The Virgo interferometer was constructed to identify gravitational waves that were theorized by the general theory of relativity. It is a large Michelson interferometer that is impervious to external disturbances, with its mirrors and instrumentation being suspended and its laser beam operating within a vacuum. The device's two arms span 3 kilometers and are situated in Santo Stefano a Macerata, in proximity to the city of Pisa, Italy.
The European Gravitational Observatory (EGO), a consortium established by the French CNRS and Italian INFN, houses the Virgo instrument. More than 650 members make up the Virgo Collaboration, which runs the detector, representing 119 institutions in 14 countries. CIEMAT is a member of Virgo since July 2022.
Our commitments in Virgo are:
- Participate in the data analysis of run O4.
- Develop and deploy an off-site computing cluster for low-latency data analysis.
- Design, build and install new payload hardware providing tuning of the Radius of Curvature (ROC) of the mirror of the Interferometer Mode Cleaner (IMC) of Virgo for O5.
- Design, produce and install stray light baffles in the output port tower of the Virgo interferometer.
Data analysis of run O4
Our activities belong to Virgo’s Searches for Signals from Compact Binary Coalescences analysis effort, and we are focusing on three main aspects:
- Search for primordial black holes as dark matter candidates: black hole binaries (or other compact object binaries) with component masses below the Chandrasekar mass constitute a speculative source of dark matter candidates suggested by inflation models. Such systems might possibly constitute some fraction of the dark matter. A search for sub-solar mass binaries could reveal the existence of a new class of object, or place stronger constraints on the fraction of dark matter explained by sub-solar mass black hole binaries. This goal is pursued in collaboration with the IFT Virgo group, which is at the forefront of the theoretical and phenomenological groups proposing such a possibility.
- Measuring the Hubble constant using gravitational waves: by observing extra-galactic binary coalescences both electromagnetically and through gravitational waves, and measuring the redshift of the host galaxy, an estimate of the Hubble constant can be obtained. As more observations are made, this method is expected to become a competitive and independent way to measure it, aiming to provide further information to hopefully solve the current tension between local and cosmological measurements. In addition, the measurement can also be performed in the absence of identified counterparts by using a statistical approach involving spatial correlations with a galaxy catalogue.
- Development and testing of Artificial intelligence tools toward improving the understanding of data and the template banks: we aim to use explainable artificial intelligence algorithms (XAI) in the analysis of gravitational waves. Currently, AI-based decisions are not transparent to humans, but by using XAI, the decision-making process of the AI can be characterised and the input data that is important for a specific decision made by a neural network can be identified and visualised. This potentially allows for new understanding and knowledge of the detector, which can be used to improve all the analysis chains. The focus of the research is on using XAI to identify detector signals and distinguish them from noise captured by detectors, which is an essential tool for avoiding biased learning in Neural Networks.
We also plan to use Generative Neural Networks (GAN) able to generate realistic synthetic images. We propose to exploit the GAN capability to generate realistic objects for replacing expensive simulations for the emission due to dynamical captures in clusters of black holes. This has the additional impact of reducing the carbon footprint of simulation of low-probability events.
As a transversal activity, understanding noise sources and mitigating their effects is key for the detection of gravitational-wave signals, which involves both hardware and software efforts. In the context of data analysis, we foresee an important activity on detector characterisation, in particular using the aforementioned XAI techniques. The hardware side of this task is an integral part of the detector commissioning for run O4. We will participate in the detector commissioning at the EGO site to measure the several noise transfer functions that will eventually be incorporated into the noise model of the interferometer. These activities will be performed in collaboration with IFT.
Off-site computing cluster for low-latency data analysis
The low latency pipeline of Virgo is a system for quickly processing and transmitting data from the Virgo interferometer. It is used to identify potential gravitational wave signals and generate alerts that can be sent to other observatories for confirmation and further study. The low latency pipeline is essential for enabling rapid follow-up observations of gravitational wave events, which can help scientists better understand these phenomena' sources and the universe's nature. During the O4 run in 2023, and especially during the O5 one starting in 2026, the detection rate of the Virgo interferometer is expected to increase significantly. In order to cope with this increase, it has been proposed to develop off-site low-latency pipeline nodes to be added to the interferometer computing infrastructure. In agreement with Virgo, we will develop one of such nodes at CIEMAT and PIC facilities. This will be done in two steps: in the first one, an engineering model will be developed with Virgo experts and deployed and maintained at CIEMAT headquarters. Once the requirements on the node are fulfilled, it will be replicated and upgraded at PIC to provide its services for O5. So far the requirements and specifications for the hardware of the node have been defined together with PIC and Virgo experts, as well as the location for its installation at CIEMAT.
New payload hardware providing tuning of the Radius of Curvature of the mirror of the Interferometer Mode Cleaner
The IMC is a 300 m resonance cavity that uses a system of mirrors to clean the laser beam of the Virgo interferometer of spurious modes. The translational and rotational degrees of freedom of the mirror of the main payload of the IMC are controlled using electromagnetic actuators. However, contrary to other mirrors in the interferometer, no tuning ROC has been implemented in the IMC, which could lead to improvements in the stability and sensibility of the interferometer for O5. Similarly to other mirrors, we plan to implement the control of the ROC by using a ring heater mounted around the perimeter of the mirror that produces heat to change the shape of the mirror. The main challenge in this task is to design a support structure that keeps the static and dynamic properties of the current payload whereas it adds this ring heater. We have already started the design of the mechanical structure holding the mirror and the ring heater together with Virgos’ engineers, and have identified the necessary resources for its production. The plan is to be ready to produce the mechanics by the end of 2023 and to proceed with the first integration tests by mid-2024.
Stray light baffles in the output port tower of the Virgo interferometer
Stray light is present in the interferometer due to reflections and other sources of contamination. This stray light can interfere with the laser beam, reducing the quality of the signal and degrading the interferometer's sensitivity. In particular, light scattered from the output optics towards the vacuum chamber and then scattered back recombines with the beam and produces spurious signals. This is sensitive to environmental noise since it induces a signal correlated at least with acoustic and seismic noise. In order to reduce stray light, baffles coated with special materials that reduce reflections and scattering of light are installed around the optical path of the laser beam or on surfaces where reflection is expected to take place. Such baffles are not available in the chamber at the output port due to the lack of space for the process installation, which requires a special design. In agreement with Virgo experts, and profiting from our experience with space mechanics, we have taken responsibility for the design and production of such baffles. The design has started and we have identified the requirements regarding the reflectivity of the surfaces. We plan to identify the technology and company for the coating by mid-2023 and start production by the beginning of 2024.