arXiv subjects

Michael Pürrer

Publications and source records attributed to Michael Pürrer.

At least 37 records · Page 2Linked to original sources

arXiv2019

Regression methods in waveform modeling: a comparative study

Gravitational-wave astronomy of compact binaries relies on theoretical models of the gravitational-wave signal that is emitted as binaries coalesce. These models do not only need to be accurate, they also have to be fast to evaluate in order to be able to compare millions of signals in near real time with the data of gravitational-wave instruments. A variety of regression and interpolation techniques have been employed to build efficient waveform models, but no study has systematically compared the performance of these regression methods yet. Here we provide such a comparison of various techniques, including polynomial fits, radial basis functions, Gaussian process regression and artificial neural networks, specifically for the case of gravitational waveform modeling. We use all these techniques to regress analytical models of non-precessing and precessing binary black hole waveforms, and compare the accuracy as well as computational speed. We find that most regression methods are reasonably accurate, but efficiency considerations favour in many cases the most simple approach. We conclude that sophisticated regression methods are not necessarily needed in standard gravitational-wave modeling applications, although problems with higher complexity than what is tested here might be more suitable for machine-learning techniques and more sophisticated methods may have side benefits.

astro-ph.IM↗

arXiv2018

Surrogate model for an aligned-spin effective one body waveform model of binary neutron star inspirals using Gaussian process regression

Fast and accurate waveform models are necessary for measuring the properties of inspiraling binary neutron star systems such as GW170817. We present a frequency-domain surrogate version of the aligned-spin binary neutron star waveform model using the effective one body formalism known as SEOBNRv4T. This model includes the quadrupolar and octopolar adiabatic and dynamical tides. The version presented here is improved by the inclusion of the spin-induced quadrupole moment effect, and completed by a prescription for tapering the end of the waveform to qualitatively reproduce numerical relativity simulations. The resulting model has 14 intrinsic parameters. We reduce its dimensionality by using universal relations that approximate all matter effects in terms of the leading quadrupolar tidal parameters. The implementation of the time-domain model can take up to an hour to evaluate using a starting frequency of 20Hz, and this is too slow for many parameter estimation codes that require $O(10^7)$ sequential waveform evaluations. We therefore construct a fast and faithful frequency-domain surrogate of this model using Gaussian process regression. The resulting surrogate has a maximum mismatch of $4.5\times 10^{-4}$ for the Advanced LIGO detector, and requires 0.13s to evaluate for a waveform with a starting frequency of 20Hz. Finally, we perform an end-to-end test of the surrogate with a set of parameter estimation runs, and find that the surrogate accurately recovers the parameters of injected waveforms.

gr-qc↗

arXiv2018

Parameterized tests of the strong-field dynamics of general relativity using gravitational wave signals from coalescing binary black holes: Fast likelihood calculations and sensitivity of the method

Thanks to the recent discoveries of gravitational wave signals from binary black hole mergers by Advanced Laser Interferometer Gravitational Wave Observatory and Advanced Virgo, the genuinely strong-field dynamics of spacetime can now be probed, allowing for stringent tests of general relativity (GR). One set of tests consists of allowing for parametrized deformations away from GR in the template waveform models and then constraining the size of the deviations, as was done for the detected signals in previous work. In this paper, we construct reduced-order quadratures so as to speed up likelihood calculations for parameter estimation on future events. Next, we explicitly demonstrate the robustness of the parametrized tests by showing that they will correctly indicate consistency with GR if the theory is valid. We also check to what extent deviations from GR can be constrained as information from an increasing number of detections is combined. Finally, we evaluate the sensitivity of the method to possible violations of GR.

gr-qc↗

arXiv2017

The most powerful astrophysical events: Gravitational-wave peak luminosity of binary black holes as predicted by numerical relativity

