Journées LISA France
Amphithéatre Henri Mineur
IAP
Les journées LISA France ont pour objectif de rassembler la communauté LISA France et de partager les dernières avancées qui concernent la mission :
- Instrumentation
- Science
- Traitement des données
- Théorie et modélisation
Le programme est élaboré via un appel à contribution ouvert à tous. Les jeunes chercheuses et jeunes chercheurs (doctorant.e.s, post doctorant.e.s, chercheuses et chercheurs en début de carrière) sont fortement encouragés à présenter leur travaux à la communauté.
Les journées se tiendra en mode hybride depuis l'amphithéâtre de l'IAP. Il est nécessaire de s'inscrire pour participer. Le lien zoom ne sera fourni qu'aux personnes inscrites. La date limite pour les inscriptions est le 24 avril 23:59.
-
-
Informations généralesPrésident de session: Martin Boutelier (CNES)
-
1
Introduction - Informations généralesOrateur: Martin Boutelier (CNES)
-
2
News from ESAOrateur: Dr Oliver Jennrich (ESA)
-
3
Avancement du projet LISA en FranceOrateur: Jean-Charles Damery (CNES)
-
4
Avancement DDPCOrateurs: Anne-Therese Nguyen, Antoine Petiteau (CEA/IRFU/DPhP)
- 5
- 6
-
1
-
16:05
Pause Kfé
-
Session 1Président de session: Martin Boutelier (CNES)
- 7
-
8
Gravitational radiation reaction for compact binary systems at 4.5 post-Newtonian order
Compact binaries are the primary sources of gravitational waves measured by gravitational-wave detectors. In this context, obtaining the equations of motion in different approximation schemes is essential for producing waveform templates with the accuracy required by future LISA detections.
In this presentation, I derive the gravitational radiation-reaction force on a compact binary system at the fourth-and-a-half post-Newtonian (4.5PN) order of general relativity, i.e., 2PN orders beyond the leading 2.5PN radiation-reaction term in harmonic coordinates. This result is derived using the 4.5PN gravitational radiation-reaction force obtained recently in the Burke–Thorne (BT) coordinate system.
This represents a significant improvement, as harmonic coordinates provide a manifestly Lorentz-invariant formulation, and offer a much simpler expression compared to the BT coordinates. Moreover, they pave the way for comparisons with other approximation schemes (such as post-Minkowskian) at 4.5PN.
Using the harmonic radiation-reaction acceleration, we also derive from first principles the flux-balance laws and the center-of-mass position, in a general frame up to the 4.5PN order.Orateur: Emeric Seraille (LPENS) -
9
Separation of Milky Way bulge and disk in LISA residual data
LISA is a space mission to observe millihertz gravitational waves (GWs), planned for 2035. Millions of white dwarf binaries in the Milky Way (MW) will contribute to LISA data, of which a few thousand are expected to be individually resolved. The unresolved signal, dubbed the Galactic foreground, can be astrophysically interpreted through careful analysis. We assess LISA's capability to separate the MW bulge and disk foreground components. We generate mock catalogues of Galactic binaries (GBs) in the MW, using axisymmetric density templates while varying total mass and mass ratio. For each catalogue, we simulate the LISA foreground through an iterative technique. We then use the BLIP pipeline to fit the signal to a Whittle likelihood model, characterizing the data as colored Gaussian noise of parametrized spectral shape, with known nonstationarities induced by the GW source anisotropy. We show that the bulge and disk signals can be distinguished. From a simple model where their spectra differ only in amplitude, we are able to measure total Galactic binary population mass within 2% and the bulge/disk ratio within 8%, matching the precision of photometric and kinematic data. Our work demonstrates that resolving MW structure is feasible in a LISA Global Fit, providing complementary information to electromagnetic observations.
