{
  "id": 63027,
  "title": "Extra ghost cell event000001000/19932?",
  "url": "/competitions/trackml-particle-identification/discussion/63027",
  "author_name": "",
  "post_date": "2018-08-10T13:57:44.544593200Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello,</p>\n\n<p>I cannot explain the momentum for certain cells, e.g. event000001000/19932</p>\n\n<p>I am attaching my calculation how the given truth momentum is travelling through the cells.</p>\n\n<ol>\n<li><p>from the pixel-detector-cells kernel using 19932 hit there is a faint cell that's barely visible but that is probably OK, the orange cell on the side corner is the suspicious one\n<img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10027/19932.png\" alt=\"from the pixel-detector-cells kernel using 19932 hit\"></p>\n\n<p>2) top view looks OK <img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10028/19932_top.png\" alt=\"front view\"></p>\n\n<p>3) front view misses one cell <img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10029/19932_side.png\" alt=\"front view\"></p></li>\n</ol>\n\n<p>If my calculation is correct then it should not touch cell_19932_35_1090. Is someone able to confirm if there is another bug in the data are the cell/momentum calculations OK? I wasted 3 days on this already :(</p>\n\n<p>Given this confirmed bug <strong>Sensor thickness twice too big (but radiation length is OK)</strong></p>\n\n<ol>\n<li>Should we assume that the detector height is double, correct, or half of the thickness?</li>\n<li>Is the momentum measured at the tx, ty, tz coordinates or at one side of the detector (one thread said that the sensitive surface is at one side)?</li>\n</ol>\n\n<p>Thank you</p>",
  "messages": [
    {
      "id": "368696",
      "postDate": "08/10/2018 13:57:44",
      "content": "<p>Hello,</p>\n\n<p>I cannot explain the momentum for certain cells, e.g. event000001000/19932</p>\n\n<p>I am attaching my calculation how the given truth momentum is travelling through the cells.</p>\n\n<ol>\n<li><p>from the pixel-detector-cells kernel using 19932 hit there is a faint cell that's barely visible but that is probably OK, the orange cell on the side corner is the suspicious one\n<img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10027/19932.png\" alt=\"from the pixel-detector-cells kernel using 19932 hit\"></p>\n\n<p>2) top view looks OK <img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10028/19932_top.png\" alt=\"front view\"></p>\n\n<p>3) front view misses one cell <img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10029/19932_side.png\" alt=\"front view\"></p></li>\n</ol>\n\n<p>If my calculation is correct then it should not touch cell_19932_35_1090. Is someone able to confirm if there is another bug in the data are the cell/momentum calculations OK? I wasted 3 days on this already :(</p>\n\n<p>Given this confirmed bug <strong>Sensor thickness twice too big (but radiation length is OK)</strong></p>\n\n<ol>\n<li>Should we assume that the detector height is double, correct, or half of the thickness?</li>\n<li>Is the momentum measured at the tx, ty, tz coordinates or at one side of the detector (one thread said that the sensitive surface is at one side)?</li>\n</ol>\n\n<p>Thank you</p>",
      "rawMarkdown": "Hello,\n\nI cannot explain the momentum for certain cells, e.g. event000001000/19932\n\nI am attaching my calculation how the given truth momentum is travelling through the cells.\n\n 1. from the pixel-detector-cells kernel using 19932 hit there is a faint cell that's barely visible but that is probably OK, the orange cell on the side corner is the suspicious one\n![from the pixel-detector-cells kernel using 19932 hit][1]\n\n    \n    2) top view looks OK ![front view][2]\n    \n    3) front view misses one cell ![front view][3]\n\nIf my calculation is correct then it should not touch cell_19932_35_1090. Is someone able to confirm if there is another bug in the data are the cell/momentum calculations OK? I wasted 3 days on this already :(\n\nGiven this confirmed bug **Sensor thickness twice too big (but radiation length is OK)**\n\n 1. Should we assume that the detector height is double, correct, or half of the thickness?\n 2. Is the momentum measured at the tx, ty, tz coordinates or at one side of the detector (one thread said that the sensitive surface is at one side)?\n\nThank you\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10027/19932.png\n  [2]: https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10028/19932_top.png\n  [3]: https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10029/19932_side.png",
      "votes": null
    },
    {
      "id": "369374",
      "postDate": "08/12/2018 20:15:58",
      "content": "<p>no waste of time at all . actually in silicon detectors you don't need the particle to physically pass through the cell. Check the image <img src=\"http://www.quantumdiaries.org/wp-content/uploads/2008/07/siliconsteps3.png\" alt=\"particle passing through matter\"></p>",
