The recent release of the second Gravitational-Wave Transient Catalog (GWTC-2) has increased significantly the number of known GW events, enabling unprecedented constraints on formation models of compact binaries. One pressing question is to understa
nd the fraction of binaries originating from different formation channels, such as isolated field formation versus dynamical formation in dense stellar clusters. In this paper, we combine the $texttt{COSMIC}$ binary population synthesis suite and the $texttt{CMC}$ code for globular cluster evolution to create a mixture model for black hole binary formation under both formation scenarios. For the first time, these code bodies are combined self-consistently, with $texttt{CMC}$ itself employing $texttt{COSMIC}$ to track stellar evolution. We then use a deep-learning enhanced hierarchical Bayesian analysis to constrain the mixture fraction $f$ between formation models, while simultaneously constraining the common envelope efficiency $alpha$ assumed in $texttt{COSMIC}$ and the initial cluster virial radius $r_v$ assumed in $texttt{CMC}$. Under specific assumptions about other uncertain aspects of isolated binary and globular cluster evolution, we report the median and $90%$ confidence interval of three physical parameters $(f,alpha,r_v)=(0.20^{+0.32}_{-0.18},2.26^{+2.65}_{-1.84},2.71^{+0.83}_{-1.17})$. This simultaneous constraint agrees with observed properties of globular clusters in the Milky Way and is an important first step in the pathway toward learning astrophysics of compact binary formation from GW observations.
Primordial black holes (PBHs) might be formed in the early Universe and could comprise at least a fraction of the dark matter. Using the recently released GWTC-2 dataset from the third observing run of the LIGO-Virgo Collaboration, we investigate whe
ther current observations are compatible with the hypothesis that all black hole mergers detected so far are of primordial origin. We constrain PBH formation models within a hierarchical Bayesian inference framework based on deep learning techniques, finding best-fit values for distinctive features of these models, including the PBH initial mass function, the fraction of PBHs in dark matter, and the accretion efficiency. The presence of several spinning binaries in the GWTC-2 dataset favors a scenario in which PBHs accrete and spin up. Our results indicate that PBHs may comprise only a fraction smaller than $0.3 %$ of the total dark matter, and that the predicted PBH abundance is still compatible with other constraints.
Computing signal-to-noise ratios (SNRs) is one of the most common tasks in gravitational-wave data analysis. While a single SNR evaluation is generally fast, computing SNRs for an entire population of merger events could be time consuming. We compute
SNRs for aligned-spin binary black-hole mergers as a function of the (detector-frame) total mass, mass ratio and spin magnitudes using selected waveform models and detector noise curves, then we interpolate the SNRs in this four-dimensional parameter space with a simple neural network (a multilayer perceptron). The trained network can evaluate $10^6$ SNRs on a 4-core CPU within a minute with a median fractional error below $10^{-3}$. This corresponds to average speed-ups by factors in the range $[120,,7.5times10^4]$, depending on the underlying waveform model. Our trained network (and source code) is publicly available at https://github.com/kazewong/NeuralSNR, and it can be easily adapted to similar multidimensional interpolation problems.
We combine hierarchical Bayesian modeling with a flow-based deep generative network, in order to demonstrate that one can efficiently constraint numerical gravitational wave (GW) population models at a previously intractable complexity. Existing tech
niques for comparing data to simulation,such as discrete model selection and Gaussian process regression, can only be applied efficiently to moderate-dimension data. This limits the number of observable (e.g. chirp mass, spins.) and hyper-parameters (e.g. common envelope efficiency) one can use in a population inference. In this study, we train a network to emulate a phenomenological model with 6 observables and 4 hyper-parameters, use it to infer the properties of a simulated catalogue and compare the results to the phenomenological model. We find that a 10-layer network can emulate the phenomenological model accurately and efficiently. Our machine enables simulation-based GW population inferences to take on data at a new complexity level.
We report on advances to interpret current and future gravitational-wave events in light of astrophysical simulations. A machine-learning emulator is trained on numerical population-synthesis predictions and inserted into a Bayesian hierarchical fram
ework. In this case study, a modest but state-of-the-art suite of simulations of isolated binary stars is interpolated across two event parameters and one population parameter. The validation process of our pipelines highlights how omitting some of the event parameters might cause errors in estimating selection effects, which propagates as systematics to the final population inference. Using LIGO/Virgo data from O1 and O2 we infer that black holes in binaries are most likely to receive natal kicks with one-dimensional velocity dispersion $sigma$ = 105+44 km/s. Our results showcase potential applications of machine-learning tools in conjunction with population-synthesis simulations and gravitational-wave data.
Cunha et al. (2018) recently reexamined the possibility of detecting gravitational waves from exoplanets, claiming that three ultra-short period systems would be observable by LISA. We revisit their analysis and conclude that the currently known exop
lanetary systems are unlikely to be detectable, even assuming a LISA observation time $T_{rm obs}=4$ yrs. Conclusive statements on the detectability of one of these systems, GP Com b, will require better knowledge of the systems properties, as well as more careful modeling of both LISAs response and the galactic confusion noise. Still, the possibility of exoplanet detection with LISA is interesting enough to warrant further study, as gravitational waves could yield dynamical properties that are difficult to constrain with electromagnetic observations.
Light bosons, proposed as a possible solution to various problems in fundamental physics and cosmology, include a broad class of candidates for beyond the Standard Model physics, such as dilatons and moduli, wave dark matter and axion-like particles.
If light bosons exist in nature, they will spontaneously form clouds by extracting rotational energy from rotating massive black holes through superradiance, a classical wave amplification process that has been studied for decades. The superradiant growth of the cloud sets the geometry of the final black hole, and the black hole geometry determines the shape of the cloud. Hence, both the black hole geometry and the cloud encode information about the light boson. For this reason, measurements of the gravitational field of the black hole/cloud system (as encoded in gravitational waves) are over-determined. We show that a single gravitational wave measurement can be used to verify the existence of light bosons by model selection, rule out alternative explanations for the signal, and measure the boson mass. Such measurements can be done generically for bosons in the mass range $[10^{-16.5},10^{-14}]$ eV using LISA observations of extreme mass-ratio inspirals.
We show how LIGO is expected to detect coalescing binary black holes at $z>1$, that are lensed by the intervening galaxy population. Gravitational magnification, $mu$, strengthens gravitational wave signals by $sqrt{mu}$, without altering their frequ
encies, which if unrecognised leads to an underestimate of the event redshift and hence an overestimate of the binary mass. High magnifications can be reached for coalescing binaries because the region of intense gravitational wave emission during coalescence is so small ($sim$100km), permitting very close projections between lensing caustics and gravitational-wave events. Our simulations incorporate accurate waveforms convolved with the LIGO power spectral density. Importantly, we include the detection dependence on sky position and orbital orientation, which for the LIGO configuration translates into a wide spread in observed redshifts and chirp masses. Currently we estimate a detectable rate of lensed events rateEarly{}, that rises to rateDesign{}, at LIGOs design sensitivity limit, depending on the high redshift rate of black hole coalescence.