Orateur: M. Solano Felicio (Observatoire de la Côte d'Azur) -
10
Presentation of the External Data work package
The LISA External Data work package is coordinated by IRAP and L2IT in the framework of the DDPC. Its primary objective is to collect all data from sources outside LISA that may be relevant to the mission, and to provide the software infrastructure necessary for their distribution in a standardized way that is compatible with the software requirements of the DDPC.
A first prototype is currently under development. It consists of a database of verification binaries and its API, and is intended to be used in the generation of future LISA data challenges (starting with Mojito Heavy) and to generate informative priors for global fit pipelines.
In this presentation, we will describe the goals of the External Data project, its use cases, and its current status. As a transverse project that aims to provide a service to the LISA community, it will also be an occasion to collect feedback and specific needs from the community in order to include new features in future versions of the software.
Orateur: Dr Adrian MACQUET (IRAP) -
11
Summary of contributions of CEA to DDPC and P&O
In this talk, we will provide an update on the contributions of CEA to DDPC and P&O.
Within DDPC, CEA is involved in CU L01 and CU L2A. We will present the latest updates on our work on the common dataset, as well as on source detection and parameter estimation in the context of the Low Latency Alert Pipeline (LLAP).
As the lead of CU L01 and technical deputy of CU L2A, we will also share updates from these CUs.
For the P&O group, we will provide recent updates on its activities.
Orateur: Dr Nicolas Dagoneau (CEA Paris-Saclay) -
12
Fast localization of EMRI sources using the coronagraphic TDI variable for LISA
The LISA constellation offers a unique opportunity to detect EMRIs, with a sensitivity band that is particularly well suited to these sources. Their detection will not only allow us to probe the intrinsic properties and immediate environments of black holes in the local Universe, but also to use these sources as “dark sirens” for cosmological studies \cite{colpi2024lisa}.
To fully exploit EMRI observations, rapid and reliable sky localization is crucial. It enables the triggering of alerts for electromagnetic observatories, which complements the observation of these complex systems through a multi-messenger approach \cite{kocsis2011observable}. In this context, we present a fast and agnostic localization method.
This approach relies on the construction of a TDI observable, the coronagraphic variable $\kappa$, derived from linear combinations of Sagnac-type channels \cite{costa2025sky}. The principle is as follows: when the direction and frequency parameters injected into the observable exactly match the true source parameters, the gravitational-wave signal is canceled within the $\kappa$ variable. Localization is thus performed by scanning the celestial sphere to identify the configuration that minimizes the spectral power of the signal.
This method also allows for constraining the source localization parameters in an agnostic way, without directly relying on precise waveform models.
The implementation of this approach within the Nullstream pipeline has already demonstrated promising results for MBHBs \cite{costa2025coronagraphic}. This study validates the extension of the method to EMRI signals. In the presence of noise, we show that the method remains robust, notably because EMRIs exhibit a slow frequency evolution, allowing them to remain for long durations within the optimal sensitivity band.
Finally, this work opens perspectives on handling non-stationary noise and raises an important question for LISA data analysis: to what extent can the persistent and complex signal of EMRIs influence or bias the detection and characterization of other transient sources, such as MBHBs?
