TV-GWD Detects

High Frequency

Gravitational Waves ?

That’s the headline I would like to see eventually (but alas, not yet). After the LIGO Scientific Collaboration announced that LIGO detected a neutron star collision GW190425, I decide to do some work. The gif animation of the spectrogram they displayed on their website inspired me to create my own animations from data collected from my new version 6 TV-GWD. I captured 0.1 seconds of data and then did a CWT (Continuous Wavelet Transform) on a random subset of that data just to show that, given a sufficient amount of motivation, bias and CPU time, I could easily find chirp signals out of purely random noise. I created GIF animations that highlights three chirp signals. The first is a 4.7 usec. chirp with a frequency sweep of 2.5 MHz to 8 MHz, the second is a 2.2 usec. chirp with a sweep of 6 MHz to 15 MHz, and the third is a 14 usec. chirp sweeping from 1 MHz to 10 MHz.


Maybe I lucked-out and detected the merger of miniature, primordial black holes left over from the Big-bang, or dark matter particles colliding, or maybe I just see patterns in random noise because I want to!


Also… It may have something to do with the fact that Spectrograms and CWTs are plotted with the y axis displayed using a log scale so that higher frequencies are compressed more than lower frequencies. Plus, lower frequencies are stretched more along the x (time) axis because lower frequencies have larger wavelengths thereby causing poorer time resolution. And a given amount of power in the lower frequencies is spread out over a larger area so it looks dimmer. This gives the illusion that the power level in the chirp signal is increasing with frequency.


It's unlikely that AI (Artificial Intelligence) or Machine Learning will help because the search algorithms will have the same bias as the programmer.

4.7 usec. chirp with a frequency sweep of 2.5 MHz to 8MHz
2.2 usec. chirp with a sweep of 6 MHz to 15 MHz
14 usec. chirp with a sweep of 1 MHz to 10 MHz

2020 by Peter C.M. Hahn C.E.T.