Davide Gerosa

When inferring black holes from gravitational-wave data, we tend to do two things, one after the other. First, we consider each event individually and measure its parameters (masses, spins, etc). Then we consider all the events together and measure the population properties. This is what we do all the time, but, actually, if objects are now part of a population, those parameters should be looked at again in light of all the others. This full problem (all parameters of all the events plus the population parameters) is daunting, and in the past we used an indirect and somewhat convoluted approach. We got back to it now, and this time, we managed to do it head-on. Let me introduce this giant 500-dimensional sampling of the full population problem!

Michele Mancarella, Davide Gerosa.
arXiv:2502.12156 [gr-qc].

Long gravitational-was signals are, well, long. And long often means painful, as more data need to be stored and processed. Kind of intuitively, the solution might be that of cutting things into chunks, so that long becomes short. Here we apply this idea to the popular inner product entering all gravitational-wave pipelines; this is a key building block of everything we do. The answer is that using SFTs, “Short-time Fourier Transforms”, can make things faster by more than 3 orders of magnitudes, sometimes 5. We think this is the solution to future gravitational-wave data analysis problems (think LISA and 3G…).

Rodrigo Tenorio, Davide Gerosa.
arXiv:2502.11823 [gr-qc].

General relativity has this beautiful property that coordinates are meaningless. You can change them at will, which means they don’t contain any physics. And, believe it or not, some of the popular formulations we use to write down the dynamics of eccentric binary black holes still have coordinates in them. They go away if you take an average of an orbit (Peters, the man!) but that’s killing some information. In this paper we go back to those old results and show how those gauges can actually be absorbed into the formulation itself. The paper is on the maths-heavy side of things, but the results are great. Peters, you were basically right, but not quite.

Giulia Fumagalli, Nicholas Loutrel, Davide Gerosa, Matteo Boschini.
arXiv:2502.06952 [gr-qc].

Quasar 3C 186 strikes back! Matteo and I got interested in this funny quasar last year (see this one). When our paper hit the arxiv, we got contacted by the real astronomers who take actual data, who told us they had even more beautiful data. We ended up contributing with our relativistic model and… well… everything seems to work. 3C 186 is indeed a recoiling black hole (it might be a rare one, but we’ve observed it nonetheless). The abstract says “decisive,” and this is indeed the right word.

Marco Chiaberge, Takahiro Morishita, Matteo Boschini, Stefano Bianchi, Alessandro Capetti, Gianluca Castignani, Davide Gerosa, Masahiro Konishi, Shuhei Koyama, Kosuke Kushibiki, Erini Lambrides, Eileen T. Meyer, Kentaro Motohara, Massimo Stiavelli, Hidenori Takahashi, Grant R. Tremblay, Colin Norman.
arXiv:2501.18730 [astro-ph.GA].

Gravitational-wave population people talk all the time about parametric vs non-parametric methods. Parametric methods mean imposing our astrophysical knowledge on how we look at GW data. This is great, we do want to extract astrophysical knowledge, but what if we don’t know what to look for? The statisticians tell us to go non-parametric, which means using a flexible model that can fit whatever you want. That’s great, but what do we learn then? In other words, where’s the boundary between flexibility and interpretability? Today’s paper shows that one can conceptually separate these two processes and extract parametric results from non-parametric fits. I’m very proud of this piece of work, which was Cecilia Fabbri‘s MSc thesis project and was actually kickstarted by one of my previous students, Alessandro Santini. We even wrote a poem about this!

Cecilia Maria Fabbri, Davide Gerosa, Alessandro Santini, Matthew Mould, Alexandre Toubiana, Jonathan Gair.
arXiv:2501.17233 [astro-ph.HE].

Our “popfisher” paper is finally out! (and now Viola can submit her PhD thesis). This is about next-generation (aka 3G) gravitational wave detectors. Those beasts will measure millions of black holes… and with so many of them who cares about each source individually. The important thing will be the population of objects, i.e. how those black holes are distributed as a whole. Measuring populations is an interesting but convoluted statistical problem. Here we implement a quick shortcut (the Fisher matrix) and show that yes, 3G detectors will be amazing… but more amazing for some things than for others.

Viola De Renzis, Francesco Iacovelli, Davide Gerosa, Michele  Mancarella, Costantino Pacilio.
arXiv:2410.17325 [astro-ph.HE].

LISA will see a gazillion white dwarfs, but we won’t, or at least not individually. Those signals will actually pile up together in a mashed potato thing called foreground. But this mashed potato won’t be smooth (translate: the gravitational-wave signal won’t be stationary and Gaussian) and this structure can indeed be precious for extracting more information from LISA. But first, let’s taste this with today’s paper, i.e. characterize the foreground.

Riccardo Buscicchio, Antoine Klein, Valeriya Korol, Francesco Di Renzo, Christopher J. Moore, Davide Gerosa, Alessandro Carzaniga.
arXiv:2410.08263 [astro-ph.HE].

ps. This started as the student project of Alessandro Carzaniga, great it’s finally out!

All right I think this is great (but it took me a long time to convince myself and the others that’s the case!) In gravitational-wave astronomy we measure binaries, that is, pairs of two objects. Our signals have information about the pair as a whole. At the same time, we care very much about separating those two objects and measuring the properties of individual black holes and neutron stars. We always do that operation without thinking twice, just say that for each posterior sample object “1” is that with the larger mass and object “2” is that with the lower mass. But is that ok? Surely it’s a choice, but is it the best one? What does it even mean to pick the “best” labels? I think machine learning can help us here and that this problem can be framed using the language of semi-supervised clustering. The results? Well, they seem very significant. Measurements of the black-hole spins are more accurate, you can tell more easily if that’s a black hole or a neutron star, and overall the posterior distributions just look nicer (go away nasty multimodalities and non-Gaussianities!).

Davide Gerosa, Viola De Renzis, Federica Tettoni, Matthew Mould, Alberto Vecchio, Costantino Pacilio
arXiv:2409.07519 [gr-qc].