter recovery. Furthermore, both the sign and the amplitude of the resonance-induced jumps, hence, their dependence on the orbital phases, must be carefully modeled.

Triple systems consisting of a double white dwarf (DWD) and a planet or a brown dwarf, although considered abundant in literature, are yet to be observed. The existence of a third body causes a periodic doppler-shifted perturbation in the DWD waveform. Recent furtherance in the field has shown that LISA has the potential to detect such sub-stellar objects (SSOs). LISA’s detection of triples will allow us a unique perspective of these planetary systems, diversifying our knowledge of the SSO population.

We aim to incorporate a search pipeline dedicated to the detection and characterization of these sources into the LISA Global Fit. To determine LISA's efficacy in detecting triple systems, we simulated GW emission from DWDs and triples, and put them through statistical tests. Our findings demonstrate ~ hundreds of detectable SSOs, deeming the addition of LISA triples into the Global Fit necessary. In this talk, we will outline the methodologies employed and the results of our analysis.

Direct-collapse black holes (DCBHs) are an important component of the massive black hole population of the early universe, and their formation and early mergers will be prominent in the data stream of the Laser Interferometer Space Antenna (LISA). However, the population and binary properties of these early black holes are poorly understood, with masses, mass ratios, spins, and orbital eccentricities strongly dependent on the details of their formation, and the properties of the remaining exterior material (baryonic and non-baryonic), which may be substantial to the point of merger.

We report on progress on our project to take collapsing cosmological halos from the Renaissance Simulations generated by the Enzo code, evolving them to the onset of the relativistic regime with the MESA stellar evolution code, and on to full collapse to intermediate-mass black holes using the Einstein Toolkit. The aims of this project are to assess the ability of LISA to detect and identify the collapse gravitational waveforms, and also to inform massive black hole populations available for later mergers in the LISA regime.

Dual AGN at kpc separations are the parent population MBHBs that eventually merge and produce the GW events. Studying these systems is of great importance to: 1) predict the LISA GW event rate and the PTA background; 2) provide an astrophysical interpretation of the GW data; 3) test many assumptions of the galaxy/black hole formation and co-evolution models.

We will present the results of a large, on-going search for compact (separations between 0.2” and 0.8”, corresponding to 1.5-7 kpc) dual AGN at 0.5z3.5, where the merging activity reach its peak. We are using Gaia and Euclid to select large sample of candidates, and a number of space (HST, JWST, Chandra) and ground-based telescopes (VLT/MUSE, VLT/ERIS, Keck, LBT, VLBA, LOFAR, and others) to classify them and study their properties. This intense observational effort has allowed us to collect, to the first time, a sizable sample of dual AGN outside the local universe. We will present the physical properties of these systems in terms of dual fraction and BH mass distribution, and how they compare with the predictions of current models of galaxy formation and galaxy/SMBH co-evolution.

This is a remote presentation

Similarly to electromagnetic (EM) signals, gravitational lensing by intervening galaxies can also affect gravitational waves (GWs). In this work, we estimate the strong-lensing rate of massive black hole mergers observed with LISA. Given the uncertainties in the source populations as well as in the population of galaxies at high redshift, we consider: six different source population models, including light and heavy seeds, as well as three lens population models, including redshift-independent and redshift-dependent evolution properties. Among all the scenarios explored, the expected number of strong lensed events detected in a 4-year observation time in LISA ranges between 0.13-231 with most of them having two (one) images detectable in the heavy (light) seed scenarios. The event numbers obtained correspond to 0.2%–0.9% of all detected unlensed events. Out of all the detectable strong-lensed events, up to 61% (in the light-seed scenario) and 1% (in the heavy-seed scenario) of them are above the detectability threshold solely due to strong lensing effects and would otherwise be undetectable. For detectable pairs of strong-lensed events by galaxy lenses, we also find between 72%–81% of them to have time delays from 1 week to 1 year.

ArXiv number: 2510.02061

This is a remote presentation

The combination of cosmological observations and strong gravitational fields measurements provides a unique opportunity to test fundamental physics. we present constraints on variations of fundamental constants derived from astrophysical systems and discuss their implications for strong gravitational fields and gravitational-wave sources in the LISA band. These effects may influence binary evolution, merger rates, and waveform properties in scenarios involving modified gravity or scalar fields. This work emphasizes the synergy between LISA and traditional astrophysical probes in exploring the nature of gravity and fundamental interactions.

The Laser Interferometer Space Antenna (LISA) will probe a new regime of low-frequency gravitational waves collectively from both local and high redshift sources. Massive black hole binaries (MBHBs)–a notable population detectable by LISA–are expected to be observable in the electromagnetic (EM) regime. These prospects motivate the need for an informed EM follow-up strategy that includes understanding the types of photometrically variable Galactic binaries that may pose as EM “imposters” for the higher redshift MBHB. I will present the framework for determining the number of potential EM imposters for a case study MBHB of 10^5 solar masses. The framework begins by simulating the Galactic binary population, identifying EM imposters for the MBHB and then selecting regions of the Milky Way that would be projected onto the error region of a MBHB detected by LISA. I then calculate the theoretical total number of EM imposters that would be projected onto the MBHB error box and finally estimate the number of imposters that could be detected optically. For this case study, I find that there may be greater than 10^5 Galactic binaries in the foreground of the MBHB of which only a fraction will be bright EM sources contributing to the confusion.