For a brief moment, a binary black hole (BBH) merger can be the most powerful astrophysical event in the visible universe. Here we present a model fit for this gravitational-wave peak luminosity of nonprecessing quasicircular BBH systems as a function of the masses and spins of the component black holes, based on numerical relativity (NR) simulations and the hierarchical fitting approach introduced by X. Jiménez-Forteza et al. [Phys. Rev. D 95, 064024 (2017), arXiv:1611.00332]. This fit improves over previous results in accuracy and parameter-space coverage and can be used to infer posterior distributions for the peak luminosity of future astrophysical signals like GW150914 and GW151226. The model is calibrated to the l<=6 modes of 378 nonprecessing NR simulations up to mass ratios of 18 and dimensionless spin magnitudes up to 0.995, and includes unequal-spin effects. We also constrain the fit to perturbative numerical results for large mass ratios. Studies of key contributions to the uncertainty in NR peak luminosities, such as (i) mode selection, (ii) finite resolution, (iii) finite extraction radius, and (iv) different methods for converting NR waveforms to luminosity, allow us to use NR simulations from four different codes as a homogeneous calibration set. This study of systematic fits to combined NR and large-mass-ratio data, including higher modes, also paves the way for improved inspiral-merger-ringdown waveform models.

gr-qc↗

arXiv2017

Statistical Gravitational Waveform Models: What to Simulate Next?

Models of gravitational waveforms play a critical role in detecting and characterizing the gravitational waves (GWs) from compact binary coalescences. Waveforms from numerical relativity (NR), while highly accurate, are too computationally expensive to produce to be directly used with Bayesian parameter estimation tools like Markov-chain-Monte-Carlo and nested sampling. We propose a Gaussian process regression (GPR) method to generate accurate reduced-order-model waveforms based only on existing accurate (e.g. NR) simulations. Using a training set of simulated waveforms, our GPR approach produces interpolated waveforms along with uncertainties across the parameter space. As a proof of concept, we use a training set of IMRPhenomD waveforms to build a GPR model in the 2-d parameter space of mass ratio $q$ and equal-and-aligned spin $χ_1=χ_2$. Using a regular, equally-spaced grid of 120 IMRPhenomD training waveforms in $q\in[1,3]$ and $χ_1 \in [-0.5,0.5]$, the GPR mean approximates IMRPhenomD in this space to mismatches below $4.3\times 10^{-5}$. Our approach can alternatively use training waveforms directly from numerical relativity. Beyond interpolation of waveforms, we also present a greedy algorithm that utilizes the errors provided by our GPR model to optimize the placement of future simulations. In a fiducial test case we find that using the greedy algorithm to iteratively add simulations achieves GPR errors that are $\sim 1$ order of magnitude lower than the errors from using Latin-hypercube or square training grids.

astro-ph.HE↗

arXiv2017

An improved effective-one-body model of spinning, nonprecessing binary black holes for the era of gravitational-wave astrophysics with advanced detectors

We improve the accuracy of the effective-one-body (EOB) waveforms that were employed during the first observing run of Advanced LIGO for binaries of spinning, nonprecessing black holes by calibrating them to a set of 141 numerical-relativity (NR) waveforms. The NR simulations expand the domain of calibration towards larger mass ratios and spins, as compared to the previous EOBNR model. Merger-ringdown waveforms computed in black-hole perturbation theory for Kerr spins close to extremal provide additional inputs to the calibration. For the inspiral-plunge phase, we use a Markov-chain Monte Carlo algorithm to efficiently explore the calibration space. For the merger-ringdown phase, we fit the NR signals with phenomenological formulae. After extrapolation of the calibrated model to arbitrary mass ratios and spins, the (dominant-mode) EOBNR waveforms have faithfulness --- at design Advanced-LIGO sensitivity --- above $99\%$ against all the NR waveforms, including 16 additional waveforms used for validation, when maximizing only on initial phase and time. This implies a negligible loss in event rate due to modeling for these binary configurations. We find that future NR simulations at mass ratios $\gtrsim 4$ and double spin $\gtrsim 0.8$ will be crucial to resolve discrepancies between different ways of extrapolating waveform models. We also find that some of the NR simulations that already exist in such region of parameter space are too short to constrain the low-frequency portion of the models. Finally, we build a reduced-order version of the EOBNR model to speed up waveform generation by orders of magnitude, thus enabling intensive data-analysis applications during the upcoming observation runs of Advanced LIGO.

