by Serena Vinciguerra
A neuroscience perspective on the gravitational wave community.
INSIDE OUT is not only a Pixar cartoon, but also a very intelligent slogan. I am not talking about emotions, but more generally about our brain. A more common view of our brain might be OUTSIDE IN: we use the brain to interpret the inputs we receive from outside. However, the brain is also the most powerful computer ever known, so why not try the INSIDE OUT modality, and be inspired by our brains as computational models?
The brain is a biological network composed of nerve cells (neurons) connected to each other. We can imagine neurons as calculation units which compute a weighted sum of the received electric inputs. If this sum reaches a particular threshold, a new electric signal is generated, propagated and finally transmitted to other neurons.
Artificial neural networks (ANNs) and their success clearly represent the strength of applying the mechanisms which drive our mind to other subjects. ANNs find many applications in research, including in the science of gravitational waves (GW). In searches for un-modelled GW transients, ANNs have been used to classify noisy events, to search for GWs associated with short gamma ray bursts as well as for signal classification. What are the eyes, the ears, the nose and the mouth which make up an identifiable face in GW transients or glitches? These are the kind of questions ANNs have to answer to perform classification/identification tasks. To find out how good they are, take a look to these papers [1, 2]
The Gravitational Wave Brain
Different brain areas are devoted to different tasks. Inside the LIGO-Virgo collaboration, the electromagnetic (EM) group is dedicated to facilitate the exchange of information between the collaboration and the astronomy partners who have signed a Memorandum of Understanding. The final aim is to collect the available inputs (electromagnetic and gravitational) and develop a coherent multi-messenger astronomy. In our GW brain, we can think of the EM group as a part of the sensory nervous system, the apparatus which leads the direct connections with partners outside the GW community. In our brain, as in the GW community, specialised areas are however not isolated, but more like clusters of neurons in a bigger network. Networks are crucial, in our brains as well as in research. Networks can be classified in different ways. Some networks are proximity-based, like the people working in the same institute, other are type-based. Together with 13 other PhD students, investigating GW related topics, we form the Initial Training Network, funded by European Commission under FP7-Marie Curie Actions, GraWIToN. We are -almost- newly born neurons in the relatively mature GW brain (neurogenesis is present also in adult brains). We specialise on our research topics, but we often have the possibility of interacting and enriching ourself not only from a “declarative” knowledge point of view, but also from the diversity of our reality.
Interestingly our mind has the ability of changing the reality. Inputs can be interpreted differently if we change our prior model, this is how we update our conception of reality. This ability of physically changing our mind is called neuroplasticity. Aware or not, neuroplascticity is present in research too. Just think of the impact of the first discovery of GWs. GW150914 has radically modified the way we look at Universe, for example many studies have changed the reference black hole binary from a 10-10 to a 30-30 solar masses. This has motivated several new investigations, such as studies on the EM emissions generated by black hole mergers. Internally to our GW brain, we have now focused more attention away from neutron star – neutron star and towards black hole – black hole systems.
Another property that heavily impacts the brain activity is the speed of the signal transmissions. In our brain, as well as in GW community, different kinds of information are processed at different speeds. For example, since they have very different goals (detection versus source characterisation), online analyses for GW transients are much faster than pipelines devoted to the parameter estimation (PE).
Online pipelines, in terms of speed, process the data as our auditory system, while off-line PE pipelines as our visual system. This explains why, despite the speed of light travelling much faster than the speed of sound, in races a sound (gun shoot) marks the start of the competition: our response to sound inputs is much faster than our response to photons. But for GW PE, we are competing against time, especially in view of EM follow ups, so for equal analysis quality, the faster, the better. Many studies have therefore been devoted to speeding up the PE analyses. Among them, I have explored an optimisation of the sampling procedure for the GW waveform computation, one of the most computationally expensive operation of the Bayes analysis at the base of most PE pipelines. Our analysis suppresses unnecessary steps, such as generating waveform samples at low frequencies, similarly to myelin, substance which reduces the electrical conductivity of some axon’s areas, forcing signals to propagate through the faster saltatory conduction.
Take a look at our paper  to discover more about our method and results! Connect, interact and be inspired!
 Vinciguerra et al, 2017 Class. Quantum Grav.
 S. Rampone et al, 2013 International Journal of Modern
Physics C, vol 24, n 11,1350084
 Vinciguerra et al 2017 Class. Quantum Grav.
Read the full article in Classical and Quantum Gravity:
Accelerating gravitational wave parameter estimation with multi-band template interpolation
Serena Vinciguerra et al 2017 Class. Quantum Grav. 34 115006
Sign up to follow CQG+ by entering your email address in the ‘follow’ box at the foot of this page for instant notification of new entries.
CQG papers are selected for promotion based on the content of the referee reports. The papers you read about on CQG+ have been rated ‘high quality’ by your peers.
This work is licensed under a Creative Commons Attribution 3.0 Unported License.