![\"event_ex\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1636313%2F43b591393d9be45b4efb1e9ce6347f5a%2FScreenshot%202023-01-22%20at%2012.42.07%20PM.png?generation=1674409516323981&alt=media\")
As an example, if you look at row #109966314 in the train meta parquet file, it points you to an event that plots like so:
\n
You can see above that in the auxillary=False plot, the pulses are found in a perfectly vertical (z values are all the same) line, and on top of that the size of the pulses looks the same too.
Is something going on here? Or am I overthinking?
", "messages": [ { "id": "2111224", "postDate": "01/22/2023 17:46:53", "content": "As an example, if you look at row #109966314 in the train meta parquet file, it points you to an event that plots like so:
\n
You can see above that in the auxillary=False plot, the pulses are found in a perfectly vertical (z values are all the same) line, and on top of that the size of the pulses looks the same too.
Is something going on here? Or am I overthinking?
", "rawMarkdown": "As an example, if you look at row #109966314 in the train meta parquet file, it points you to an event that plots like so:\n\n\n\nYou can see above that in the **`auxillary=False`** plot, the pulses are found in a perfectly vertical (z values are all the same) line, and on top of that the size of the pulses looks the same too.\n\nIs something going on here? Or am I overthinking?", "votes": null }, { "id": "2111237", "postDate": "01/22/2023 17:59:20", "content": "I think it would be physically possible. If a neutrino collision in the ice created a muon that happened to travel vertically, it could make a signature like that. The fact that the colour/time in your left plot would agree with this makes me think it's real. I might be completely wrong though
", "rawMarkdown": "I think it would be physically possible. If a neutrino collision in the ice created a muon that happened to travel vertically, it could make a signature like that. The fact that the colour/time in your left plot would agree with this makes me think it's real. I might be completely wrong though", "votes": null }, { "id": "2111319", "postDate": "01/22/2023 19:27:46", "content": "What if auxillary=True used?
", "rawMarkdown": "What if auxillary=True used?", "votes": null }, { "id": "2111441", "postDate": "01/22/2023 22:31:04", "content": "Neither pure noise nor are you overthinking it.
\nIt is mainly an artifact of cost-effective placement of the sensors. For increasing x and y density, they have to drill another hole in the ice(!!!) so very expensive. For z density, they just need to add a sensor to the already existing giant string of sensors. So small z movement is MUCH easier for the sensors to detect than small x or small y movement. Since it's 125 meters for xy distance vs 17 meters for z distance, triggering 3-4 vertically stacked and that's still far less distance apart than a neighboring xy string.
\nBased on the official website's science tab, and my favorite two paragraphs are below:
\n\"The in-ice component of IceCube consists of 5,160 digital optical modules (DOMs), each with a ten-inch photomultiplier tube and associated electronics. The DOMs are attached to vertical “strings,” frozen into 86 boreholes, and arrayed over a cubic kilometer from 1,450 meters to 2,450 meters depth. The strings are deployed on a hexagonal grid with 125 meters spacing and hold 60 DOMs each. The vertical separation of the DOMs is 17 meters.
\nEight of these strings at the center of the array were deployed more compactly, with a horizontal separation of about 70 meters and a vertical DOM spacing of 7 meters. This denser configuration forms the DeepCore subdetector, which lowers the neutrino energy threshold to about 10 GeV, creating the opportunity to study neutrino oscillations.\"
\nThese two paragraphs combined with any favorite EDA notebook like @shlomoron 's excellent notebook here, and combined it gives a really good understanding of the physical layout.