gr-qc↗

arXiv2017

Hierarchical data-driven approach to fitting numerical relativity data for nonprecessing binary black holes with an application to final spin and radiated energy

Numerical relativity is an essential tool in studying the coalescence of binary black holes (BBHs). It is still computationally prohibitive to cover the BBH parameter space exhaustively, making phenomenological fitting formulas for BBH waveforms and final-state properties important for practical applications. We describe a general hierarchical bottom-up fitting methodology to design and calibrate fits to numerical relativity simulations for the three-dimensional parameter space of quasicircular nonprecessing merging BBHs, spanned by mass ratio and by the individual spin components orthogonal to the orbital plane. Particular attention is paid to incorporating the extreme-mass-ratio limit and to the subdominant unequal-spin effects. As an illustration of the method, we provide two applications, to the final spin and final mass (or equivalently: radiated energy) of the remnant black hole. Fitting to 427 numerical relativity simulations, we obtain results broadly consistent with previously published fits, but improving in overall accuracy and particularly in the approach to extremal limits and for unequal-spin configurations. We also discuss the importance of data quality studies when combining simulations from diverse sources, how detailed error budgets will be necessary for further improvements of these already highly accurate fits, and how this first detailed study of unequal-spin effects helps in choosing the most informative parameters for future numerical relativity runs.

gr-qc↗

arXiv2017

Measuring neutron star tidal deformability with Advanced LIGO: a Bayesian analysis of neutron star - black hole binary observations

The discovery of gravitational waves (GW) by Advanced LIGO has ushered us into an era of observational GW astrophysics. Compact binaries remain the primary target sources for LIGO, of which neutron star-black hole (NSBH) binaries form an important subset. GWs from NSBH sources carry signatures of (a) the tidal distortion of the neutron star by its companion black hole during inspiral, and (b) its potential tidal disruption near merger. In this paper, we present a Bayesian study of the measurability of neutron star tidal deformability $Λ_\mathrm{NS}\propto (R/M)^{5}$ using observation(s) of inspiral-merger GW signals from disruptive NSBH coalescences, taking into account the crucial effect of black hole spins. First, we find that if non-tidal templates are used to estimate source parameters for an NSBH signal, the bias introduced in the estimation of non-tidal physical parameters will only be significant for loud signals with signal-to-noise ratios $> 30$. For similarly loud signals, we also find that we can begin to put interesting constraints on $Λ_\mathrm{NS}$ (factor of 1-2) with individual observations. Next, we study how a population of realistic NSBH detections will improve our measurement of neutron star tidal deformability. For astrophysical populations of $disruptive$ NSBH mergers, we find 20-35 events to be sufficient to constrain $Λ_\mathrm{NS}$ within $\pm 25-50\%$, depending on the chosen equation of state. In this we also assume that LIGO will detect black holes with masses within the astrophysical $mass$-$gap$. If the mass-gap remains preserved in NSBHs detected by LIGO, we estimate that $25\%$ $additional$ detections will furnish comparable tidal measurement accuracy. In both cases, we find that the loudest 5-10 events to provide most of the tidal information, thereby facilitating targeted follow-ups of NSBHs in the upcoming LIGO-Virgo runs.

gr-qc↗

arXiv2016

Fast and Accurate Inference on Gravitational Waves from Precessing Compact Binaries

Inferring astrophysical information from gravitational waves emitted by compact binaries is one of the key science goals of gravitational-wave astronomy. In order to reach the full scientific potential of gravitational-wave experiments we require techniques to mitigate the cost of Bayesian inference, especially as gravitational-wave signal models and analyses become increasingly sophisticated and detailed. Reduced order models (ROMs) of gravitational waveforms can significantly reduce the computational cost of inference by removing redundant computations. In this paper we construct the first reduced order models of gravitational-wave signals that include the effects of spin-precession, inspiral, merger, and ringdown in compact object binaries, and which are valid for component masses describing binary neutron star, binary black hole and mixed binary systems. This work utilizes the waveform model known as "IMRPhenomPv2". Our ROM enables the use of a fast reduced order quadrature (ROQ) integration rule which allows us to approximate Bayesian probability density functions at a greatly reduced computational cost. We find that the ROQ rule can be used to speed up inference by factors as high as 300 without introducing systematic bias. This corresponds to a reduction in computational time from around half a year to a half a day, for the longest duration/lowest mass signals. The ROM and ROQ rule are available with the main inference library of the LIGO Scientific Collaboration, LALInference.

