Searching for long-duration transient gravitational waves from spinning neutron stars

Resumo

While gravitational wave (GW) detections from compact binary coalescences are becoming routine, with 90 events detected by the LIGO—Virgo—KAGRA collaborations throughout the first three observing runs, GW signals from individual neutron stars (NSs) are yet to be discovered. Pulsars, i.e. rotating NSs which emit electromagnetic beams, are incredibly dense astronomical objects which could also emit long-lasting, quasi-monochromatic continuous waves (CWs). Most of the observed pulsars are very stable clocks, but some exhibit anomalies in their evolution called glitches, consisting of a sudden increase in the rotational frequency followed by a relaxation phase with timescales from days to months. Pulsar glitches are one of the few instances during which we can indirectly examine the interior of a NS. These rare phenomena could trigger GW emission, similar to CWs but limited in time, namely transient continuous waves (tCWs). The main subject and objective of this thesis is the study and development of methods to search for tCWs from glitching pulsars. The first part comprises three introductory chapters providing both physical phenomenology and methodological means to understand these types of searches: GW generation and propagation, description of pulsars and characterization of pulsar glitches, and finally the delineation of different methods used in this thesis to search for tCWs. In the second part I present original scientific results separated into three chapters. The first one is the study of the prospects to detect these types of signals from glitching pulsars during the next LIGO—Virgo—KAGRA observing runs and with third generation detectors, Einstein Telescope and Cosmic Explorer. The analysis consisted in collecting data from pulsar glitch catalogues, computing the indirect energy upper limit for each glitch and comparing these values to the sensitivities of different detectors. The results are promising, especially for third generation detectors which would be able to detect (or at least physically constrain) 35—40% of past glitches thanks to their expected improved sensitivity. The second results chapter covers the results of a tCW search based on the existing F-statistic matched-filtering method, on LIGO—Virgo data during the third observing run. Two outliers coming from the pulsar J0537—6910 were reported, but after thorough follow-up inspection no confident detection could be claimed. For each glitch, upper limits on the GW amplitude as a function of the duration of the signal were set, and none surpass the physical indirect upper limit of the corresponding glitch. Finally, in the last results chapter I develop and present the results of a tCW search partially based on deep learning. For this search I designed and implemented convolutional neural networks which receive as input intermediate matched-filter quantities, called “F-statistic atoms”, and which reach sensitivities similar to those of standard detection statistics at a lower computational cost.