Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userA robust machine learning algorithm to search for continuous gravitational waves
376
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
Many continuous gravitational wave searches are affected by instrumental spectral lines that could be confused with a continuous astrophysical signal. Several techniques have been developed to limit the effect of these lines by penalising signals that appear in only a single detector. We have developed a general method, using a convolutional neural network, to reduce the impact of instrumental artefacts on searches that use the SOAP algorithm. The method can identify features in corresponding frequency bands of each detector and classify these bands as containing a signal, an instrumental line, or noise. We tested the method against four different data-sets: Gaussian noise with time gaps, data from the final run of Initial LIGO (S6) with signals added, the reference S6 mock data challenge data set and signals injected into data from the second advanced LIGO observing run (O2). Using the S6 mock data challenge data set and at a 1% false alarm probability we showed that at 95% efficiency a fully-automated SOAP search has a sensitivity corresponding to a coherent signal-to-noise ratio of 110, equivalent to a sensitivity depth of 10 Hz$^{-1/2}$, making this automated search competitive with other searches requiring significantly more computing resources and human intervention.
rate research
Read More
We present an improved method of targeting continuous gravitational-wave signals in data from the LIGO and Virgo detectors with a higher efficiency than the time-domain Bayesian pipeline used in many previous searches. Our spectral interpolation algorithm, SplInter, removes the intrinsic phase evolution of the signal from source rotation and relative detector motion. We do this in the frequency domain and generate a time series containing only variations in the signal due to the antenna pattern. Although less flexible than the classic heterodyne approach, SplInter allows for rapid analysis of putative signals from isolated (and some binary) pulsars, and efficient follow-up searches for candidate signals generated by other search methods. The computational saving over the heterodyne approach can be many orders of magnitude, up to a factor of around fifty thousand in some cases, with a minimal impact on overall sensitivity for most targets.
We consider a machine learning algorithm to detect and identify strong gravitational lenses on sky images. First, we simulate different artificial but very close to reality images of galaxies, stars and strong lenses, using six different methods, i.e. two for each class. Then we deploy a convolutional neural network architecture to classify these simulated images. We show that after neural network training process one achieves about 93 percent accuracy. As a simple test for the efficiency of the convolutional neural network, we apply it on an real Einstein cross image. Deployed neural network classifies it as gravitational lens, thus opening a way for variety of lens search applications of the deployed machine learning scheme.
All-sky surveys for isolated continuous gravitational waves present a significant data-analysis challenge. Semicoherent search methods are commonly used to efficiently perform the computationally-intensive task of searching for these weak signals in the noisy data of gravitational-wave detectors such as LIGO and Virgo. We present a new implementation of a semicoherent search method, Weave, that for the first time makes full use of a parameter-space metric to generate banks of search templates at the correct resolution, combined with optimal lattices to minimize the required number of templates and hence the computational cost of the search. We describe the implementation of Weave and associated design choices, and characterize its behavior using semi-analytic models.
We present a comprehensive study of the effectiveness of Convolution Neural Networks (CNNs) to detect long duration transient gravitational-wave signals lasting $O(hours-days)$ from isolated neutron stars. We determine that CNNs are robust towards signal morphologies that differ from the training set, and they do not require many training injections/data to guarantee good detection efficiency and low false alarm probability. In fact, we only need to train one CNN on signal/noise maps in a single 150 Hz band; afterwards, the CNN can distinguish signals/noise well in any band, though with different efficiencies and false alarm probabilities due to the non-stationary noise in LIGO/Virgo. We demonstrate that we can control the false alarm probability for the CNNs by selecting the optimal threshold on the outputs of the CNN, which appears to be frequency dependent. Finally we compare the detection efficiencies of the networks to a well-established algorithm, the Generalized FrequencyHough (GFH), which maps curves in the time/frequency plane to lines in a plane that relates to the initial frequency/spindown of the source. The networks have similar sensitivities to the GFH but are orders of magnitude faster to run and can detect signals to which the GFH is blind. Using the results of our analysis, we propose strategies to apply CNNs to a real search using LIGO/Virgo data to overcome the obstacles that we would encounter, such as a finite amount of training data. We then use our networks and strategies to run a real search for a remnant of GW170817, making this the first time ever that a machine learning method has been applied to search for a gravitational wave signal from an isolated neutron star.
By now, tens of gravitational-wave (GW) events have been detected by the LIGO and Virgo detectors. These GWs have all been emitted by compact binary coalescence, for which we have excellent predictive models. However, there might be other sources for which we do not have reliable models. Some are expected to exist but to be very rare (e.g., supernovae), while others may be totally unanticipated. So far, no unmodeled sources have been discovered, but the lack of models makes the search for such sources much more difficult and less sensitive. We present here a search for unmodeled GW signals using semi-supervised machine learning. We apply deep learning and outlier detection algorithms to labeled spectrograms of GW strain data, and then search for spectrograms with anomalous patterns in public LIGO data. We searched $sim 13%$ of the coincident data from the first two observing runs. No candidates of GW signals were detected in the data analyzed. We evaluate the sensitivity of the search using simulated signals, we show that this search can detect spectrograms containing unusual or unexpected GW patterns, and we report the waveforms and amplitudes for which a $50%$ detection rate is achieved.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