gr-qc↗

arXiv2015

Measuring intermediate mass black hole binaries with advanced gravitational wave detectors

We perform a systematic study to explore the accuracy with which the parameters of intermediate-mass black-hole binary systems can be measured from their gravitational wave (GW) signatures using second-generation GW detectors. We make use of the most recent reduced-order models containing inspiral, merger and ringdown signals of aligned-spin effective-one-body waveforms (SEOBNR) to significantly speed up the calculations. We explore the phenomenology of the measurement accuracies for binaries with total masses between 50 and 500 $M_\odot$ and mass ratios between 0.1 and 1. We find that (i) at total masses below ~200 $M_\odot$, where the signal-to-noise-ratio is dominated by the inspiral portion of the signal, the chirp mass parameter can be accurately measured; (ii) at higher masses, the information content is dominated by the ringdown, and total mass is measured more accurately; (iii) the mass of the lower-mass companion is poorly estimated, especially at high total mass and more extreme mass ratios; (iv) spin cannot be accurately measured for our injection set with non-spinning components. Most importantly, we find that for binaries with non-spinning components at all values of the mass ratio in the considered range and at network signal-to-noise ratio of 15, analyzed with spin-aligned templates, the presence of an intermediate-mass black hole with mass >100 $M_\odot$ can be confirmed with 95% confidence in any binary that includes a component with a mass of 130 $M_\odot$ or greater.

astro-ph.HE↗

arXiv2015

Can we measure individual black-hole spins from gravitational-wave observations?

Measurements of black-hole spins from gravitational-wave observations of black-hole binaries with ground-based detectors are expected to be hampered by partial degeneracies in the gravitational-wave phasing: between the two component spins, and between the spins and the binary's mass ratio, at least for signals that are dominated by the binary's inspiral. Through the merger and ringdown, however, a different set of degeneracies apply. This suggests the possibility that, if the inspiral, merger and ringdown are all within the sensitive frequency band of a detector, we may be able to break these degeneracies and more accurately measure both spins. In this work we investigate our ability to measure individual spins for non-precessing binaries, for a range of configurations and signal strengths, and conclude that in general the spin of the larger black hole will be measurable (at best) with observations from Advanced LIGO and Virgo. This implies that in many applications waveform models parameterized by only one \emph{effective spin} will be sufficient. Our work does not consider precessing binaries or sub-dominant harmonics, although we provide some arguments why we expect that these will not qualitatively change our conclusions.

gr-qc↗

arXiv2015

Frequency domain reduced order model of aligned-spin effective-one-body waveforms with generic mass-ratios and spins

I provide a frequency domain reduced order model (ROM) for the aligned-spin effective-one-body (EOB) model "SEOBNRv2" for data analysis with second and third generation ground based gravitational wave (GW) detectors. SEOBNRv2 models the dominant mode of the GWs emitted by the coalescence of black hole (BH) binaries. The large physical parameter space (dimensionless spins $-1 \leq χ_i \leq 0.99$ and symmetric mass-ratios $0.01 \leq η\leq 0.25$) requires sophisticated reduced order modeling techniques, including patching in the parameter space and in frequency. I find that the time window over which the inspiral-plunge and the merger-ringdown waveform in SEOBNRv2 are connected is discontinuous when the spin of the deformed Kerr BH $χ=0.8$ or the symmetric mass-ratio $η\sim 0.083$. This discontinuity increases resolution requirements for the ROM. The ROM can be used for compact binary systems with total masses of $2 M_\odot$ or higher for the advanced LIGO (aLIGO) design sensitivity and a $10$ Hz lower cutoff frequency. The ROM has a worst mismatch against SEOBNRv2 of $\sim 1\%$, but in general mismatches are better than $\sim 0.1\%$. The ROM is crucial for key data analysis applications for compact binaries, such as GW searches and parameter estimation carried out within the LIGO Scientific Collaboration (LSC).