REFERENCES :
@inproceedings{costa2025sky,
title={Sky Localization of Massive Black Hole Binaries with LISA: Applying Coronagraphic Time-delay Interferometry to Low-latency Searches},
author={Costa Barroso, R and Lemi{`e}re, Y and Mauger, F and Baghi, Q},
booktitle={SF2A-2025: Proceedings of the Annual meeting of the French Society of Astronomy and Astrophysics. Eds.: A. Siebert},
pages={413--416},
year={2025}
}@article{colpi2024lisa,
title={LISA definition study report},
author={Colpi, Monica and Danzmann, Karsten and Hewitson, Martin and Holley-Bockelmann, Kelly and Jetzer, Philippe and Nelemans, Gijs and Petiteau, Antoine and Shoemaker, David and Sopuerta, Carlos and Stebbins, Robin and others},
journal={arXiv preprint arXiv:2402.07571},
year={2024}
}@article{kocsis2011observable,
title={Observable signatures of extreme mass-ratio inspiral black hole binaries embedded in thin accretion disks},
author={Kocsis, Bence and Yunes, Nicol{\'a}s and Loeb, Abraham},
journal={Physical Review D—Particles, Fields, Gravitation, and Cosmology},
volume={84},
number={2},
pages={024032},
year={2011},
publisher={APS}
}@article{costa2025coronagraphic,
title={Coronagraphic time-delay interferometry: characterization and updated geometric properties},
author={Costa Barroso, Raissa and Lemi{`e}re, Yves and Mauger, Fran{\c{c}}ois and Baghi, Quentin},
journal={Classical and Quantum Gravity},
volume={42},
number={11},
pages={115016},
year={2025},
publisher={IOP Publishing}
}Orateur: Hugo Alexandre (LPC Caen)
-
-
-
08:30
Accueil Kfé
-
Session 2Président de session: Astrid Lamberts (Observatoire de la Côte d'Azur)
-
13
Waveform systematics for massive black hole binaries in LISA
Massive black hole binaries (MBHBs) are one of the main targets of the LISA mission, with numerous science objectives ranging from astrophysics to comsology. These signals will reach very high signal-to-noise ratios (SNRs), from hundreds up to thousands, opening the possibility of high-precision scientific measurements, such as the search for deviations from General Relativity. However, in order to enable the full potential of the instrument, waveform models must be accurate enough to keep systematic errors subdominant with respect to statistical errors. Inaccurate waveform models could cause biases in the parameter estimation, as well as leave residuals in the data stream after signal subtraction, affecting the LISA global fit. Focusing on spin-aligned circular systems, we present a first exploration of these biases for example MBHB signals, using a full simulated Bayesian inference, and investigate the relation between large SNRs and accuracty requirements.
Orateur: Sylvain Marsat (L2I Toulouse, CNRS/IN2P3, Université de Toulouse) -
14
What’s new in L01?
The LISA L01 pipeline plays a central role in transforming raw measurements into calibrated, science-ready Time-Delay Interferometry (TDI) observables. This stage encompasses data preprocessing for calibration, noise mitigation, and clock synchronization—key steps required before higher-level scientific analysis and global fitting.
In this talk, I will present recent updates of the L01 pipeline developed within the DDPC CU L01, led by the French and UK community with major contributions from LTE and CEA. I will highlight the processing pipelines implemented and foreseen for the Mojito common datasets. I will also discuss ongoing developments that extend beyond Mojito, aiming to address future challenges in noise modeling and estimation, mitigating other instrumental effects (eg. clock errors), robustness (eg. against gaps or glitches), scalability and performance (eg. to deliver low-latency L1 data for alerts), and the development of alternative pipelines (eg. based on principle-component analysis or time-domain linear algebra).
Orateur: Jean-Baptiste Bayle (CEA) -
15
Advancing the Erebor Global Fit for the "Mojito" LISA Data Challenge: Machine Learning, GPU Acceleration, and Population Inference
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. To prepare for future data streams, global fit pipelines must be tested against realistic scenarios. The most recent LISA Data Challenge, "Mojito," provides the ideal testbed, introducing full global fit tasks with unprecedented realism, including time-varying noise, realistic orbits, and detailed populations of interacting and detached GBs. In this work, we present the most recent advancements to the GPU-accelerated global fit pipeline, Erebor. More specifically, we focus on the analysis of the resolved population of GBs and demonstrate a methodology to perform rigorous astrophysical inference on the entire population of GBs.
To extract the resolved GBs, the pipeline now utilizes GPU-enhanced and computationally efficient waveform and likelihood computations using the recently developed TDI-on-the-fly approach (Littenberg & Cornish 2025). Furthermore, to navigate the vast parameter space associated with fitting tens of thousands of binaries, we integrate machine learning-enhanced proposals utilizing normalizing flows, which significantly accelerate the sampling convergence of the resolved GB block. Using this updated GB block, we present the results of preliminary runs on the Mojito data challenge and provide an outlook on future developments for the Erebor pipeline.
