{
  "id": 611375,
  "title": "This competition has officially pushed science forward. Congrats!!!",
  "url": "/competitions/leap-atmospheric-physics-ai-climsim/discussion/611375",
  "author_name": "Jerry Lin",
  "post_date": "2025-10-10T19:35:25.633000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>Just wanted to share that the results from this competition have already resulted in a paper that pushes the field of hybrid physics-ML climate simulation forward! The architecture from the 5th place winner, difference loss from the 2nd place winner, and the confidence loss design from the 1st place winner were used to create hybrid physics-ML climate simulations in ICON that combines ML and a traditional physics-based parameterization based on the uncertainty from confidence loss:</p>\n<p><a href=\"https://arxiv.org/abs/2510.08107\" target=\"_blank\">https://arxiv.org/abs/2510.08107</a></p>\n<p>EDIT: the paper I've been working on is finally in review!</p>\n<p>Here's the blog post summarizing it:</p>\n<p><a href=\"https://coupledsystems.leaflet.pub/3m6zutjos4s2e\" target=\"_blank\">https://coupledsystems.leaflet.pub/3m6zutjos4s2e</a></p>\n<p>and the preprint itself:</p>\n<p><a href=\"https://arxiv.org/abs/2511.20963\" target=\"_blank\">https://arxiv.org/abs/2511.20963</a></p>\n<p>Best regards,</p>\n<p>Jerry</p>",
  "messages": [
    {
      "id": 3300583,
      "postDate": "2025-10-10T19:35:25.633Z",
      "content": "<p>Hi everyone,</p>\n<p>Just wanted to share that the results from this competition have already resulted in a paper that pushes the field of hybrid physics-ML climate simulation forward! The architecture from the 5th place winner, difference loss from the 2nd place winner, and the confidence loss design from the 1st place winner were used to create hybrid physics-ML climate simulations in ICON that combines ML and a traditional physics-based parameterization based on the uncertainty from confidence loss:</p>\n<p><a href=\"https://arxiv.org/abs/2510.08107\" target=\"_blank\">https://arxiv.org/abs/2510.08107</a></p>\n<p>EDIT: the paper I've been working on is finally in review!</p>\n<p>Here's the blog post summarizing it:</p>\n<p><a href=\"https://coupledsystems.leaflet.pub/3m6zutjos4s2e\" target=\"_blank\">https://coupledsystems.leaflet.pub/3m6zutjos4s2e</a></p>\n<p>and the preprint itself:</p>\n<p><a href=\"https://arxiv.org/abs/2511.20963\" target=\"_blank\">https://arxiv.org/abs/2511.20963</a></p>\n<p>Best regards,</p>\n<p>Jerry</p>",
      "rawMarkdown": "Hi everyone,\n\nJust wanted to share that the results from this competition have already resulted in a paper that pushes the field of hybrid physics-ML climate simulation forward! The architecture from the 5th place winner, difference loss from the 2nd place winner, and the confidence loss design from the 1st place winner were used to create hybrid physics-ML climate simulations in ICON that combines ML and a traditional physics-based parameterization based on the uncertainty from confidence loss:\n\nhttps://arxiv.org/abs/2510.08107\n\nEDIT: the paper I've been working on is finally in review!\n\nHere's the blog post summarizing it:\n\nhttps://coupledsystems.leaflet.pub/3m6zutjos4s2e\n\nand the preprint itself:\n\nhttps://arxiv.org/abs/2511.20963\n\nBest regards,\n\nJerry\n\n",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3300583": "Hi everyone,\n\nJust wanted to share that the results from this competition have already resulted in a paper that pushes the field of hybrid physics-ML climate simulation forward! The architecture from the 5th place winner, difference loss from the 2nd place winner, and the confidence loss design from the 1st place winner were used to create hybrid physics-ML climate simulations in ICON that combines ML and a traditional physics-based parameterization based on the uncertainty from confidence loss:\n\nhttps://arxiv.org/abs/2510.08107\n\nEDIT: the paper I've been working on is finally in review!\n\nHere's the blog post summarizing it:\n\nhttps://coupledsystems.leaflet.pub/3m6zutjos4s2e\n\nand the preprint itself:\n\nhttps://arxiv.org/abs/2511.20963\n\nBest regards,\n\nJerry\n\n"
  }
}