{
  "id": 58703,
  "title": "Conic slicing for small angles/non-straight tracks",
  "url": "/competitions/trackml-particle-identification/discussion/58703",
  "author_name": "",
  "post_date": "2018-06-12T17:31:32.842181300Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi,</p>\n\n<p>I noticed that at small angles, selecting hits that are on a cone returns many tracks with only a few hits. This happens because those tracks have a few points at those angles but the following points of the tracks change direction and aren't included in the slice.</p>\n\n<p>As cone slicing seems to be one of the most promising ways to deal with the large amount of tracks pointing in different directions, any solution which uses cone slicing won't be able to detect these tracks.</p>\n\n<p>How could this problem be addressed?</p>\n\n<p>Thanks</p>",
  "messages": [
    {
      "id": "342005",
      "postDate": "06/12/2018 17:31:32",
      "content": "<p>Hi,</p>\n\n<p>I noticed that at small angles, selecting hits that are on a cone returns many tracks with only a few hits. This happens because those tracks have a few points at those angles but the following points of the tracks change direction and aren't included in the slice.</p>\n\n<p>As cone slicing seems to be one of the most promising ways to deal with the large amount of tracks pointing in different directions, any solution which uses cone slicing won't be able to detect these tracks.</p>\n\n<p>How could this problem be addressed?</p>\n\n<p>Thanks</p>",
      "rawMarkdown": "Hi,\n\nI noticed that at small angles, selecting hits that are on a cone returns many tracks with only a few hits. This happens because those tracks have a few points at those angles but the following points of the tracks change direction and aren't included in the slice.\n\nAs cone slicing seems to be one of the most promising ways to deal with the large amount of tracks pointing in different directions, any solution which uses cone slicing won't be able to detect these tracks.\n\nHow could this problem be addressed?\n\nThanks",
      "votes": null
    },
    {
      "id": "342047",
      "postDate": "06/12/2018 18:51:59",
      "content": "<p>Don't use cone slicing ;)</p>",
      "rawMarkdown": "Don't use cone slicing ;)",
      "votes": null
    },
    {
      "id": "342572",
      "postDate": "06/13/2018 18:18:47",
      "content": "<p>Assuming that when you say small angles, you're meaning a narrow cone that is tighter around z, you can change your slicing code to include a wider range for smaller angles.</p>\n\n<p>For example, an iteration on an angle of 45 might include angles 44-46, but an angle of 5 might inlude 0-10?</p>\n\n<p>I didnt think the tracks move too far from their \"cone\", but it makes me think that it might be a good next step to check how much the angle (wrt z axis) can vary, and create a more accurate criteria for a 'cone'. </p>",
      "rawMarkdown": "Assuming that when you say small angles, you're meaning a narrow cone that is tighter around z, you can change your slicing code to include a wider range for smaller angles.\n\nFor example, an iteration on an angle of 45 might include angles 44-46, but an angle of 5 might inlude 0-10?\n\nI didnt think the tracks move too far from their \"cone\", but it makes me think that it might be a good next step to check how much the angle (wrt z axis) can vary, and create a more accurate criteria for a 'cone'.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 342047,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "06/12/2018 18:51:59",
      "content": "<p>Don't use cone slicing ;)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 342572,
      "author_name": "macfarll",
      "author_url": "",
      "post_date": "06/13/2018 18:18:47",
      "content": "<p>Assuming that when you say small angles, you're meaning a narrow cone that is tighter around z, you can change your slicing code to include a wider range for smaller angles.</p>\n\n<p>For example, an iteration on an angle of 45 might include angles 44-46, but an angle of 5 might inlude 0-10?</p>\n\n<p>I didnt think the tracks move too far from their \"cone\", but it makes me think that it might be a good next step to check how much the angle (wrt z axis) can vary, and create a more accurate criteria for a 'cone'. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "342005": "Hi,\n\nI noticed that at small angles, selecting hits that are on a cone returns many tracks with only a few hits. This happens because those tracks have a few points at those angles but the following points of the tracks change direction and aren't included in the slice.\n\nAs cone slicing seems to be one of the most promising ways to deal with the large amount of tracks pointing in different directions, any solution which uses cone slicing won't be able to detect these tracks.\n\nHow could this problem be addressed?\n\nThanks",
    "342047": "Don't use cone slicing ;)",
    "342572": "Assuming that when you say small angles, you're meaning a narrow cone that is tighter around z, you can change your slicing code to include a wider range for smaller angles.\n\nFor example, an iteration on an angle of 45 might include angles 44-46, but an angle of 5 might inlude 0-10?\n\nI didnt think the tracks move too far from their \"cone\", but it makes me think that it might be a good next step to check how much the angle (wrt z axis) can vary, and create a more accurate criteria for a 'cone'."
  },
  "source": "meta"
}