{
  "id": 60182,
  "title": "Feedback from organizers on simulation bugs please.",
  "url": "/competitions/trackml-particle-identification/discussion/60182",
  "author_name": "",
  "post_date": "2018-07-01T08:58:28.250543400Z",
  "votes": 8,
  "comment_count": 9,
  "views": 0,
  "content": "<p>It would be good to know about the simulator bugs.  I reported one before, but here is a beautiful one.  The particle litterally bounces.  It is particle 815158129623826432 in event train_1/event000001003\n<img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351078/9745/weird.png\" alt=\"weird\">\nColor is z axis.</p>\n\n<p>Well, I say it is a bug because the introductory document says deviation from an helix trajectory should be few millimeters only.</p>\n\n<p>Of course, reconstructing a trajectory like this is beyond hope.</p>",
  "messages": [
    {
      "id": "351078",
      "postDate": "07/01/2018 08:58:28",
      "content": "<p>It would be good to know about the simulator bugs.  I reported one before, but here is a beautiful one.  The particle litterally bounces.  It is particle 815158129623826432 in event train_1/event000001003\n<img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351078/9745/weird.png\" alt=\"weird\">\nColor is z axis.</p>\n\n<p>Well, I say it is a bug because the introductory document says deviation from an helix trajectory should be few millimeters only.</p>\n\n<p>Of course, reconstructing a trajectory like this is beyond hope.</p>",
      "rawMarkdown": "It would be good to know about the simulator bugs.  I reported one before, but here is a beautiful one.  The particle litterally bounces.  It is particle 815158129623826432 in event train_1/event000001003\n![weird][1]\nColor is z axis.\n\nWell, I say it is a bug because the introductory document says deviation from an helix trajectory should be few millimeters only.\n\nOf course, reconstructing a trajectory like this is beyond hope.\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/351078/9745/weird.png",
      "votes": null
    },
    {
      "id": "351099",
      "postDate": "07/01/2018 10:18:37",
      "content": "<p>And this one originates close to the origin, yet has z starting high and moving closer to origin as it moves away from origin on the (x,y) plane.  It is particle 459370597965627392 in same event.  The coordinates of its vertex are clearly wrong.  Moreover its trajectory is far from being a straight line.</p>\n\n<p><img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9747/Screenshot-2018-7-1%20xgb-002.png\" alt=\"data\"></p>\n\n<p><img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9748/weird2.png\" alt=\"xy\"></p>",
      "rawMarkdown": "And this one originates close to the origin, yet has z starting high and moving closer to origin as it moves away from origin on the (x,y) plane.  It is particle 459370597965627392 in same event.  The coordinates of its vertex are clearly wrong.  Moreover its trajectory is far from being a straight line.\n\n![data][1]\n\n![xy][2]\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9747/Screenshot-2018-7-1%20xgb-002.png\n  [2]: https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9748/weird2.png",
      "votes": null
    },
    {
      "id": "351105",
      "postDate": "07/01/2018 10:35:57",
      "content": "<p>Particle 585472418324152320 in train_1/event000001003 also bounces.\n<img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351105/9749/weird3.png\" alt=\"xy\">\nSeems I found a way to spot wrong trajectories, not sure it can help improving score though.</p>",
      "rawMarkdown": "Particle 585472418324152320 in train_1/event000001003 also bounces.\n![xy][1]\nSeems I found a way to spot wrong trajectories, not sure it can help improving score though.\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/351105/9749/weird3.png",
      "votes": null
    },
    {
      "id": "351412",
      "postDate": "07/02/2018 06:12:55",
      "content": "<p>What is your estimation on the frequency of the bugs?</p>\n\n<p>My estimation is that 3% of particles have wrong (impossible) tracks.</p>",
      "rawMarkdown": "What is your estimation on the frequency of the bugs?\n\nMy estimation is that 3% of particles have wrong (impossible) tracks.",
      "votes": null
    },
    {
      "id": "351434",
      "postDate": "07/02/2018 07:28:08",
      "content": "<p>I didn't compute it using the filter I'm using, but this seems consistent with what I'm seeing.  Issue is these 3% hurt supervised machine learning.</p>",
