![\"000a5b6e5c\"](\"https://i.imgur.com/hqvf0PE.png\")
First let us check a strong positive (GW pattern visible to human-like objects), many notebooks use this sample to display the \"chirp\" in the time-frequency diagram (CQT, CWT, spectrogram, etc)
\n
There is a strong spike around 612hz:
\n![\"000a5b6e5c_energy\"](\"https://i.imgur.com/vqVe58x.png\")
We might think, there is the energy signature of the GW!
\nHowever, here is a false positive (if we use the regular 0.5 threshold)
\n![\"c92bc5ce6b\"](\"https://i.imgur.com/tYobZoO.png\")
![\"c92bc5ce6b_spectrum\"](\"https://i.imgur.com/64RsaHn.png\")
Huh?
\nNote this will embody itself as a white band (visible to human-like objects) in 2D images (CQT, CWT, spectrogram, etc), but this only exists in LIGOs (in the US), not the Virgo (in Europe).
\nFor the models, I filtered out that frequency and the CV improved about 0.001-0.002. So it may be useful to top teams.
", "messages": [ { "id": "1524522", "postDate": "09/26/2021 16:06:42", "content": "First let us check a strong positive (GW pattern visible to human-like objects), many notebooks use this sample to display the \"chirp\" in the time-frequency diagram (CQT, CWT, spectrogram, etc)
\n
There is a strong spike around 612hz:
\n
We might think, there is the energy signature of the GW!
\nHowever, here is a false positive (if we use the regular 0.5 threshold)
\n
Huh?
\nNote this will embody itself as a white band (visible to human-like objects) in 2D images (CQT, CWT, spectrogram, etc), but this only exists in LIGOs (in the US), not the Virgo (in Europe).
\nFor the models, I filtered out that frequency and the CV improved about 0.001-0.002. So it may be useful to top teams.
", "rawMarkdown": "First let us check a strong positive (GW pattern visible to human-like objects), many notebooks use this sample to display the \"chirp\" in the time-frequency diagram (CQT, CWT, spectrogram, etc)\n\n\n\nThere is a strong spike around 612hz:\n\n\n\nWe might think, there is the energy signature of the GW!\n\nHowever, here is a false positive (if we use the regular 0.5 threshold)\n\n\n\n\n\nHuh?\n\nNote this will embody itself as a white band (visible to human-like objects) in 2D images (CQT, CWT, spectrogram, etc), but this only exists in LIGOs (in the US), not the Virgo (in Europe).\n\nFor the models, I filtered out that frequency and the CV improved about 0.001-0.002. So it may be useful to top teams.", "votes": null }, { "id": "1524582", "postDate": "09/26/2021 17:14:55", "content": "![\"\"](\"https://i.ibb.co/1bBTvZ1/ASDz.png\")
This is what the mean ASDs looks like across the entire ve- portion of the training dataset (~280k samples). The spike you mentioned is visible there as well, and I believe there was another topic that discussed it as well a while back. There was also a nice phd dissertation Heng posted that shares about 10 causes for the spikes at various frequencies. Notably, the generated data we have is missing a lot of the spikes present in actual detector data (they did not model in some known noise sources).
", "rawMarkdown": "\n\nThis is what the mean ASDs looks like across the entire ve- portion of the training dataset (~280k samples). The spike you mentioned is visible there as well, and I believe there was another topic that discussed it as well a while back. There was also a nice phd dissertation Heng posted that shares about 10 causes for the spikes at various frequencies. Notably, the generated data we have is missing a lot of the spikes present in actual detector data (they did not model in some known noise sources).", "votes": null }, { "id": "1524644", "postDate": "09/26/2021 18:13:58", "content": "i thought i read in some post that it is safe to assume all GW waves in the competition are below 500 Hz?
", "rawMarkdown": "i thought i read in some post that it is safe to assume all GW waves in the competition are below 500 Hz?", "votes": null }, { "id": "1524724", "postDate": "09/26/2021 19:16:09", "content": "The hosts never confirmed that. In fact, they basically said in the welcome thread something to the effect of 'we take all liberties to produce any astronomically possible signal.' The <300 and <500 hz values are mostly from literature and experiments in terms of what people cut off at.
", "rawMarkdown": "The hosts never confirmed that. In fact, they basically said in the welcome thread something to the effect of 'we take all liberties to produce any astronomically possible signal.' The <300 and <500 hz values are mostly from literature and experiments in terms of what people cut off at.", "votes": null }, { "id": "1524735", "postDate": "09/26/2021 19:38:39", "content": "i make high-resolution cqt and then inspect by eye.
\ni never find anything above 600hz.
(but it is possible that I miss the chirp since I am not trained in GW)
", "rawMarkdown": "i make high-resolution cqt and then inspect by eye.\ni never find anything above 600hz.\n\n(but it is possible that I miss the chirp since I am not trained in GW)", "votes": null }, { "id": "1524745", "postDate": "09/26/2021 19:52:47", "content": "![\"\"](\"https://i.ibb.co/ynKztDn/fakwbh.png\")
This is a fake (but totally possible I suppose) BH merger in FD:
\nOh my god…this is such a huge hint…too late for me to train new models 😂
", "rawMarkdown": "Oh my god...this is such a huge hint...too late for me to train new models 😂", "votes": null }, { "id": "1525188", "postDate": "09/27/2021 08:01:25", "content": "@scaomath are above parameters of Preprocessing or some thing ?
", "rawMarkdown": "scaomath are above parameters of Preprocessing or some thing ?", "votes": null }, { "id": "1525210", "postDate": "09/27/2021 08:35:54", "content": "@authman posted what a simulated black hole merger would produce, without noise.