Crucially, this integration drastically accelerates sampling convergence, making it computationally tractable to perform joint population inference. This population inference must be intrinsic to the global fit, because traditional post-processing methods used in LVK data analysis fail due to the transdimensional nature of extracting an unknown number of sources. To address this, 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. This algorithm dynamically jumps between competing population synthesis models, altering the overarching astrophysical priors and the Poisson distributions governing the expected number of resolved systems. This approach rigorously quantifies and improves our understanding of LISA’s sensitivity to specific astrophysical processes and explores physical degeneracies between astrophysical hyperparameters.
Orateur: Coen Rondeel (Observatoire de la Côte d'Azur) -
16
Detectability of the gravitational memory effect with LISA
The gravitational memory effect is an unobserved prediction of General Relativity. In our work, we investigate the prospects for detecting this memory effect with LISA from a massive black hole binary merger. Using a Bayesian analysis, we assess the conditions under which the memory signal can be confidently identified in LISA data. Our results highlight a region of the source parameter space in which memory detection is likely, and identify a simplified criterion that enables the detectability of individual events to be quickly estimated. We also compare these results with simulated populations of supermassive black hole binaries in order to estimate the number of memory events that are expected to be detectable over the lifetime of LISA. This enables us to contextualise the results within a realistic astrophysical framework. Overall, our analysis shows that LISA could detect the memory effect and constrain its amplitude, providing a probe for fundamental physics.
Orateur: Adrien Cogez (CEA/IRFU/DPhP) -
17
Towards systematic searches for LISA white dwarf binaries with multiband photometry
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 (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.
Orateur: Alice Perego (Observatoire de la Côte d'Azur) -
18
Machine Learning for Global Fit
The immense scientific potential of LISA hinges on solving an unprecedented data analysis challenge: the Global Fit problem. This involves the simultaneous inference of numerous overlapping signals and instrument noise, framed in a high-dimensional Bayesian setting.
Current approaches rely on computationally intensive Markov chain Monte Carlo (MCMC) techniques with block Gibbs sampling across source classes. Yet, these methods suffer from poor scalability and slow convergence, especially in the presence of source confusion and uncertainty in source number. To address these issues, we introduce GWINESS (Gravitational Wave Inference using NEural Source Separation), a machine learning-based framework inspired by music source separation. Using an encoder-decoder neural architecture, GWINESS aims to perform blind source separation of overlapping gravitational-wave signals—analogous to isolating vocals, drums, and bass in a song.
This talk will present the core principles behind GWINESS, and discuss current limitations, and future directions for integrating ML methods in the Global Fit.
Orateur: Antsa Rasamoela (L2I Toulouse, CNRS/IN2P3, Université de Toulouse)
-
13
-
11:00
Pause Kfé
-
Session 3
-
19
Mitigation of the flexing-filtering effect in time-delay interferometry
The LISA constellation employs three spacecrafts that exchange laser beams to make interferometric measurements over a distance of 2.5 million kms. The measurements will then telemetered down to Earth at a lower sampling frequency. Anti-aliasing filters will be used on board to limit spectral folding of out-of-band laser noise. The primary noise concern in these measurements is laser frequency noise, caused by the variability in the arm-length of the constellation via the disproportionate movement of the spacecrafts and the Doppler effect. Minimization of this noise requires virtual time-shifting of the data using delay operators to build a beam path that simulates equal-arm interferometers. The non-commutativity of these delay operators and on-board filters manifests as a noise (flexing-filtering) that significantly contributes to the noise budget. This non-commutativity is a consequence of the non-flatness of the filter in-band. Attenuation of this noise requires high-order and computationally expensive filters, putting additional demands on the spacecraft.
