Modular global-fit pipeline for LISA data analysis

Author(s)

Deng, Senwen, Babak, Stanislav, Le Jeune, Maude, Marsat, Sylvain, Plagnol, Éric, Sartirana, Andrea

Abstract

We anticipate that the data acquired by the Laser Interferometer Space Antenna (LISA) will be dominated by the gravitational wave signals from several astrophysical populations. The analysis of these data is a new challenge and is the main focus of this paper. Numerous gravitational wave signals overlap in the time and/or frequency domain, and the possible correlation between them has to be taken into account during their detection and characterization. In this work, we present a method to address the LISA data analysis challenge; it is flexible and scalable for a number of sources and across several populations. Its performance is demonstrated on the simulated data LDC2a.

Figures

Architecture of the pipeline. Arrows indicate the flow of information. Black arrows mean the piece of information is passed only once in the full process, whereas the other colours indicate the information is transfered for each iteration cycle. Inside each cycle, the MBHB and GB blocks run concurrently and the noise block runs sequentially after other blocks are finished.

Architecture of the pipeline. Arrows indicate the flow of information. Black arrows mean the piece of information is passed only once in the full process, whereas the other colours indicate the information is transfered for each iteration cycle. Inside each cycle, the MBHB and GB blocks run concurrently and the noise block runs sequentially after other blocks are finished.


The narrow frequency band in the GB block. The shown data contains 3 GB blocks (``GB block data''), each composed of a core (blue tile) and two wings (white and grey tiles). The core takes half of the width of the band in the centre. During the data preparation of a GB block, GB signals in the central live catalogue that belong to the neighbouring but not the current and the adjacent core bands are subtracted. Analyses are performed in the full band (GB block data) but only the results with the central frequency in the core are synchronised to the central live catalogue and the central storage of the chains.

The narrow frequency band in the GB block. The shown data contains 3 GB blocks (``GB block data''), each composed of a core (blue tile) and two wings (white and grey tiles). The core takes half of the width of the band in the centre. During the data preparation of a GB block, GB signals in the central live catalogue that belong to the neighbouring but not the current and the adjacent core bands are subtracted. Analyses are performed in the full band (GB block data) but only the results with the central frequency in the core are synchronised to the central live catalogue and the central storage of the chains.


The 16 frequency bands used for the noise model sampling with their location and size. The solid black line shows an expectation based on \cite{karnesis_characterization_2021}.

The 16 frequency bands used for the noise model sampling with their location and size. The solid black line shows an expectation based on \cite{karnesis_characterization_2021}.


Time series signal of the Sangria dataset zoomed in around the merger of the MBHB 0 and the reconstructed signal for that MBHB. In the second and the fourth panels (from the top), we plot the residuals after the removal of the reconstructed signal in TDI A and TDI E channels. Note that the data is not whitened in this plot.

Time series signal of the Sangria dataset zoomed in around the merger of the MBHB 0 and the reconstructed signal for that MBHB. In the second and the fourth panels (from the top), we plot the residuals after the removal of the reconstructed signal in TDI A and TDI E channels. Note that the data is not whitened in this plot.


Whitened time series signal of the full Sangria dataset and the reconstruction of all the 15 MBHBs. In the second and the fourth panels, we plot the whitened residuals after the removal of the reconstructed signals in TDI A, E.

Whitened time series signal of the full Sangria dataset and the reconstruction of all the 15 MBHBs. In the second and the fourth panels, we plot the whitened residuals after the removal of the reconstructed signals in TDI A, E.


Marginalised posterior distribution of the chirp mass and the mass ratio (\textit{Left}); individual masses (\textit{Right}) for the MBHB sources in the ``Sangria'' dataset. We plot a relative uncertainty and zero corresponds to the true parameter values of injected signals. The left plot: the left histogram (blue) is for the chirp mass \( \mathcal{M}_\text{c} / \mathcal{M}_{\text{c,tr}} -1 \), the right (green) histogram corresponds to the mass ratio \(q / q_\text{tr} - 1\). The right plot: \( m_i / m_\text{i,tr} -1 \), the blue (left) for the primary (heavier) MBH, the green (right) is for the secondary. The dashed lines indicate the quartiles of the distribution.