", "rawMarkdown": "authman posted what a simulated black hole merger would produce, without noise.", "votes": null }, { "id": "1525375", "postDate": "09/27/2021 11:11:21", "content": "@scaomath What hint do you mean? 😄
", "rawMarkdown": "scaomath What hint do you mean? 😄", "votes": null }, { "id": "1525390", "postDate": "09/27/2021 11:25:08", "content": "While shape is correct, your frequencies are off by the factor of 2, probably wrong sample rate. Anyway, here is thread about this artifact - https://www.kaggle.com/c/g2net-gravitational-wave-detection/discussion/263995
\nTL/DR: Noise from dampening systems, not related to signal.
probably he meant about this: \"all GW waves in the competition are below 500 Hz\"
", "rawMarkdown": "probably he meant about this: \"all GW waves in the competition are below 500 Hz\"", "votes": null }, { "id": "1525645", "postDate": "09/27/2021 14:48:04", "content": "@imeintanis I meant the tiny phase difference shown in the simulation, useful a priori information for augmentations.
", "rawMarkdown": "imeintanis I meant the tiny phase difference shown in the simulation, useful a priori information for augmentations.", "votes": null }, { "id": "1528596", "postDate": "09/29/2021 19:19:56", "content": "\n\ni make high-resolution cqt and then inspect by eye.
\n
\n i never find anything above 600hz.
Cutoff at 640hz.
\n![\"\"](\"https://i.imgur.com/Oh4ZzAg.png\")
yes, this is also roughly what i get.
", "rawMarkdown": "yes, this is also roughly what i get.", "votes": null }, { "id": "1559746", "postDate": "10/27/2021 07:11:22", "content": "Hey All,
\nThank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey https://forms.gle/QP9L16niPexozyhu5.
\nThank you all,
\nRegards,
\nChris
This is what the mean ASDs looks like across the entire ve- portion of the training dataset (~280k samples). The spike you mentioned is visible there as well, and I believe there was another topic that discussed it as well a while back. There was also a nice phd dissertation Heng posted that shares about 10 causes for the spikes at various frequencies. Notably, the generated data we have is missing a lot of the spikes present in actual detector data (they did not model in some known noise sources).
", "votes": null, "replies": [] }, { "id": 1524644, "author_name": "hengck23", "author_url": "", "post_date": "09/26/2021 18:13:58", "content": "i thought i read in some post that it is safe to assume all GW waves in the competition are below 500 Hz?
", "votes": null, "replies": [ { "id": 1524724, "author_name": "authman", "author_url": "", "post_date": "09/26/2021 19:16:09", "content": "The hosts never confirmed that. In fact, they basically said in the welcome thread something to the effect of 'we take all liberties to produce any astronomically possible signal.' The <300 and <500 hz values are mostly from literature and experiments in terms of what people cut off at.
", "votes": null, "replies": [] }, { "id": 1524735, "author_name": "hengck23", "author_url": "", "post_date": "09/26/2021 19:38:39", "content": "i make high-resolution cqt and then inspect by eye.
\ni never find anything above 600hz.
(but it is possible that I miss the chirp since I am not trained in GW)
", "votes": null, "replies": [] }, { "id": 1524745, "author_name": "authman", "author_url": "", "post_date": "09/26/2021 19:52:47", "content": "
This is a fake (but totally possible I suppose) BH merger in FD:
\nOh my god…this is such a huge hint…too late for me to train new models 😂
", "votes": null, "replies": [] }, { "id": 1525188, "author_name": "jaideepvalani", "author_url": "", "post_date": "09/27/2021 08:01:25", "content": "@scaomath are above parameters of Preprocessing or some thing ?
", "votes": null, "replies": [] }, { "id": 1525210, "author_name": "scaomath", "author_url": "", "post_date": "09/27/2021 08:35:54", "content": "@authman posted what a simulated black hole merger would produce, without noise.
", "votes": null, "replies": [] }, { "id": 1525375, "author_name": "sggpls", "author_url": "", "post_date": "09/27/2021 11:11:21", "content": "@scaomath What hint do you mean? 😄
", "votes": null, "replies": [ { "id": 1525393, "author_name": "imeintanis", "author_url": "", "post_date": "09/27/2021 11:26:52", "content": "probably he meant about this: \"all GW waves in the competition are below 500 Hz\"
", "votes": null, "replies": [] } ] }, { "id": 1525645, "author_name": "scaomath", "author_url": "", "post_date": "09/27/2021 14:48:04", "content": "@imeintanis I meant the tiny phase difference shown in the simulation, useful a priori information for augmentations.
", "votes": null, "replies": [] }, { "id": 1528596, "author_name": "fffrrt", "author_url": "", "post_date": "09/29/2021 19:19:56", "content": "\n\ni make high-resolution cqt and then inspect by eye.
\n
\n i never find anything above 600hz.
Cutoff at 640hz.
\n
yes, this is also roughly what i get.
", "votes": null, "replies": [] } ] }, { "id": 1525390, "author_name": "denisbsu", "author_url": "", "post_date": "09/27/2021 11:25:08", "content": "While shape is correct, your frequencies are off by the factor of 2, probably wrong sample rate. Anyway, here is thread about this artifact - https://www.kaggle.com/c/g2net-gravitational-wave-detection/discussion/263995
\nTL/DR: Noise from dampening systems, not related to signal.
Hey All,
\nThank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey https://forms.gle/QP9L16niPexozyhu5.
\nThank you all,
\nRegards,
\nChris