The Laser Interferometer Space Antenna (LISA) is expected to have a source rich data stream containing signals from large numbers of many different types of source. This will include both individually resolvable signals and overlapping stochastic backgrounds, a regime intermediate between current ground-based detectors and pulsar timing arrays. The resolved sources and backgrounds will be fitted together in a high dimensional Global Fit. To extract information about the astrophysical populations to which the sources belong, we need to decode the information in the Global Fit, which requires new methodology that has not been required for the analysis of current gravitational wave detectors. In this talk I will present a hierarchical Bayesian framework designed to infer the properties of astrophysical populations directly from the output of a LISA Global Fit, consistently accounting for information encoded in both the resolved sources and the unresolved background. We will also present some results demonstrating the effectiveness of the method using a simplified model of the Global Fit. Our approach provides a practical foundation for population inference using LISA data.

The inspiral, merger, and ringdown of Massive Black Hole Binaries (MBHBs) are among the primary sources of gravitational waves for the future Laser Interferometer Space Antenna (LISA), which is expected to observe these systems across the observable universe. Detecting and accurately characterizing these signals requires robust and efficient data analysis techniques. We present a simulation-based inference approach based on Sequential Neural Likelihood, where neural density estimators, implemented via normalizing flows, are trained on simulated MBHB signals and noise to construct a surrogate likelihood. This approach avoids repeated forward simulations during inference and enables posterior sampling via Markov Chain Monte Carlo at a significantly reduced computational cost. The iterative nature of the method allows for targeted refinement of models for specific events, achieving accurate posterior distributions with fewer than 2% of the simulator calls required by standard techniques. We analyze the performance of the method, emphasizing the critical role of data compression in determining both accuracy and efficiency. Comparisons with traditional MCMC results are presented, and we discuss strategies for further improvement and extensions to more realistic scenarios, including higher-order waveform modes, non-stationary noise, glitches, and data gaps.

This work investigates when two gravitational-wave sources can be confused with a single effective source in LISA data. The problem is formulated in the noise-weighted Hilbert space of detector waveforms, where the single-source waveform family defines a manifold and multi-source signals are represented as superpositions in the same space. In the present work we developed a geometric framework for source confusion based on a distance-to-manifold viewpoint tied directly to likelihood optimization, with particular emphasis on nearby sources and on the role of the extrinsic geometry of the single-source waveform manifold. The goal is to provide a mathematically well-founded description of multi-to-single degeneracies relevant to crowded LISA signal environments.

The LISA data will be non-stationary due to unresolved galactic signals and time varying instrument noise. Wavelet domain, time-frequency methods are a promising approach to handling non-stationary data, while also allowing for efficient representation of signals and fast likelihood evaluations. I will discuss how we have been incorporating time-frequency analysis in the GLASS global fit algorithms.

LISA's data will contain every mHz compact binary system in the Milky Way — the Galactic binaries (GBs) — in one of two ways: as ~10,000 individually-observable signals (resolved GBs), and as a prominent confusion noise from the remaining ~30 million systems (unresolved GBs, i.e., the Galactic foreground). Together, the resolved GBs and the Galactic foreground comprise a complete sample — a rare opportunity in astronomy, full of promise for Galactic and stellar astrophysics. However, current LVK population inference methods will not suffice for LISA due to the transdimensional, global nature of LISA data analysis and the circular dependence of the resolved/unresolved GBs. We present a method for GB population inference within the global fit, demonstrate a prototype GPU-accelerated population inference module for the global fit, and lay out a roadmap for — and progress towards — an astrophysically-motivated LISA global fit with embedded population inference.

We introduce a highly efficient Bayesian global fit for LISA that combines a maximum-likelihood galactic binary search with a simultaneous massive black hole binary search. The maximum likelihood solutions initialize the reversible-jump MCMC sampler, bypassing burn-in entirely and bringing the total cost to roughly 10 USD. Applied to the latest LISA data challenge dataset, the pipeline achieves fast convergence and robustly recovers the Bayesian posterior across all source classes.

To meet mission goals, the LISA hardware must satisfy stringent requirements on the Power Spectral Density (PSD) of instrument noise, which propagates linearly into the observatory sensitivity. Experimental verification of these requirements is far from straightforward, as achieving high frequency resolution demands extremely long measurements, impractical on the ground. With measurement durations from a fraction of a day to a few days, frequency resolution is limited; any attempt to improve on it comes at the expense of precision in PSD estimates, for a given time series. Moreover, the resolution set by these durations still yields a very large number of frequencies, making joint statistics impractical.