", "rawMarkdown": "Neither pure noise nor are you overthinking it.\n\nIt is mainly an artifact of cost-effective placement of the sensors. For increasing x and y density, they have to drill another hole in the ice(!!!) so very expensive. For z density, they just need to add a sensor to the already existing giant string of sensors. So small z movement is MUCH easier for the sensors to detect than small x or small y movement. Since it's 125 meters for xy distance vs 17 meters for z distance, triggering 3-4 vertically stacked and that's still far less distance apart than a neighboring xy string.\n\nBased on [the official website's science tab](https://icecube.wisc.edu/science/icecube/), and my favorite two paragraphs are below:\n\n\"The in-ice component of IceCube consists of 5,160 digital optical modules (DOMs), each with a ten-inch photomultiplier tube and associated electronics. The DOMs are attached to vertical “strings,” frozen into 86 boreholes, and arrayed over a cubic kilometer from 1,450 meters to 2,450 meters depth. The strings are deployed on a hexagonal grid with 125 meters spacing and hold 60 DOMs each. The vertical separation of the DOMs is 17 meters.\n\nEight of these strings at the center of the array were deployed more compactly, with a horizontal separation of about 70 meters and a vertical DOM spacing of 7 meters. This denser configuration forms the DeepCore subdetector, which lowers the neutrino energy threshold to about 10 GeV, creating the opportunity to study neutrino oscillations.\"\n\nThese two paragraphs combined with any favorite EDA notebook like @shlomoron 's [excellent notebook here](https://www.kaggle.com/code/shlomoron/icecube-eda-pca-baseline-cv-1-28-lb-1-274), and combined it gives a really good understanding of the physical layout.", "votes": null }, { "id": "2111445", "postDate": "01/22/2023 22:34:50", "content": "So the long line you posted would be unusual, but it still might be caused by around 0.1% of all possible random angles, instead of 0.0001% if the sensors were more evenly spaced in all 3 directions. (Made up numbers of course)
", "rawMarkdown": "So the long line you posted would be unusual, but it still might be caused by around 0.1% of all possible random angles, instead of 0.0001% if the sensors were more evenly spaced in all 3 directions. (Made up numbers of course)", "votes": null }, { "id": "2111448", "postDate": "01/22/2023 22:36:41", "content": "Oh, and aux=False the \"neighbor\" logic might increase the chance of this artifact, again because of how 'neighbor' interacts with the unusual spacing. Vastly more likely that neighboring z sensors will both trigger at about the same time, compared with neighboring xy sensors triggering at about the same time.
", "rawMarkdown": "Oh, and aux=False the \"neighbor\" logic might increase the chance of this artifact, again because of how 'neighbor' interacts with the unusual spacing. Vastly more likely that neighboring z sensors will both trigger at about the same time, compared with neighboring xy sensors triggering at about the same time.", "votes": null }, { "id": "2153746", "postDate": "02/21/2023 16:05:22", "content": "Based on my looks at the data and predictions my models make - they are both Noise and Signals.
", "rawMarkdown": "Based on my looks at the data and predictions my models make - they are both Noise and Signals.", "votes": null }, { "id": "2158190", "postDate": "02/24/2023 17:22:50", "content": "Old question, but this seems to me like an atmospheric muon along with it triggering auxiliary=False (possibly due to pulses happening close together in time and on the same string).
\nFirst thing, note that the neutrino direction (the line) is different from the track direction. This means that it is a muon caused by something other than the neutrino itself.
\nSecond, it is coming straight down. Atmospheric muons tend to come straight down (I've read somewhere - I think the webinar - that this is one of the possible ways to differentiate them from the muons created during the neutrino interactions).
\nThird (well, not much of an argument), the top of the detectors is opaque to avoid these muons from creating too much noise, however, light still bounces around them in the ice and reaches the transparent side below.
\nSo, that is it, an atmospheric muon interfering with a netrino event.
\nMy personal guess is that this event would be pretty hard to reconstruct, because it looks like the charges detected because of the atmospheric muon are greater than the charges detected because of the neutrino itself, maybe a model that assumes that atmospheric muons are coming straight down would account for that.
