{
  "id": 93866,
  "title": "Distance vs Type vs Scalar Coupling Distance",
  "url": "/competitions/champs-scalar-coupling/discussion/93866",
  "author_name": "",
  "post_date": "2019-05-30T16:50:42.904318800Z",
  "votes": 9,
  "comment_count": 2,
  "views": 0,
  "content": "<p>typo: Instead of SCC, we should read SCD on the top right corner of the image.\n<img src=\"https://i.imgur.com/ZPxnqVt.png\" alt=\"distance vs type vs SCC\"></p>\n\n<p>Before all, thanks for the competition and thanks <a href=\"/artgor\">@artgor</a> and <a href=\"/inversion\">@inversion</a> for the public kernels I'm using.\nFrom the first exploration, it seems that the distance is clearly separable by the major types (1, 2 and 3). \nDoes this mean:\n- We should create individual models for each of the types?\n- Is the distance between pairs linearly correlated with scalar coupling distance?\n(EDIT: By this question, i mean linearly correlated by each of the types: 1JHC, 1JHN. ... )</p>\n\n<p>What do you guys think is the most crucial point to tackle? \nGood luck everyone!</p>",
  "messages": [
    {
      "id": "539886",
      "postDate": "05/30/2019 16:50:42",
      "content": "<p>typo: Instead of SCC, we should read SCD on the top right corner of the image.\n<img src=\"https://i.imgur.com/ZPxnqVt.png\" alt=\"distance vs type vs SCC\"></p>\n\n<p>Before all, thanks for the competition and thanks <a href=\"/artgor\">@artgor</a> and <a href=\"/inversion\">@inversion</a> for the public kernels I'm using.\nFrom the first exploration, it seems that the distance is clearly separable by the major types (1, 2 and 3). \nDoes this mean:\n- We should create individual models for each of the types?\n- Is the distance between pairs linearly correlated with scalar coupling distance?\n(EDIT: By this question, i mean linearly correlated by each of the types: 1JHC, 1JHN. ... )</p>\n\n<p>What do you guys think is the most crucial point to tackle? \nGood luck everyone!</p>",
      "rawMarkdown": "typo: Instead of SCC, we should read SCD on the top right corner of the image.\n![distance vs type vs SCC](https://i.imgur.com/ZPxnqVt.png)\n\nBefore all, thanks for the competition and thanks @artgor and @inversion for the public kernels I'm using.\nFrom the first exploration, it seems that the distance is clearly separable by the major types (1, 2 and 3). \nDoes this mean:\n- We should create individual models for each of the types?\n- Is the distance between pairs linearly correlated with scalar coupling distance?\n(EDIT: By this question, i mean linearly correlated by each of the types: 1JHC, 1JHN. ... )\n\n\nWhat do you guys think is the most crucial point to tackle? \nGood luck everyone!",
      "votes": null
    },
    {
      "id": "539899",
      "postDate": "05/30/2019 17:07:18",
      "content": "<p>This is a different way to look at the same data. I was having the same thought as you though. It does appear that the distribution of Scalar Coupling Distance is very different between types.</p>\n\n<p><img src=\"https://i.imgur.com/QMyaBFk.png\" alt=\"\"></p>",
      "rawMarkdown": "This is a different way to look at the same data. I was having the same thought as you though. It does appear that the distribution of Scalar Coupling Distance is very different between types.\n\n![](https://i.imgur.com/QMyaBFk.png)",
      "votes": null
    },
    {
      "id": "539906",
      "postDate": "05/30/2019 17:20:07",
      "content": "<p>That's right. It's a similar thought. And this also goes in line with the very n visualization <a href=\"/artgor\">@artgor</a> created (we just need to be careful about the scales):\n<img src=\"https://i.imgur.com/c6GdBkJ.png\" alt=\"\"></p>\n\n<p>Time to connect the dots on the other variables: dipole moments, mulliken charges, ...</p>",
      "rawMarkdown": "That's right. It's a similar thought. And this also goes in line with the very n visualization @artgor created (we just need to be careful about the scales):\n![](https://i.imgur.com/c6GdBkJ.png)\n\nTime to connect the dots on the other variables: dipole moments, mulliken charges, ...",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 539899,
      "author_name": "robikscube",
      "author_url": "",
      "post_date": "05/30/2019 17:07:18",
      "content": "<p>This is a different way to look at the same data. I was having the same thought as you though. It does appear that the distribution of Scalar Coupling Distance is very different between types.</p>\n\n<p><img src=\"https://i.imgur.com/QMyaBFk.png\" alt=\"\"></p>",
      "votes": null,
      "replies": [
        {
          "id": 539906,
          "author_name": "gilbertomrsilva",
          "author_url": "",
          "post_date": "05/30/2019 17:20:07",
          "content": "<p>That's right. It's a similar thought. And this also goes in line with the very n visualization <a href=\"/artgor\">@artgor</a> created (we just need to be careful about the scales):\n<img src=\"https://i.imgur.com/c6GdBkJ.png\" alt=\"\"></p>\n\n<p>Time to connect the dots on the other variables: dipole moments, mulliken charges, ...</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "539886": "typo: Instead of SCC, we should read SCD on the top right corner of the image.\n![distance vs type vs SCC](https://i.imgur.com/ZPxnqVt.png)\n\nBefore all, thanks for the competition and thanks @artgor and @inversion for the public kernels I'm using.\nFrom the first exploration, it seems that the distance is clearly separable by the major types (1, 2 and 3). \nDoes this mean:\n- We should create individual models for each of the types?\n- Is the distance between pairs linearly correlated with scalar coupling distance?\n(EDIT: By this question, i mean linearly correlated by each of the types: 1JHC, 1JHN. ... )\n\n\nWhat do you guys think is the most crucial point to tackle? \nGood luck everyone!",
    "539899": "This is a different way to look at the same data. I was having the same thought as you though. It does appear that the distribution of Scalar Coupling Distance is very different between types.\n\n![](https://i.imgur.com/QMyaBFk.png)",
    "539906": "That's right. It's a similar thought. And this also goes in line with the very n visualization @artgor created (we just need to be careful about the scales):\n![](https://i.imgur.com/c6GdBkJ.png)\n\nTime to connect the dots on the other variables: dipole moments, mulliken charges, ..."
  },
  "source": "meta"
}