Logo

Open Data Workshop #3 - Day 1 - Shared screen with speaker view - Recording 1/2
Ezequiel
29:12
Oh, my…I was pretty much afraid that I missed the timezone or so.
Ezequiel
29:16
>_<
Ezequiel
29:23
Hello everyone!
Alan W
01:20:26
clap clap!
Ezequiel
01:21:22
I’ve a side question, since I’m not a physicist. Somebody knows what unit of measure is “Mpc”?
Ezequiel
01:21:33
(I saw it from the observation runs)
Jonah Kanner
01:21:42
“Mega parsec”
Alan W
01:21:44
IM will discuss the polarization deconvolution in my talk tomorrow
Leonard Barraclough
01:21:45
^^
Alan W
01:22:03
1 parsec is 3.26 lightyears
Ezequiel
01:22:14
Awesome! thank you!
Ezequiel
01:26:34
My question will be: how does we know from “where” the gw comes from? do we have a way to know this? (it’s a general question, no details needed. An introduction will do… ^_^)
Jonah Kanner
01:27:00
@Ezequiel: yes. Sylvia will discuss this tomorrow.
Ezequiel
01:27:10
Oh, then I will lower my hand.
Simonetta Frittelli
01:27:13
Doesn't the length of the arms of the interferometer limit the observable frequency from above? So the longer the arms, the lower the upper frequency observable? Can someone verify or correct?
Alan W
01:29:16
@Simonetta yes! That’s why our noise curve rises at higher frequewncies
Santanu Ganguly
01:29:36
Hi, just a quick one: is there any plans for multi-messenger type of detections for electromagnetic fields?
Sidharth S Nair
01:29:44
KAGRA is experimenting on cryogenic mirror tech afaik
Alan W
01:30:59
@Santanu - We observed GW170817 with telescopes at all EM wavelengths. Dozens of papers were written. We look forward to doing this again!
Alan W
01:31:18
KAGRA is cryogenic, yes.
Santanu Ganguly
01:31:45
Aweome! @Alan W. I shall look that up!
Ezequiel
01:31:49
Question: why we need “observational runs”, can’t we have the interferometers at all times running? is there a huge cost in its operation or what are the factors?
Alan W
01:32:37
@Santanu https://iopscience.iop.org/article/10.3847/2041-8213/aa91c9
Alan W
01:33:13
There is a good deal of work in progress to attempt to detect the memory effect
Abhishek MS
01:33:40
why dose light is not effected by gw
Ezequiel
01:34:26
@Alan
Sylvia Biscoveanu
01:35:09
Re: plotting noise curves, pygwinc in a nice tool to turn models of the interferometer into noise curves https://git.ligo.org/gwinc/pygwinc
Alan W
01:35:21
Here’s a popular article on memory , with links to papers - https://www.space.com/gravitational-waves-memory-space-time.html
Neel Vadodaria
01:35:52
yes
2016
01:35:52
yes
Uddeepta Deka
01:35:54
yes
Leonard Barraclough
01:35:54
ye
Marissa Walker
01:37:44
@Abhishek light is affected by gravitational waves, there’s a good explanation of why we can still use laser light to detect gravitational waves here: https://www.ligo.caltech.edu/page/faq
Alan W
01:40:12
@Ezequiel (1) we need to periodically update our detectors to improve sensitivity. If we do this right, we win. (2) We bring the detectors down one day a week for necessary maintenance. (3) The detectors are very complex, and a bit too sensitive to environmental disturbances, like earthquakes and storms, “knocking the detectors “out of lock” (observational mode). Unfortunately, this can happen numerous times per week. Duty cycle in O2: https://www.gw-openscience.org/detector_status/O2/
LUIS CARDENAS
01:49:41
Dear Professor Kanner
Santanu Ganguly
01:51:35
In terms of Machine Learning, do you use AstroML package at all? Just wondering where NOT to get lost for my own benefit :-)
Abhishek MS
01:51:37
may I know how Einsten @ Home works
Alan W
01:52:33
@Santanu there are dozens of people in the LIGO Scientific Collaboration attempting to use ML to improve our analyses. They use every ML toolkit on the market!
Santanu Ganguly
01:53:16
@Alan Thanks a lot Sir!
Sylvia Biscoveanu
01:53:26
Here’s a nice review article of all the different applications of ML to gravitational-wave data analysis https://arxiv.org/abs/2005.03745
Simonetta Frittelli
01:53:37
Is the gravity spy data set tool the same as the one available at Zooniverse?
Alan W
01:53:46
@Abhishek E@H uses BOINC to distribute chunks of LIGO data to users (hundreds of thousands of them) to look for continuous GWs.
LUIS CARDENAS
01:54:06
Why the SNR data of events are changing so frecuently? LCN gratings to Professor Kanner
Alan W
01:54:09
https://einsteinathome.org/
Alan W
01:54:26
@Simonetta yes
Santanu Ganguly
01:54:37
Thanks a lot! @Sylvia
Alan W
01:55:37
@LUIS each event is from a source at a different distance and with different masses. More distant, lower mass sources have low SNR (and are often undetectable), and vice versa. We expect a broad spread of event SNRs
petra
01:55:37
what python package do you use for getting the posterior distributions? pymc3?
Ezequiel
01:55:45
Question
Ezequiel
01:56:17
Is there a way to map the gravity spy dataset to the openly strain data?
Alan W
01:56:30
@petra Bilby uses dynesty. LALInference uses it’s own MCMC code. Sylvia will talk about it
Ezequiel
01:56:40
(I saw that the GPS time is hashed, but, there is a *way*?)
Plamen Fiziev
01:57:07
What kind of information about instrumental injections in the data for every event we can find in gwosc ? Only frequencies of the injected signal or more detailed information?
Ezequiel
01:57:10
(*openly available strain data)
Sylvia Biscoveanu
01:57:40
@petra Bilby also interfaces with many other samplers including pymc3
Sylvia Biscoveanu
01:58:46
I will cover different PSD estimation methods tomorrow
Alan W
01:58:52
@plamen Hi! Info on hardware injections is here: https://www.gw-openscience.org/o2_inj/
Plamen Fiziev
01:59:39
Hi Alan! Thank you!
Alan W
02:01:08
@Ezequiel gravity spy gives you a GPS time, and the open data in gw-openscience.org are listed / indexed by GPS time as well; so it should be “easy” to find the former in the latter.
LUIS CARDENAS
02:03:58
Dear Alan, I am refering for the same event, exemple de g w150914 has chenched from 24 to only 19, why?
shobhit ranjan
02:04:48
When a new detection is made, LSC announces it almost immediately for EM follow up. Is the data for the said detection put up on gwosc or elsewhere immediately as well?
Abhishek MS
02:05:53
if i was try to model an unmodeled signal what can i do