", "rawMarkdown": "Old question, but this seems to me like an atmospheric muon along with it triggering auxiliary=False (possibly due to pulses happening close together in time and on the same string).\n\nFirst thing, note that the neutrino direction (the line) is different from the track direction. This means that it is a muon caused by something other than the neutrino itself.\n\nSecond, it is coming straight down. Atmospheric muons tend to come straight down (I've read somewhere - I think the webinar - that this is one of the possible ways to differentiate them from the muons created during the neutrino interactions).\n\nThird (well, not much of an argument), the top of the detectors is opaque to avoid these muons from creating too much noise, however, light still bounces around them in the ice and reaches the transparent side below.\n\nSo, that is it, an atmospheric muon interfering with a netrino event.\n\nMy personal guess is that this event would be pretty hard to reconstruct, because it looks like the charges detected because of the atmospheric muon are greater than the charges detected because of the neutrino itself, maybe a model that assumes that atmospheric muons are coming straight down would account for that.", "votes": null } ], "comments": [ { "id": 2111237, "author_name": "anjum48", "author_url": "", "post_date": "01/22/2023 17:59:20", "content": "I think it would be physically possible. If a neutrino collision in the ice created a muon that happened to travel vertically, it could make a signature like that. The fact that the colour/time in your left plot would agree with this makes me think it's real. I might be completely wrong though
", "votes": null, "replies": [] }, { "id": 2111319, "author_name": "walrickroy", "author_url": "", "post_date": "01/22/2023 19:27:46", "content": "What if auxillary=True used?
", "votes": null, "replies": [] }, { "id": 2111441, "author_name": "roberthatch", "author_url": "", "post_date": "01/22/2023 22:31:04", "content": "Neither pure noise nor are you overthinking it.
\nIt is mainly an artifact of cost-effective placement of the sensors. For increasing x and y density, they have to drill another hole in the ice(!!!) so very expensive. For z density, they just need to add a sensor to the already existing giant string of sensors. So small z movement is MUCH easier for the sensors to detect than small x or small y movement. Since it's 125 meters for xy distance vs 17 meters for z distance, triggering 3-4 vertically stacked and that's still far less distance apart than a neighboring xy string.
\nBased on the official website's science tab, and my favorite two paragraphs are below:
\n\"The in-ice component of IceCube consists of 5,160 digital optical modules (DOMs), each with a ten-inch photomultiplier tube and associated electronics. The DOMs are attached to vertical “strings,” frozen into 86 boreholes, and arrayed over a cubic kilometer from 1,450 meters to 2,450 meters depth. The strings are deployed on a hexagonal grid with 125 meters spacing and hold 60 DOMs each. The vertical separation of the DOMs is 17 meters.
\nEight of these strings at the center of the array were deployed more compactly, with a horizontal separation of about 70 meters and a vertical DOM spacing of 7 meters. This denser configuration forms the DeepCore subdetector, which lowers the neutrino energy threshold to about 10 GeV, creating the opportunity to study neutrino oscillations.\"
\nThese two paragraphs combined with any favorite EDA notebook like @shlomoron 's excellent notebook here, and combined it gives a really good understanding of the physical layout.
", "votes": null, "replies": [ { "id": 2111445, "author_name": "roberthatch", "author_url": "", "post_date": "01/22/2023 22:34:50", "content": "So the long line you posted would be unusual, but it still might be caused by around 0.1% of all possible random angles, instead of 0.0001% if the sensors were more evenly spaced in all 3 directions. (Made up numbers of course)
", "votes": null, "replies": [] }, { "id": 2111448, "author_name": "roberthatch", "author_url": "", "post_date": "01/22/2023 22:36:41", "content": "Oh, and aux=False the \"neighbor\" logic might increase the chance of this artifact, again because of how 'neighbor' interacts with the unusual spacing. Vastly more likely that neighboring z sensors will both trigger at about the same time, compared with neighboring xy sensors triggering at about the same time.
", "votes": null, "replies": [] } ] }, { "id": 2153746, "author_name": "pcjimmmy", "author_url": "", "post_date": "02/21/2023 16:05:22", "content": "Based on my looks at the data and predictions my models make - they are both Noise and Signals.
", "votes": null, "replies": [] }, { "id": 2158190, "author_name": "araraonline", "author_url": "", "post_date": "02/24/2023 17:22:50", "content": "Old question, but this seems to me like an atmospheric muon along with it triggering auxiliary=False (possibly due to pulses happening close together in time and on the same string).
\nFirst thing, note that the neutrino direction (the line) is different from the track direction. This means that it is a muon caused by something other than the neutrino itself.
\nSecond, it is coming straight down. Atmospheric muons tend to come straight down (I've read somewhere - I think the webinar - that this is one of the possible ways to differentiate them from the muons created during the neutrino interactions).
\nThird (well, not much of an argument), the top of the detectors is opaque to avoid these muons from creating too much noise, however, light still bounces around them in the ice and reaches the transparent side below.