gr-qc↗

arXiv2015

Impact of gravitational radiation higher order modes on single aligned-spin gravitational wave searches for binary black holes

Current template-based gravitational wave searches for compact binary coalescences (CBC) use waveform models that neglect the higher order modes content of the gravitational radiation emitted, considering only the quadrupolar $(\ell,|m|)=(2,2)$ modes. We study the effect of such a neglection for the case of aligned-spin CBC searches for equal-spin (and non-spinning) binary black holes in the context of two versions of Advanced LIGO: the upcoming 2015 version, known as early Advanced LIGO (eaLIGO) and its Zero-Detuned High Energy Power version, that we will refer to as Advanced LIGO (AdvLIGO). In addition, we study the case of a non-spinning search for initial LIGO (iLIGO). We do this via computing the effectualness of the aligned-spin SEOBNRv1 ROM waveform family, which only considers quadrupolar modes, towards hybrid post-Newtonian/Numerical Relativity waveforms which contain higher order modes. We find that for all LIGO versions, losses of more than $10\%$ of events occur for mass ratio $q\geq6$ and $M \geq 100M_\odot$ due to the neglection of higher modes. Moreover, for iLIGO and eaLIGO, losses notably increase up to $(39,23)\%$ respectively for the highest mass $(220M_\odot)$ and mass ratio ($q=8$) studied. For the case of early AdvLIGO, losses of $10\%$ occur for $M>50M_\odot$ and $q\geq6$. Neglection of higher modes leads to observation-averaged systematic parameter biases towards lower spin, total mass and chirp mass. For completeness, we perform a preliminar, non-exhaustive comparison of systematic biases to statistical errors. We find that, for a given SNR, systematic biases dominate over statistical errors at much lower total mass for eaLIGO than for AdvLIGO.

gr-qc↗

arXiv2015

Frequency-domain gravitational waves from non-precessing black-hole binaries. I. New numerical waveforms and anatomy of the signal

In this paper we discuss the anatomy of frequency-domain gravitational-wave signals from non-precessing black-hole coalescences with the goal of constructing accurate phenomenological waveform models. We first present new numerical-relativity simulations for mass ratios up to 18 including spins. From a comparison of different post-Newtonian approximants with numerical-relativity data we select the uncalibrated SEOBNRv2 model as the most appropriate for the purpose of constructing hybrid post-Newtonian/numerical-relativity waveforms, and we discuss how we prepare time-domain and frequency-domain hybrid data sets. We then use our data together with results in the literature to calibrate simple explicit expressions for the final spin and radiated energy. Equipped with our prediction for the final state we then develop a simple and accurate merger-ringdown-model based on modified Lorentzians in the gravitational wave amplitude and phase, and we discuss a simple method to represent the low frequency signal augmenting the TaylorF2 post-Newtonian approximant with terms corresponding to higher orders in the post-Newtonian expansion. We finally discuss different options for modelling the small intermediate frequency regime between inspiral and merger-ringdown. A complete phenomenological model based on the present work is presented in a companion paper.

gr-qc↗

arXiv2015

Frequency-domain gravitational waves from non-precessing black-hole binaries. II. A phenomenological model for the advanced detector era