      "rawMarkdown": "I didn't compute it using the filter I'm using, but this seems consistent with what I'm seeing.  Issue is these 3% hurt supervised machine learning.",
      "votes": null
    },
    {
      "id": "354325",
      "postDate": "07/09/2018 10:15:53",
      "content": "<p>I thought I had found weird tracks but here is one that beats them all.  It is particle 747599599727804416 in event train_1/event000001002</p>\n\n<p><img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/354325/9825/weird4.png\" alt=\"weirdest\"></p>",
      "rawMarkdown": "I thought I had found weird tracks but here is one that beats them all.  It is particle 747599599727804416 in event train_1/event000001002\n\n![weirdest][1]\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/354325/9825/weird4.png",
      "votes": null
    },
    {
      "id": "359161",
      "postDate": "07/19/2018 14:48:15",
      "content": "<p>This one (particle 459370597965627392) is identified as an electron and affected by the scattering parameter scaling problem.</p>",
      "rawMarkdown": "This one (particle 459370597965627392) is identified as an electron and affected by the scattering parameter scaling problem.",
      "votes": null
    },
    {
      "id": "359162",
      "postDate": "07/19/2018 14:49:19",
      "content": "<p>I'll feed that into my re-simulation and log the trajectory i order to 'draw' the particle path for further inspection.</p>",
      "rawMarkdown": "I'll feed that into my re-simulation and log the trajectory i order to 'draw' the particle path for further inspection.",
      "votes": null
    },
    {
      "id": "365502",
      "postDate": "08/02/2018 19:40:05",
      "content": "<p>Victor! Sorry for writing here. Kaggle don't let me write private messages. :( \n\"You cannot contact users until you reach the Contributor tier\"</p>\n\n<p>I want to propose you to make a team in this competition. Hope you will read this...</p>",
      "rawMarkdown": "Victor! Sorry for writing here. Kaggle don't let me write private messages. :( \n\"You cannot contact users until you reach the Contributor tier\"\n\nI want to propose you to make a team in this competition. Hope you will read this...",
      "votes": null
    },
    {
      "id": "367010",
      "postDate": "08/06/2018 22:32:58",
      "content": "<p>This competition requires much effort to merge different solutions. So, teaming up was reasonable at the beginning, but not now (IMHO).</p>",
      "rawMarkdown": "This competition requires much effort to merge different solutions. So, teaming up was reasonable at the beginning, but not now (IMHO).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 351099,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "07/01/2018 10:18:37",
      "content": "<p>And this one originates close to the origin, yet has z starting high and moving closer to origin as it moves away from origin on the (x,y) plane.  It is particle 459370597965627392 in same event.  The coordinates of its vertex are clearly wrong.  Moreover its trajectory is far from being a straight line.</p>\n\n<p><img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9747/Screenshot-2018-7-1%20xgb-002.png\" alt=\"data\"></p>\n\n<p><img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9748/weird2.png\" alt=\"xy\"></p>",
      "votes": null,
      "replies": [
        {
          "id": 359161,
          "author_name": "asalzburger",
          "author_url": "",
          "post_date": "07/19/2018 14:48:15",
          "content": "<p>This one (particle 459370597965627392) is identified as an electron and affected by the scattering parameter scaling problem.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 351105,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "07/01/2018 10:35:57",
      "content": "<p>Particle 585472418324152320 in train_1/event000001003 also bounces.\n<img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/351105/9749/weird3.png\" alt=\"xy\">\nSeems I found a way to spot wrong trajectories, not sure it can help improving score though.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 351412,
      "author_name": "nedelko",
      "author_url": "",
      "post_date": "07/02/2018 06:12:55",
      "content": "<p>What is your estimation on the frequency of the bugs?</p>\n\n<p>My estimation is that 3% of particles have wrong (impossible) tracks.</p>",
      "votes": null,
      "replies": [
        {
          "id": 351434,
