{
  "id": 61116,
  "title": "Properties of helices with respect to axis?",
  "url": "/competitions/trackml-particle-identification/discussion/61116",
  "author_name": "Kha Vo",
  "post_date": "2018-07-14T12:08:07.925000",
  "votes": 0,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi Organisers,\nI have one question regarding the magnetic field direction. </p>\n\n<p>As the code of helix unroll is around there in the kernel section, I assume that we can cluster the hits that lie on an arc, in the XY plane. It is because the angle Phi is calculated by arctan(y/x).</p>\n\n<p>As a result, let us consider the case that a true particle forms a clear arc in the XZ plane. Also assuming that this particle generates hits that all have Y=0.  As a result, in the XY plane, this particle becomes a straight line that lies on the X-axis. Therefore, this particle will be better clustered by an algorithm that calculates phi from the XZ plane instead of XY plane.</p>\n\n<p>What I'm concerned is that, is there any assumption on the magnetic field which cause the particles to form helices with common properties wrt any axis? </p>\n\n<p>If the answer of the above question is No, then we can assume that a particle can form any kind of helices in any direction. Hence, we can develop a clustering algorithm that can generalise better, than just using phi calculated from the XY-plane.</p>\n\n<p>Thanks. </p>",
  "messages": [
    {
      "id": 356767,
      "postDate": "2018-07-14T12:26:49.430Z",
      "content": "<p>The magnetic field is to a good approximation parallel to the z axis, hence the trajectories being  arcs of circle on the XY plane. So the XY plane does have a special role.</p>",
      "rawMarkdown": "The magnetic field is to a good approximation parallel to the z axis, hence the trajectories being  arcs of circle on the XY plane. So the XY plane does have a special role.",
      "replies": [
        {
          "id": 356780,
          "postDate": "2018-07-14T12:37:54.920Z",
          "content": "<p>Thanks for your answer.</p>",
          "rawMarkdown": "Thanks for your answer."
        }
      ]
    },
    {
      "id": 356757,
      "postDate": "2018-07-14T12:08:07.927Z",
      "content": "<p>Hi Organisers,\nI have one question regarding the magnetic field direction. </p>\n\n<p>As the code of helix unroll is around there in the kernel section, I assume that we can cluster the hits that lie on an arc, in the XY plane. It is because the angle Phi is calculated by arctan(y/x).</p>\n\n<p>As a result, let us consider the case that a true particle forms a clear arc in the XZ plane. Also assuming that this particle generates hits that all have Y=0.  As a result, in the XY plane, this particle becomes a straight line that lies on the X-axis. Therefore, this particle will be better clustered by an algorithm that calculates phi from the XZ plane instead of XY plane.</p>\n\n<p>What I'm concerned is that, is there any assumption on the magnetic field which cause the particles to form helices with common properties wrt any axis? </p>\n\n<p>If the answer of the above question is No, then we can assume that a particle can form any kind of helices in any direction. Hence, we can develop a clustering algorithm that can generalise better, than just using phi calculated from the XY-plane.</p>\n\n<p>Thanks. </p>",
      "rawMarkdown": "Hi Organisers,\nI have one question regarding the magnetic field direction. \n\nAs the code of helix unroll is around there in the kernel section, I assume that we can cluster the hits that lie on an arc, in the XY plane. It is because the angle Phi is calculated by arctan(y/x).\n\nAs a result, let us consider the case that a true particle forms a clear arc in the XZ plane. Also assuming that this particle generates hits that all have Y=0.  As a result, in the XY plane, this particle becomes a straight line that lies on the X-axis. Therefore, this particle will be better clustered by an algorithm that calculates phi from the XZ plane instead of XY plane.\n\nWhat I'm concerned is that, is there any assumption on the magnetic field which cause the particles to form helices with common properties wrt any axis? \n\nIf the answer of the above question is No, then we can assume that a particle can form any kind of helices in any direction. Hence, we can develop a clustering algorithm that can generalise better, than just using phi calculated from the XY-plane.\n\nThanks. "
    },
    {
      "id": 455130,
      "postDate": "2019-01-13T03:18:47.273Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 356767,
      "author_name": "David Rousseau",
      "author_url": "",
      "post_date": "2018-07-14T12:26:49.430000",
      "content": "<p>The magnetic field is to a good approximation parallel to the z axis, hence the trajectories being  arcs of circle on the XY plane. So the XY plane does have a special role.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 356780,
          "author_name": "Kha Vo",
          "author_url": "",
          "post_date": "2018-07-14T12:37:54.920000",
          "content": "<p>Thanks for your answer.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 455130,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-01-13T03:18:47.273000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "356767": "The magnetic field is to a good approximation parallel to the z axis, hence the trajectories being  arcs of circle on the XY plane. So the XY plane does have a special role.",
    "356757": "Hi Organisers,\nI have one question regarding the magnetic field direction. \n\nAs the code of helix unroll is around there in the kernel section, I assume that we can cluster the hits that lie on an arc, in the XY plane. It is because the angle Phi is calculated by arctan(y/x).\n\nAs a result, let us consider the case that a true particle forms a clear arc in the XZ plane. Also assuming that this particle generates hits that all have Y=0.  As a result, in the XY plane, this particle becomes a straight line that lies on the X-axis. Therefore, this particle will be better clustered by an algorithm that calculates phi from the XZ plane instead of XY plane.\n\nWhat I'm concerned is that, is there any assumption on the magnetic field which cause the particles to form helices with common properties wrt any axis? \n\nIf the answer of the above question is No, then we can assume that a particle can form any kind of helices in any direction. Hence, we can develop a clustering algorithm that can generalise better, than just using phi calculated from the XY-plane.\n\nThanks. ",
    "455130": ""
  }
}