My presentation will explore an alternative method to reduce this flexing filtering noise via the introduction of a modified delay operator -- accounting for the non-commutativity with the filter in the delay operation itself. This approach allows us to reduce the flexing-filtering noise by over three orders of magnitude whilst reducing the dependency on the flatness of the filter. The work is supplemented by numerical simulations of the data processing chain that compare the results with those of the standard approach.
Orateur: Dr Shivani Harer
-
19
-
Les produits LISAPrésidents de session: Antoine Petiteau (CEA/IRFU/DPhP), Astrid Lamberts (Observatoire de la Côte d'Azur)
-
12:50
Déjeuner - Buffet
-
Session 4Président de session: Antoine Petiteau (CEA/IRFU/DPhP)
-
20
Inférence par simulation pour l’estimation rapide de paramètres d’ondes gravitationnelles de binaires de trous noirs supermassifs
Dans le cadre de l’analyse de données de LISA et du développement du pipeline d’alertes rapides (LLAP) qui impose de fortes contraintes de temps, où la détection et la caractérisation rapide des binaires de trous noirs supermassifs (MBHB) constituent un enjeu central, je présenterai mes travaux sur l’estimation de leurs paramètres.
Je montrerai ainsi une approche basée sur une simulation de réponse rapide de l’instrument et l’inférence par simulation (simulation-based-inference, SBI) un outil d’intelligence artificielle permettant d’amortir l’inférence et d’estimer rapidement les paramètres des MBHB ainsi que leurs incertitudes, tout en respectant les fortes contraintes de temps relatives au LLAP.
Je montrerai en particulier des résultats combinant SBI et des méthodes d’importance sampling puis expliquerai quelles sont mes pistes d’amélioration pour l’échantillonnage des distributions des paramètres.
Orateur: Louis Le Saulnier (CEA) -
21
Low-Latency detection and parameter estimation of massive black hole binaries for LISA using flow matching
A primary target for LISA will be massive black hole binaries (MBHBs), the inspiral of which will be detectable in a few days to months before coalescence. If they occur in gas-rich environments, they may produce electromagnetic (EM) counterparts, allowing for multimessenger follow-up.
To provide triggers to the EM community rapid parameter estimation is needed, specifically of merger time, sky localization, distance and masses, ideally already during the inspiral phase.
However, the Global Fit's detection and parameter estimation of all sources at the same time will not be suitable for near real-time requirements, motivating the development of a low-latency alert pipeline.We present a simulation-based inference (SBI) approach, that avoids likelihood evaluations entirely and shifts the computational cost to the training phase.
We employ conditional flow matching with optimal transport, in which the trajectory from a base distribution to the parameters’ posteriors is learnt conditioning on data.
SBI is suitable not only because of the short inference time thanks to amortisation, but also because it can easily allow for realistic data conditions. Future developments will include real noise, instrumental artifacts and data gaps into the training process.We focus on a sub-population of MBHBs, with masses between 10^4 and 10^5 solar masses, astrophysically motivated by the possibility of observing an EM signature for such systems .
We train and evaluate our network on two data representations: time series and time-frequency.
We compare various time-frequency representations and investigate the use of autoencoders to project the data into a space where the galactic binary foreground is ignored.