\nSo, that is it, an atmospheric muon interfering with a netrino event.
\nMy personal guess is that this event would be pretty hard to reconstruct, because it looks like the charges detected because of the atmospheric muon are greater than the charges detected because of the neutrino itself, maybe a model that assumes that atmospheric muons are coming straight down would account for that.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2111224": "As an example, if you look at row #109966314 in the train meta parquet file, it points you to an event that plots like so:\n\n\n\nYou can see above that in the **`auxillary=False`** plot, the pulses are found in a perfectly vertical (z values are all the same) line, and on top of that the size of the pulses looks the same too.\n\nIs something going on here? Or am I overthinking?", "2111237": "I think it would be physically possible. If a neutrino collision in the ice created a muon that happened to travel vertically, it could make a signature like that. The fact that the colour/time in your left plot would agree with this makes me think it's real. I might be completely wrong though", "2111319": "What if auxillary=True used?", "2111441": "Neither pure noise nor are you overthinking it.\n\nIt is mainly an artifact of cost-effective placement of the sensors. For increasing x and y density, they have to drill another hole in the ice(!!!) so very expensive. For z density, they just need to add a sensor to the already existing giant string of sensors. So small z movement is MUCH easier for the sensors to detect than small x or small y movement. Since it's 125 meters for xy distance vs 17 meters for z distance, triggering 3-4 vertically stacked and that's still far less distance apart than a neighboring xy string.\n\nBased on [the official website's science tab](https://icecube.wisc.edu/science/icecube/), and my favorite two paragraphs are below:\n\n\"The in-ice component of IceCube consists of 5,160 digital optical modules (DOMs), each with a ten-inch photomultiplier tube and associated electronics. The DOMs are attached to vertical “strings,” frozen into 86 boreholes, and arrayed over a cubic kilometer from 1,450 meters to 2,450 meters depth. The strings are deployed on a hexagonal grid with 125 meters spacing and hold 60 DOMs each. The vertical separation of the DOMs is 17 meters.\n\nEight of these strings at the center of the array were deployed more compactly, with a horizontal separation of about 70 meters and a vertical DOM spacing of 7 meters. This denser configuration forms the DeepCore subdetector, which lowers the neutrino energy threshold to about 10 GeV, creating the opportunity to study neutrino oscillations.\"\n\nThese two paragraphs combined with any favorite EDA notebook like @shlomoron 's [excellent notebook here](https://www.kaggle.com/code/shlomoron/icecube-eda-pca-baseline-cv-1-28-lb-1-274), and combined it gives a really good understanding of the physical layout.", "2111445": "So the long line you posted would be unusual, but it still might be caused by around 0.1% of all possible random angles, instead of 0.0001% if the sensors were more evenly spaced in all 3 directions. (Made up numbers of course)", "2111448": "Oh, and aux=False the \"neighbor\" logic might increase the chance of this artifact, again because of how 'neighbor' interacts with the unusual spacing. Vastly more likely that neighboring z sensors will both trigger at about the same time, compared with neighboring xy sensors triggering at about the same time.", "2153746": "Based on my looks at the data and predictions my models make - they are both Noise and Signals.", "2158190": "Old question, but this seems to me like an atmospheric muon along with it triggering auxiliary=False (possibly due to pulses happening close together in time and on the same string).\n\nFirst thing, note that the neutrino direction (the line) is different from the track direction. This means that it is a muon caused by something other than the neutrino itself.\n\nSecond, it is coming straight down. Atmospheric muons tend to come straight down (I've read somewhere - I think the webinar - that this is one of the possible ways to differentiate them from the muons created during the neutrino interactions).\n\nThird (well, not much of an argument), the top of the detectors is opaque to avoid these muons from creating too much noise, however, light still bounces around them in the ice and reaches the transparent side below.\n\nSo, that is it, an atmospheric muon interfering with a netrino event.\n\nMy personal guess is that this event would be pretty hard to reconstruct, because it looks like the charges detected because of the atmospheric muon are greater than the charges detected because of the neutrino itself, maybe a model that assumes that atmospheric muons are coming straight down would account for that." }, "source": "meta" }