Marginalised posterior distribution of the chirp mass and the mass ratio (\textit{Left}); individual masses (\textit{Right}) for the MBHB sources in the ``Sangria'' dataset. We plot a relative uncertainty and zero corresponds to the true parameter values of injected signals. The left plot: the left histogram (blue) is for the chirp mass \( \mathcal{M}_\text{c} / \mathcal{M}_{\text{c,tr}} -1 \), the right (green) histogram corresponds to the mass ratio \(q / q_\text{tr} - 1\). The right plot: \( m_i / m_\text{i,tr} -1 \), the blue (left) for the primary (heavier) MBH, the green (right) is for the secondary. The dashed lines indicate the quartiles of the distribution.


Marginalised posterior distribution of the chirp mass and the mass ratio (\textit{Left}); individual masses (\textit{Right}) for the MBHB sources in the ``Sangria'' dataset. We plot a relative uncertainty and zero corresponds to the true parameter values of injected signals. The left plot: the left histogram (blue) is for the chirp mass \( \mathcal{M}_\text{c} / \mathcal{M}_{\text{c,tr}} -1 \), the right (green) histogram corresponds to the mass ratio \(q / q_\text{tr} - 1\). The right plot: \( m_i / m_\text{i,tr} -1 \), the blue (left) for the primary (heavier) MBH, the green (right) is for the secondary. The dashed lines indicate the quartiles of the distribution.

Marginalised posterior distribution of the chirp mass and the mass ratio (\textit{Left}); individual masses (\textit{Right}) for the MBHB sources in the ``Sangria'' dataset. We plot a relative uncertainty and zero corresponds to the true parameter values of injected signals. The left plot: the left histogram (blue) is for the chirp mass \( \mathcal{M}_\text{c} / \mathcal{M}_{\text{c,tr}} -1 \), the right (green) histogram corresponds to the mass ratio \(q / q_\text{tr} - 1\). The right plot: \( m_i / m_\text{i,tr} -1 \), the blue (left) for the primary (heavier) MBH, the green (right) is for the secondary. The dashed lines indicate the quartiles of the distribution.


Marginalised posterior distribution of the spin projections for the MBHB sources in the Sangria dataset (settings and colours are similar to the right plot of \cref{fig:violin-chirp-ratio}). The dashed lines indicate the quartiles of the distribution. The \(y\)-axis is the spin projection uncertainty, defined as \(\chi_{i} - \chi_{i, \text{inj}}\).

Marginalised posterior distribution of the spin projections for the MBHB sources in the Sangria dataset (settings and colours are similar to the right plot of \cref{fig:violin-chirp-ratio}). The dashed lines indicate the quartiles of the distribution. The \(y\)-axis is the spin projection uncertainty, defined as \(\chi_{i} - \chi_{i, \text{inj}}\).


The distribution of detected GBs (cyan) with SNR above 6 compared to the catalogue (injected) sources with SNR above 6 (yellow) and 8 (green). We use the noise PSD estimated from the global-fit pipeline to compute the SNR.

The distribution of detected GBs (cyan) with SNR above 6 compared to the catalogue (injected) sources with SNR above 6 (yellow) and 8 (green). We use the noise PSD estimated from the global-fit pipeline to compute the SNR.


The distribution of identified GBs (yellow), partially recovered (green) and false detections (cyan) in frequency (left panel) and SNR (right panel).

The distribution of identified GBs (yellow), partially recovered (green) and false detections (cyan) in frequency (left panel) and SNR (right panel).


Upper panel: CDF of detected GBs with three SNR cuts (6, 8, 10) as a function of their overlap with the catalogue sources. Lower panel: we plot the survived GBs with the same SNR cuts as a function of the overlap. A similar plot is given in \cite{katz_efficient_2024}.

Upper panel: CDF of detected GBs with three SNR cuts (6, 8, 10) as a function of their overlap with the catalogue sources. Lower panel: we plot the survived GBs with the same SNR cuts as a function of the overlap. A similar plot is given in \cite{katz_efficient_2024}.


Evolution of the noise power spectral density (PSD) for 10 first iterations of the global fit. For each iteration, we plot a dozen curves corresponding to the randomly chosen sample points (noise model parameters).

Evolution of the noise power spectral density (PSD) for 10 first iterations of the global fit. For each iteration, we plot a dozen curves corresponding to the randomly chosen sample points (noise model parameters).


