Lecture Questions - Day 2 (2023)

Questions for the Day 2 lectures (May 16, 2023) can be posted here

What is an ‘optimal Q Value’?

Is there a register of glitches in the public data? Can I find glitches in the same way I could find events in yesterday tutorials?

sir, what are the reason behind of these glitches?

how trains and fridge connected to main power creates glitches?

Could you tell us a bit more about RC tracking?

Glitch, specially blip glitch can trigger the template bank. However, time-frequency morphology of a glitch is different from the chirp signal. Gravity spy is one tool which can distinguish between cbc signal and glitch.

For the GWpy code, it is the Q giving you the tile with the highest energy

If high Q value is equivalent to neutron stars’ collision, what do low Q value and mid Q value signify? What exactly does Omicron detect to find the glitches?

Low Q (usually about 2pi) is more suited for “short” signals such as binary black holes; for the glitches, Omicron detects excesses of energy which are found to be not coherent among different detectors neither to match waveform templates, so these excesses are likely to be given by noise transients

1 Like

Hi @Shreyan!
The practical way of usage is to choose a Q “range” equal to the number of wave periods that you are willing to see. If you choose a low Q value, you can see better binary black holes systems.

1 Like

Typically a rule of thumb to decide the value of the Q-transform is to know how many cycles you expect to see in your data. Binary black holes merge around 100 Hz and they last about 0.1 s in our data. This will give you ~ 10 cycles observed. So Q-range = (5,20) is a good range for binary black holes. Higher values are better for binary neutron stars (see GW170817)


The short answer is no, because glitches are really many (much more than events) and people are not interested in analyzing them in the same systematic way they analyze events. Anyway, there are banks of glitches that are used for recognizing glitches in the data (e.g. with machine learning methods) Zooniverse

To learn more about Glitches you can check Gravity Spy

If you would like to know more about how to classify and remove glitches for scientific applications, you can check this pagee](O3 Auxiliary Channel Data Release)
and references therein.

There are a lot of causes. Typically something can interact with a component of our detectors. For trains and human activity, the motivation is that tiny ground motions are propagated to the apparatus of the interferometer. This motion propagates to our mirrors and can mimic the passage of a gravitational signal.

How many GW should be detected in order to calculate or detect GW memory effect?


That’s a very difficult question! I have found a paper here studying the possibility of detecting GW memory effect and it seems that it will take thousands of events. Maybe looking at LISA and lower frequencies will be better!

1 Like

Since the noise in GW data is non Gaussian. what are the alternatives to Gaussian assumptions to model the noise.

That’s difficult to state because glitches morphology is really huge and difficult to model (glitches are what give you non-Gaussian noise). The only thing you know is that glitches are not coherent among different detectors (i.e. generally speaking not at the same time unless spurious coincidences, but consider that coherence is a bit more complicated and not only a matter of “coincidence” in time); a pipeline trying to “model” glitches considering these are not coherent among different detectors is BayesWave Phys. Rev. D 103, 044006 (2021) - BayesWave analysis pipeline in the era of gravitational wave observations

1 Like

Hi @abhijeet!
You can see this paper for your answer. Basically we can construct a characterization of sensitivity of interefometers evaluating the noise in the frequency domain, as you will see in the second day’s tutorials.


What is the significance of FARs, considering we will check whether it is a detection anyway? How does calculating the probability help?