      "rawMarkdown": "no waste of time at all . actually in silicon detectors you don't need the particle to physically pass through the cell. Check the image ![particle passing through matter][1]\n\n\n  [1]: http://www.quantumdiaries.org/wp-content/uploads/2008/07/siliconsteps3.png",
      "votes": null
    },
    {
      "id": "369407",
      "postDate": "08/13/2018 01:21:36",
      "content": "<p>Thank you agerom for your reply. The line is above the detector module, so vertically I do not see how this could happen. Based on the pictures you posted I assume the charge can leak through from the neighbouring cell horizontally? However the closest neighbouring cell has even smaller charge. I would assume that any leakage would be some portion of the cell value, in this case it would be more than 100%.</p>\n\n<p>Either my calculation is wrong or there is another issue with the simulation. In the latter case I would assume that the organisers would want to track it down.</p>",
      "rawMarkdown": "Thank you agerom for your reply. The line is above the detector module, so vertically I do not see how this could happen. Based on the pictures you posted I assume the charge can leak through from the neighbouring cell horizontally? However the closest neighbouring cell has even smaller charge. I would assume that any leakage would be some portion of the cell value, in this case it would be more than 100%.\n\nEither my calculation is wrong or there is another issue with the simulation. In the latter case I would assume that the organisers would want to track it down.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 369374,
      "author_name": "artemiosgeromitsos",
      "author_url": "",
      "post_date": "08/12/2018 20:15:58",
      "content": "<p>no waste of time at all . actually in silicon detectors you don't need the particle to physically pass through the cell. Check the image <img src=\"http://www.quantumdiaries.org/wp-content/uploads/2008/07/siliconsteps3.png\" alt=\"particle passing through matter\"></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 369407,
      "author_name": "dthrone",
      "author_url": "",
      "post_date": "08/13/2018 01:21:36",
      "content": "<p>Thank you agerom for your reply. The line is above the detector module, so vertically I do not see how this could happen. Based on the pictures you posted I assume the charge can leak through from the neighbouring cell horizontally? However the closest neighbouring cell has even smaller charge. I would assume that any leakage would be some portion of the cell value, in this case it would be more than 100%.</p>\n\n<p>Either my calculation is wrong or there is another issue with the simulation. In the latter case I would assume that the organisers would want to track it down.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "368696": "Hello,\n\nI cannot explain the momentum for certain cells, e.g. event000001000/19932\n\nI am attaching my calculation how the given truth momentum is travelling through the cells.\n\n 1. from the pixel-detector-cells kernel using 19932 hit there is a faint cell that's barely visible but that is probably OK, the orange cell on the side corner is the suspicious one\n![from the pixel-detector-cells kernel using 19932 hit][1]\n\n    \n    2) top view looks OK ![front view][2]\n    \n    3) front view misses one cell ![front view][3]\n\nIf my calculation is correct then it should not touch cell_19932_35_1090. Is someone able to confirm if there is another bug in the data are the cell/momentum calculations OK? I wasted 3 days on this already :(\n\nGiven this confirmed bug **Sensor thickness twice too big (but radiation length is OK)**\n\n 1. Should we assume that the detector height is double, correct, or half of the thickness?\n 2. Is the momentum measured at the tx, ty, tz coordinates or at one side of the detector (one thread said that the sensitive surface is at one side)?\n\nThank you\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10027/19932.png\n  [2]: https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10028/19932_top.png\n  [3]: https://storage.googleapis.com/kaggle-forum-message-attachments/368696/10029/19932_side.png",
    "369374": "no waste of time at all . actually in silicon detectors you don't need the particle to physically pass through the cell. Check the image ![particle passing through matter][1]\n\n\n  [1]: http://www.quantumdiaries.org/wp-content/uploads/2008/07/siliconsteps3.png",
    "369407": "Thank you agerom for your reply. The line is above the detector module, so vertically I do not see how this could happen. Based on the pictures you posted I assume the charge can leak through from the neighbouring cell horizontally? However the closest neighbouring cell has even smaller charge. I would assume that any leakage would be some portion of the cell value, in this case it would be more than 100%.\n\nEither my calculation is wrong or there is another issue with the simulation. In the latter case I would assume that the organisers would want to track it down."
  },
  "source": "meta"
}