References
  • [1] M. Colpi, K. Danzmann, M. Hewitson, K. HolleyBockelmann, P. Jetzer, G. Nelemans, A. Petiteau, D. Shoemaker, C. Sopuerta, R. Stebbins, N. Tanvir, H. Ward, W. J. Weber, I. Thorpe, A. Daurskikh, A. Deep, I. F. N´u˜nez, C. G. Marirrodriga, M. Gehler, J.-P. Halain, O. Jennrich, U. Lammers, J. Larra˜naga, M. Lieser, N. L¨utzgendorf, W. Martens, L. Mondin, A. P. Ni˜no, P. Amaro-Seoane, M. A. Sedda, P. Auclair, S. Babak, Q. Baghi, V. Baibhav, T. Baker, J.-B. Bayle, C. Berry, E. Berti, G. Boileau, M. Bonetti, R. Brito, R. Buscicchio, G. Calcagni, P. R. Capelo, C. Caprini, A. Caputo, E. Castelli, H.-Y. Chen, X. Chen, A. Chua, G. Davies, A. Derdzinski, V. F. Domcke, D. Doneva, I. Dvorkin, J. M. Ezquiaga, J. Gair, Z. Haiman, I. Harry, O. Hartwig, A. Hees, A. Heffernan, S. Husa, D. Izquierdo, N. Karnesis, A. Klein, V. Korol, N. Korsakova, T. Kupfer, D. Laghi, A. Lamberts, S. Larson, M. L. Jeune, M. Lewicki, T. Littenberg, E. Madge, A. Mangiagli, S. Marsat, I. M. Vilchez, A. Maselli, J. Mathews, M. van de Meent, M. Muratore, G. Nardini, P. Pani, M. Peloso, M. Pieroni, A. Pound, H. QuelquejayLeclere, A. Ricciardone, E. M. Rossi, A. Sartirana, E. Savalle, L. Sberna, A. Sesana, D. Shoemaker, J. Slutsky, T. Sotiriou, L. Speri, M. Staab, D. Steer, N. Tamanini, G. Tasinato, J. Torrado, A. Torres-Orjuela, A. Toubiana, M. Vallisneri, A. Vecchio, M. Volonteri, K. Yagi, and L. Zwick, LISA Definition Study Report (2024), arXiv:2402.07571.
  • [2] M. Le Jeune and S. Babak, 10.5281/zenodo.7132178 (2022).
  • [3] M. L. Katz, N. Karnesis, N. Korsakova, J. R. Gair, and N. Stergioulas, An efficient GPU-accelerated multisource global fit pipeline for LISA data analysis (2024), arXiv:2405.04690 [gr-qc].
  • [4] S. Babak, M. Hewitson, and A. Petiteau, LISA Sensitivity and SNR Calculations (2021), arXiv:2108.01167 [astro-ph.IM].
  • [5] T. B. Littenberg and N. J. Cornish, Physical Review D 107, 063004 (2023), arXiv:2301.03673 [gr-qc].
  • [6] S. H. Strub, L. Ferraioli, C. Schmelzbach, S. C. St¨ahler, and D. Giardini, Global Analysis of LISA Data with Galactic Binaries and Massive Black Hole Binaries (2024), arXiv:2403.15318.
  • [7] V. Korol, S. Toonen, A. Klein, V. Belokurov, F. Vincenzo, R. Buscicchio, D. Gerosa, C. J. Moore, E. Roebber, E. M. Rossi, and A. Vecchio, Astronomy & Astrophysics 638, A153 (2020).
  • [8] G. Nelemans, L. R. Yungelson, and S. F. P. Zwart, Monthly Notices of the Royal Astronomical Society 349, 181 (2004), arXiv:astro-ph/0312193.
  • [9] A. Kr´olak, M. Tinto, and M. Vallisneri, Physical Review D 76, 069901 (2007), arXiv:gr-qc/0401108.
  • [10] A. B laut, S. Babak, and A. Kr´olak, Physical Review D 81, 063008 (2010), arXiv:0911.3020.
  • [11] A. Klein, E. Barausse, A. Sesana, A. Petiteau, E. Berti, S. Babak, J. Gair, S. Aoudia, I. Hinder, F. Ohme, and B. Wardell, Physical Review D 93, 024003 (2016).
  • [12] S. Khan, S. Husa, M. Hannam, F. Ohme, M. P¨urrer, X. J. Forteza, and A. Boh´e, Physical Review D 93, 044007 (2016), arXiv:1508.07253 [gr-qc].