          "author_name": "cpmpml",
          "author_url": "",
          "post_date": "07/02/2018 07:28:08",
          "content": "<p>I didn't compute it using the filter I'm using, but this seems consistent with what I'm seeing.  Issue is these 3% hurt supervised machine learning.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 365502,
          "author_name": "sergeyzlobin",
          "author_url": "",
          "post_date": "08/02/2018 19:40:05",
          "content": "<p>Victor! Sorry for writing here. Kaggle don't let me write private messages. :( \n\"You cannot contact users until you reach the Contributor tier\"</p>\n\n<p>I want to propose you to make a team in this competition. Hope you will read this...</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 367010,
          "author_name": "nedelko",
          "author_url": "",
          "post_date": "08/06/2018 22:32:58",
          "content": "<p>This competition requires much effort to merge different solutions. So, teaming up was reasonable at the beginning, but not now (IMHO).</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 354325,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "07/09/2018 10:15:53",
      "content": "<p>I thought I had found weird tracks but here is one that beats them all.  It is particle 747599599727804416 in event train_1/event000001002</p>\n\n<p><img src=\"https://storage.googleapis.com/kaggle-forum-message-attachments/354325/9825/weird4.png\" alt=\"weirdest\"></p>",
      "votes": null,
      "replies": [
        {
          "id": 359162,
          "author_name": "asalzburger",
          "author_url": "",
          "post_date": "07/19/2018 14:49:19",
          "content": "<p>I'll feed that into my re-simulation and log the trajectory i order to 'draw' the particle path for further inspection.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "351078": "It would be good to know about the simulator bugs.  I reported one before, but here is a beautiful one.  The particle litterally bounces.  It is particle 815158129623826432 in event train_1/event000001003\n![weird][1]\nColor is z axis.\n\nWell, I say it is a bug because the introductory document says deviation from an helix trajectory should be few millimeters only.\n\nOf course, reconstructing a trajectory like this is beyond hope.\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/351078/9745/weird.png",
    "351099": "And this one originates close to the origin, yet has z starting high and moving closer to origin as it moves away from origin on the (x,y) plane.  It is particle 459370597965627392 in same event.  The coordinates of its vertex are clearly wrong.  Moreover its trajectory is far from being a straight line.\n\n![data][1]\n\n![xy][2]\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9747/Screenshot-2018-7-1%20xgb-002.png\n  [2]: https://storage.googleapis.com/kaggle-forum-message-attachments/351099/9748/weird2.png",
    "351105": "Particle 585472418324152320 in train_1/event000001003 also bounces.\n![xy][1]\nSeems I found a way to spot wrong trajectories, not sure it can help improving score though.\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/351105/9749/weird3.png",
    "351412": "What is your estimation on the frequency of the bugs?\n\nMy estimation is that 3% of particles have wrong (impossible) tracks.",
    "351434": "I didn't compute it using the filter I'm using, but this seems consistent with what I'm seeing.  Issue is these 3% hurt supervised machine learning.",
    "354325": "I thought I had found weird tracks but here is one that beats them all.  It is particle 747599599727804416 in event train_1/event000001002\n\n![weirdest][1]\n\n\n  [1]: https://storage.googleapis.com/kaggle-forum-message-attachments/354325/9825/weird4.png",
    "359161": "This one (particle 459370597965627392) is identified as an electron and affected by the scattering parameter scaling problem.",
    "359162": "I'll feed that into my re-simulation and log the trajectory i order to 'draw' the particle path for further inspection.",
    "365502": "Victor! Sorry for writing here. Kaggle don't let me write private messages. :( \n\"You cannot contact users until you reach the Contributor tier\"\n\nI want to propose you to make a team in this competition. Hope you will read this...",
    "367010": "This competition requires much effort to merge different solutions. So, teaming up was reasonable at the beginning, but not now (IMHO)."
  },
  "source": "meta"
}