We present preliminary parameter estimation results and discuss the future development of a full, low-latency pipeline that can handle realistic LISA data.Orateur: Malvina Bellotti (Observatoire de la Côte d'Azur, Nice) -
22
Time-frequency approach to long-lived signals of LISA: an application to stellar-mass black holes and EMRIs
Among the numerous sources expected to be observed by LISA, stellar-mass black holes and EMRIs constitute two of the main classes with specific science objectives. The former, already observed by LVK detectors, would be now discovered in their early inspiral phase where, for example, eccentricity information would give us information on the environment in which they form and would give a probe of stellar-mass black hole formation channels. The latter would enable observations of quiescent galaxies alike to the Milky Way, not visible in the electromagnetic spectrum, and therefore give information on the content of galaxies similar to ours, as well a being a great probe of General Relativity and its extensions. The signal generated by those source have in common that they stay for a long time in the LISA band (months to years), therefore all source of noise have to be considered as non-stationary. Representing the signals using time-frequency transforms allows us to take this into account. I will present an approach based on the knowledge of the amplitude and the phase of a gravitational waveform to compute an approximate LISA response using Short Time Fourier Transforms, to be used to run MCMC chains to recover the parameters from a given signal, and its application to Mojito Lite simulated data.
Orateur: Gaël Servignat (LUTH) -
23
Performance, Operation et FELIX
Dans le cadre du groupe Performance & Operations, je présenterai une analyse des impacts des simplifications et corrélations au niveau constellation. Deux questions ont guidé ce travail : ces simplifications remettent-elles en cause le budget de bruit, et permettent-elles de conduire une campagne de vérification ? Les résultats montrent que si le budget de bruit reste valide, la vérification ne peut pas être conduite en l'état, motivant une reformulation de la spécification au niveau Mission.
Je présenterai également FELIX (Friendly Enhanced LISA Information eXplorer), un système RAG (basé sur un Large Language Model) que j'ai développé pour faciliter l'accès à la documentation technique LISA. FELIX permet d'interroger en langage naturel un corpus de documents LISA et d'obtenir des réponses contextualisées. Au-delà de la recherche documentaire, il intègre une approche agentique permettant d'appeler directement le modèle de performance de LISA. Actuellement déployé sur infrastructure GPU dédiée et évalué auprès d'un groupe restreint d'utilisateurs, je discuterai des choix techniques et des premiers retours d'usage.Orateur: joseph Martino (APC) -
24
Optical simulation of the Beams Simulator
The French contribution to the LISA Consortium, on the instrumental side, focuses on developing optical ground support equipment to validate LISA's performance prior to launch. One key task of the French community is to test and validate the performance of the Interferometric Detection System (IDS).
To this end, the IDS Test Set-Up is currently under development. Its objectives are twofold: to verify that picometric stability is achieved within the Interferometric Detection System, and to characterize the tilt-to-length coupling coefficient of the optical bench interferometers — that is, the coupling between the relative angle of the beams and the interferometric length readout. The test set-up comprises several test benches, including a test mass simulator and a beam simulator bench. A thorough understanding of these benches is essential to identify all systematic effects and confirm that the design meets the noise floor requirements needed to characterize the test specimen.
This contribution describes the ongoing simulation work aimed at preparing the test campaign and validating the test benches ahead of characterizing the LISA instrument.
Orateur: Maxime VINCENT (AstroParticule et Cosmologie) -
25
Testing LISA with the Beam Simulator
The presentation will review the technologies that have been developed during the last year for testing the performances of the metrological instrumentation core of LISA with the Beam Simulator, a test contraption. An introductory part will give a general description of the Beam Simulator, explaining the need at the origin of its development, the test Interferometric Detection System (IDS) setup, the key performances and functionalities to validate during the IDS test campaign as well its functional principles. Then it will focus on the technological challenges raised by particular sub-systems of the Beam Simulator, and the progress made on these aspects. Will be evoked mainly the Beam Pointing Alignment Mechanisms to align LISA OB and BSIM with micro-rad and micrometer precisions ; Rx Pointing Control System a set of two tip-tilt mirrors used to tilt the Rx beam with respect to the Ref. beam to characterized Tilt-To-Length ; and apodisation and centering masks to control beam shape and optimize detection.
Orateur: Lucas Pardessus (APC)
-
20
-
AOBPrésident de session: Martin Boutelier (CNES)
-
26
Empreinte environnementale de la contribution françaiseOrateur: M. Jean-Charles Damery (CNES)
-
26
-
08:30