  • [13] S. Marsat, J. G. Baker, and T. D. Canton, Physical Review D 103, 083011 (2021), arXiv:2003.00357 [gr-qc].
  • [14] J. A. Edlund, M. Tinto, A. Kr´olak, and G. Nelemans, Physical Review D 71, 122003 (2005), arXiv:grqc/0504112.
  • [15] M. Falxa, S. Babak, and M. Le Jeune, Physical Review D 107, 022008 (2023), arXiv:2208.04575 [astro-ph.IM].
  • [16] P. Jaranowski, A. Kr´olak, and B. F. Schutz, Physical Review D 58, 063001 (1998), arXiv:gr-qc/9804014.
  • [17] R. Prix and B. Krishnan, Classical and Quantum Gravity 26, 204013 (2009), arXiv:0907.2569 [gr-qc].
  • [18] S. Babak and A. Sartirana (2025), in preparation.
  • [19] N. J. Cornish and L. J. Rubbo, Physical Review D 67, 029905 (2003), arXiv:gr-qc/0209011.
  • [20] M. J. D. Powell, The Computer Journal 7, 155 (1964).
  • [21] S. E. Timpano, L. J. Rubbo, and N. J. Cornish, Physical Review D 73, 122001 (2006), arXiv:gr-qc/0504071.
  • [22] Z.-H. Zhan, J. Zhang, Y. Li, and H. S.-H. Chung, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics) 39, 1362 (2009).
  • [23] N. J. Cornish and J. Crowder, Physical Review D 72, 10.1103/physrevd.72.043005 (2005).
  • [24] B. Zackay, L. Dai, and T. Venumadhav, Relative binning and fast likelihood evaluation for gravitational wave parameter estimation (2018), arXiv:1806.08792 [astroph.IM].
  • [25] N. J. Cornish, Physical Review D 104, 104054 (2021), arXiv:2109.02728 [gr-qc].
  • [26] N. J. Cornish, Fast fisher matrices and lazy likelihoods (2013), arXiv:1007.4820 [gr-qc].
  • [27] N. Korsakova, S. Babak, M. L. Katz, N. Karnesis, S. Khukhlaev, and J. R. Gair, Neural density estimation for galactic binaries in LISA data analysis (2024), arXiv:2402.13701 [gr-qc].
  • [28] S. Hee, W. J. Handley, M. P. Hobson, and A. N. Lasenby, Monthly Notices of the Royal Astronomical Society 455, 2461 (2015).
  • [29] B. P. Carlin and S. Chib, Journal of the Royal Statistical Society: Series B (Methodological) 57, 473 (1995).
  • [30] P. Jaranowski and A. Kr´olak, Living Reviews in Relativity 8, 3 (2005).
  • [31] R. Buscicchio, E. Roebber, J. M. Goldstein, and C. J. Moore, Physical Review D 100, 084041 (2019), arXiv:1907.11631 [astro-ph].
  • [32] N. Karnesis, S. Babak, M. Pieroni, N. Cornish, and T. Littenberg, Physical Review D 104, 043019 (2021), arXiv:2103.14598.
  • [33] T. A. Prince, M. Tinto, S. L. Larson, and J. W. Armstrong, Physical Review D 66, 122002 (2002), arXiv:grqc/0209039.
  • [34] J. Gonzalez, Y. Low, A. Gretton, and C. Guestrin, in Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, Proceedings of Machine Learning Research, Vol. 15, edited by G. Gordon, D. Dunson, and M. Dud´ık (PMLR, Fort Lauderdale, FL, USA, 2011) pp. 324–332.
  • [35] A. Gelman and D. B. Rubin, Statistical Science 7, 457 (1992), publisher: Institute of Mathematical Statistics.
  • [36] R. Storn and K. V. Price, J. Glob. Optim. 11, 341 (1997).
  • [37] S. Marsat (2025), in preparation.
  • [38] M. Vallisneri, M. Crisostomi, A. D. Johnson, and P. M. Meyers, Rapid parameter estimation for pulsar-timingarray datasets with variational inference and normalizing flows (2024), arXiv:2405.08857 [gr-qc].