Search for: All records

We develop methods to more efficiently differentiate between gravitational wave signals from binary mergers, and detector noise. We make use of the PyCBC detection pipeline to compile larger amounts of data, including signal and noise, into SNR density plots, and we modified them so that they could be easily interpreted by an image classifier. After selecting the parameters that demonstrated features in the density plots, we created a convolutional neural network to search for these patterns. We trained and tested the neural network over increasingly large and varied data sets. 

In the post-Newtonian regime, the time it takes two black holes to orbit each other is much shorter than the time it takes their spins and the orbital angular momentum to precess about the direction of the total angular momentum, which in turn is shorter than the orbital decay time. We use the parameters quantifying the component black hole spins in and out of the orbital plane to build an interactive 3D visualization to explore the phenomenology of spin precession over these different time scales. 

Transient noise, called "glitches," can mimic and obscure real gravitational waves in the strain data channel. One machine learning software package used to classify these glitches and identify their sources, GravitySpy, is successful when the spectrogram of the glitch has a very distinct and unique shape. However, one of the most common types of glitches, called a "blip," has an indistinct shape due to so few cycles being in-band, and tends to ring off template signals of binary black hole mergers, making it especially necessary to eliminate blips for future observing runs. Here we examine blip glitches in a Q-transform spectrogram with different parameters than those used by GravitySpy to determine if there are sub-classifications of blips that might have identifiable sources, and then use Convolutional Neural Networks to sub-classify these blips. The implementation of Convolutional Neural Networks has provided compelling evidence of distinguishable differences between these hypothesized sub-classes. 

We investigate the feasibility of using trees as a seismic meta-material that could shield the LIGO detectors from seismic activity. This seismic cloak would reflect low frequency surface waves away from the detector, thereby increasing the sensitivity of the detectors. This study models the energy transfer from surface waves as they pass through the bandgap filters designed from trees in different arrangements. The attenuation and rejection will hopefully serve to cloak the LIGO detectors from seismic activity. This work could have future impact on high sensitivity detectors, leading to more detections of merger events. 

The detectable component of gravitational waves, known as the oscillatory waveform, is predicted to have a smaller, lower frequency counterpart called the memory: a permanent warping of space-time. The memory component is low-frequency (below the usual LIGO frequency band starting at 20 Hz), and low amplitude. Low frequency noise sources on earth make it difficult for ground based detectors to reach the SNR (signal to noise ratio) needed to detect this component. We use Bayesian parameter estimation on simulated events with future detector sensitivities, to determine the detector noise spectrum, event masses, and detected SNR required to detect gravitational wave memory. 

The LIGO detectors are susceptible to high magnitude teleseismic events such as earthquakes, which can disrupt proper functioning, operation and significantly reduce their duty cycle. With advanced warning of impeding tremors, the impact can be suppressed in the isolation system and the down time can be reduced at the expense of increased instrumental noise. An earthquake early- warning system has been developed relying on near real-time earthquake alerts provided by the U.S. Geological Survey (USGS) and the National Oceanic and Atmospheric Administration (NOAA). The alerts can be used to estimate arrival times and ground velocities at the gravitational-wave detectors. We use machine learning algorithms to develop a prediction model and control strategy has to reduce LIGO downtime. 

This project is focused on improving the optics in LIGO by characterizing mirror figure error that contribute to optical losses. We develop a method to measure surface deformations with in-Situ mode spectroscopy, measuring the resonant frequencies of the higher order Hermite Gaussian modes resonant in LIGO's Fabry-Perot cavities, that are shifted from their ideal spacings due to those deformations. We use this information to construct mirror phase maps. that characterize the figure error.