Transient gravitational waves at r-mode frequencies from neutron stars

A search for long transient gravitational waves associated with neutron stars is presented. The estimated length of these sources is from hours to weeks. Two types of astrophysical sources are considered: pulsar timing glitches associated with r-modes oscillations in the interior of isolated neutron stars, and Type I X-Ray bursts in neutron stars from binary systems. These signals follow the model of an e-folding sinusoid signal with a duration dependant on dissipation processes in the interior of the neutron stars and the gravitational radiation reaction. Estimations of the timescales of gravitational wave signals emitted by stable stars are presented. From this study, it is concluded that detecting signals from faster spinning neutron stars is more feasible than from slower neutron stars. The study of this type of transient gravitational wave signals is explored for the first time using an adaptation of the F-statistic gravitational wave search method used regularly in continuous gravitational wave searches. This adaptation, proposed by Prix et al, is a search methodology in which the duration of a signal plays a significant role in its detection. This code is part of the LAL/LAL-apps data analysis algorithm libraries of the LIGO and VIRGO scientific collaborations (LVC). The use of this method in the gravitational wave search presented in this thesis was implemented in two different environments: gaussian noise data and data in gravitational wave detector-like noise. For the latter, injections of long transient signals with durations ∼ 10,000 s on the LVC Engineering Run 3 were done. A comparison between the results obtained in these two studies is presented. It shows that, by having a good characterisation of unwanted noise lines, it is possible to distinguish the frequency of the injected signal within a small search band of only a few frequency bins. On the other hand, the recovery of the duration of the signal would require a broad search band over time. This estimation is set to be approximately ±τ, where τ is the damping time of the injected signal, in order to construct a complete τ distribution. For example, for a signal that last ∼ 3.5 days, an total τ interval of ∼ 6.5 − 7 days is required