We present a procedure developed within the ESA LISA Performance and Operations group for PSD requirement verification. The method reduces the number of independent frequencies while preserving useful resolution, and estimates the global Bayesian probability of compliance.

Extreme mass ratio inspirals are systems in which a compact object gradually inspirals into a much more massive companion, typically in galactic center environments. These systems are promising probes of strong field gravity and galactic dynamics and are key targets for future space based gravitational wave detectors. However, nearby objects in dense galactic centers can perturb these inspirals. In this work, we study the impact of a third body perturber near resonance using an extended analytic and computational framework that allows for general outer body orbits and multiple resonances across a range of black hole spins. Analyzing over 142,000 resonant interactions across 180 orbital systems, we find that perturbations remain within the perturbative regime but can induce waveform phase shifts of order 0.1 radians. These results highlight the importance of incorporating environmental effects into accurate EMRI waveform models.

FastEMRIWaveforms is a modular framework for accurately modeling and rapidly generating the gravitational wave (GW) signals of asymmetric-mass-ratio compact object binaries, in particular extreme mass-ratio inspirals (EMRIs). By combining multiscale self-force theory with GPU hardware acceleration, FEW enables the generation of complicated, multi-year GW signals on millisecond timescales. A key milestone in the framework’s development was the release of FastKerrEccentricEquatorialFlux, which provides waveforms with leading-order, also known as adiabatic, phase-accuracy for eccentric inspirals into rapidly rotating Kerr black holes. It currently serves as a key EMRI model in LISA data analysis pipelines and has been used for generating EMRI signals in the Mojito LISA datasets. In this talk, I will present recent advances by the FEW development team in expanding both the accuracy and scope of the framework. Key highlights include: the first precessing binary model in FEW, bridging the gap toward fully generic orbital configurations; the first waveform model with sub-radian (or post-adiabatic) phase accuracy implemented within FEW; new inspiral-merger-ringdown and hybrid waveform models; novel methods for capturing highly-eccentric and zoom-whirl dynamics; and extended support for beyond-vacuum models. I will close by outlining the roadmap and open challenges toward a fully generic—precessing and eccentric—phase-accurate waveform model within FEW, a critical milestone for realizing the full scientific potential of LISA.

This is a remote presentation

We present a new detection statistic for EMRI searches that takes advantage of several special properties of EMRI signals – properties which so far have mostly been unexploited. We present estimates of the sensitivity of our new statistic (and the corresponding search method), which are very encouraging.

Gravitational-wave signals are typically interpreted under the vacuum hypothesis, i.e. assuming negligible influence from the astrophysical environment. This assumption is expected to break down for low-frequency sources such as extreme mass ratio inspirals (EMRIs), which are prime targets for the Laser Interferometer Space Antenna (LISA) and are expected to form, at least in part, in dense environments such as Active Galactic Nuclei or dark-matter spikes/cores. Modeling environmental effects parametrically is challenging due to the large uncertainties in their underlying physics. We propose a non-parametric test for environmental effects in EMRIs, based on assessing the self-consistency of vacuum parameter posteriors inferred from different portions of the signal.

Our results demonstrate that this test can reveal statistically significant inconsistencies from vacuum signals — arising from e.g. incomplete modeling, environmental effects or deviations from General Relativity — without introducing additional parameters or assumptions about the underlying physics

The upcoming space-based gravitational-wave detector LISA is expected to observe inspiraling binary black holes, a fraction of which may retain moderate to high orbital eccentricities. Yet no established framework currently exists for testing general relativity (GR) with eccentric inspirals in the LISA band. Unlike circular binaries, eccentric systems emit gravitational waves at multiple orbital harmonics, while relativistic pericenter precession further splits each harmonic into a triplet in the frequency domain. These characteristic spectral features enable a novel test of GR. We derive a frequency-domain waveform model for eccentric binaries that includes a phenomenological non-GR parameter describing deviations in the rate of pericenter advance, which vanishes in the GR limit. Using this parametrized waveform, we perform Bayesian inference with LISABETA, incorporating the full LISA response. We show that LISA observations of eccentric binaries can place stringent constraints on deviations from general relativity, with sensitivity improving significantly as the orbital eccentricity increases. Our results provide the first demonstration of LISA’s potential to constrain deviations from GR using eccentric binary systems.

Extreme mass ratio inspirals (EMRIs), in which a stellar-mass compact object such as a black hole spirals into a supermassive black hole, are some of the most promising sources for LISA because they undergo thousands of orbits in the strong gravity of the massive black holes. Numerous theoretical studies of EMRIs have been conducted, but we are hampered by the difficulty of direct observations of compact objects at the relevant subparsec scales. However, tidal disruption events (TDEs), particularly in the Rubin era, seem promising as scouts for EMRIs in the sense that similar dynamical processes might apply to both EMRIs and TDEs. I will discuss these processes, including questions about whether channels such as binary tidal separation might dominate the observed rate of EMRIs even if two-body scattering provides the main path toward TDEs.