We present a new frequency-domain phenomenological model of the gravitational-wave signal from the inspiral, merger and ringdown of non-precessing (aligned-spin) black-hole binaries. The model is calibrated to 19 hybrid effective-one-body--numerical-relativity waveforms up to mass ratios of 1:18 and black-hole spins of $|a/m| \sim 0.85$ ($0.98$ for equal-mass systems). The inspiral part of the model consists of an extension of frequency-domain post-Newtonian expressions, using higher-order terms fit to the hybrids. The merger-ringdown is based on a phenomenological ansatz that has been significantly improved over previous models. The model exhibits mismatches of typically less than 1\% against all 19 calibration hybrids, and an additional 29 verification hybrids, which provide strong evidence that, over the calibration region, the model is sufficiently accurate for all relevant gravitational-wave astronomy applications with the Advanced LIGO and Virgo detectors. Beyond the calibration region the model produces physically reasonable results, although we recommend caution in assuming that \emph{any} merger-ringdown waveform model is accurate outside its calibration region. As an example, we note that an alternative non-precessing model, SEOBNRv2 (calibrated up to spins of only 0.5 for unequal-mass systems), exhibits mismatch errors of up to 10\% for high spins outside its calibration region. We conclude that waveform models would benefit most from a larger number of numerical-relativity simulations of high-aligned-spin unequal-mass binaries.

gr-qc↗

arXiv2015

Comparison of subdominant gravitational wave harmonics between post-Newtonian and numerical relativity calculations and construction of multi-mode hybrids

Gravitational waveforms which describe the inspiral, merger and ringdown of coalescing binaries are usually constructed by synthesising information from perturbative descriptions, in particular post-Newtonian theory and black-hole perturbation theory, with numerical solutions of the full Einstein equations. In this paper we discuss the "glueing" of numerical and post-Newtonian waveforms to produce hybrid waveforms which include subdominant spherical harmonics ("higher order modes"), and focus in particular on the process of consistently aligning the waveforms, which requires a comparison of both descriptions and a discussion of their imprecisions. We restrict to the non-precessing case, and illustrate the process using numerical waveforms of up to mass ratio $q=18$ produced with the BAM code, and publicly available waveforms from the SXS catalogue. The results also suggest new ways of analysing finite radius errors in numerical simulations.

gr-qc↗

arXiv2014

Frequency domain reduced order models for gravitational waves from aligned-spin compact binaries

Black-hole binary coalescences are one of the most promising sources for the first detection of gravitational waves. Fast and accurate theoretical models of the gravitational radiation emitted from these coalescences are highly important for the detection and extraction of physical parameters. Spinning effective-one-body (EOB) models for binaries with aligned spins have been shown to be highly faithful, but are slow to generate and thus have not yet been used for parameter estimation studies. I provide a frequency-domain singular value decomposition (SVD)-based surrogate reduced order model that is thousands of times faster for typical system masses and has a faithfulness mismatch of better than $\sim 0.1\%$ with the original SEOBNRv1 model for advanced LIGO detectors. This model enables parameter estimation studies up to signal-to-noise ratios (SNRs) of 20 and even up to SNR 50 for total masses below $50 M_\odot$. This article discusses various choices for approximations and interpolation over the parameter space that can be made for reduced order models of spinning compact binaries, provides a detailed discussion of errors arising in the construction and assesses the fidelity of such models.

gr-qc↗

arXiv2014

A simple model of complete precessing black-hole-binary gravitational waveforms

The construction of a model of the gravitational-wave (GW) signal from generic configurations of spinning-black-hole binaries, through inspiral, merger and ringdown, is one of the most pressing theoretical problems in the build-up to the era of GW astronomy. We present the first such model in the frequency domain, "PhenomP", which captures the basic phenomenology of the seven-dimensional parameter space of binary configurations with only three key physical parameters. Two of these (the binary's mass ratio and an effective total spin parallel to the orbital angular momentum, which determines the inspiral rate) define an underlying non-precessing-binary model. The non-precessing-binary waveforms are then "twisted up" with approximate expressions for the precessional motion, which require only one additional physical parameter, an effective precession spin, $χ_p$. All other parameters (total mass, sky location, orientation and polarisation, and initial phase) can be specified trivially. The model is constructed in the frequency domain, which will be essential for efficient GW searches and source measurements. We have tested the model's fidelity for GW applications by comparison against hybrid post-Newtonian-numerical-relativity waveforms at a variety of configurations --although we did not use these numerical simulations in the construction of the model. Our model can be used to develop GW searches, to study the implications for astrophysical measurements, and as a simple conceptual framework to form the basis of generic-binary waveform modelling in the advanced-detector era.

gr-qc↗