![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2F6a5f75221bd1324ab0ae18995be27a7b%2Fanimated_transit_0.gif?generation=1758890168409867&alt=media\"){kind=spacer}
Before making my writeup, I wanted to create some fun animations of the effect of limb darkening on transits. I have done that now, and the notebook for creating such figures and animations is shared here.
\nLimb darkening means that the intensity of the star is weaker toward the edge of the star than the center. Without limb darkening, the valley of the transit is very flat, because the observed light mainly depends on whether the planet is fully inside the boundary of the star or not. This is shown in the following animation:
\n
With limb darkening, the observed light continues being reduced as the planet gets closer to the center of the star, because it then covers an area with higher intensity. Here is the same animation as above, but with a large degree of limb darkening:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2Fb204454b61cda26ce99efcdb5a73601b%2Fanimated_transit_1.gif?generation=1758890395171246&alt=media\"){kind=spacer}
Note that the depth of the transit significantly deviates from the depth observed without limb darkening. There are other properties that can cause significant deviations, such as the planet not getting close to the center of the star:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2F5496157dd3312641c513cdbffb78f71b%2Fanimated_transit_2.gif?generation=1758890950937779&alt=media\"){kind=spacer}
Depending on the range of parameters used by the organizers, there may therefore be other significant outliers in the data than just those caused by clipped transits.
\nI have described the parameterization of my model in the shared notebook, and you can try out some other settings if you would like. Here are some still images from a range of other transits:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2Fc991f667e361ee5732d7b0f04d1b239a%2Ftransit_3.png?generation=1758891247197555&alt=media\"){kind=spacer}
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2F5aeff55ec00436a42a20001b5ce5df4f%2Ftransit_4.png?generation=1758891264497196&alt=media\"){kind=spacer}
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2Fa730c92c294a88fefeb61bc80561fad6%2Ftransit_5.png?generation=1758891285713259&alt=media\"){kind=spacer}
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2Fcea9bf7c029c45f63dcfd248055872f1%2Ftransit_6.png?generation=1758891302851905&alt=media\"){kind=spacer}
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2Fa30421229ba401ad8a3a35dc2cb6550c%2Ftransit_7.png?generation=1758891317199041&alt=media\"){kind=spacer}
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2F27ccdf83b6bca8c19666b753e65e3b92%2Ftransit_8.png?generation=1758891331211989&alt=media\"){kind=spacer}
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2Fe7c500ad79386db43a23b3cbee6b4a0d%2Ftransit_9.png?generation=1758891345016827&alt=media\"){kind=spacer}
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11013868%2F0535cbe04a39d6089373f76339f6cb82%2Ftransit_10.png?generation=1758891354900973&alt=media\"){kind=spacer}
Before making my writeup, I wanted to create some fun animations of the effect of limb darkening on transits. I have done that now, and the notebook for creating such figures and animations is shared here.
\nLimb darkening means that the intensity of the star is weaker toward the edge of the star than the center. Without limb darkening, the valley of the transit is very flat, because the observed light mainly depends on whether the planet is fully inside the boundary of the star or not. This is shown in the following animation:
\n
With limb darkening, the observed light continues being reduced as the planet gets closer to the center of the star, because it then covers an area with higher intensity. Here is the same animation as above, but with a large degree of limb darkening:
\n
Note that the depth of the transit significantly deviates from the depth observed without limb darkening. There are other properties that can cause significant deviations, such as the planet not getting close to the center of the star:
\n
Depending on the range of parameters used by the organizers, there may therefore be other significant outliers in the data than just those caused by clipped transits.
\nI have described the parameterization of my model in the shared notebook, and you can try out some other settings if you would like. Here are some still images from a range of other transits:
\n
Nice visualization!
\nBut what about orbital parameters like P, i, e, etc which have quite significant impact on the transit curve? :)
I don't know. I did not use any of the parameters from train_star_info.csv in my solution. It would be relatively straightforward to change the trajectory of the planet to a polynomial instead of a straight line, but I did not have time to look into stuff like that either.
", "rawMarkdown": "I don't know. I did not use any of the parameters from train_star_info.csv in my solution. It would be relatively straightforward to change the trajectory of the planet to a polynomial instead of a straight line, but I did not have time to look into stuff like that either." }, { "id": 3294772, "postDate": "2025-09-26T17:22:02.940Z", "content": "i, and a are used to compute d,min. P and a are used to compute the transit duration.
\nBoth are captured by the parameters used in these visualizations.
", "rawMarkdown": "i, and a are used to compute d,min. P and a are used to compute the transit duration.\n\nBoth are captured by the parameters used in these visualizations.", "votes": 1, "replies": [ { "id": 3294787, "postDate": "2025-09-26T18:11:13.727Z", "content": "I see. My impression was that the data in train_star_info.csv was too noisy to be useful, but I did not explicitly compute d_min and transit duration from this data.
\nI guess Oleh was not referring to the trajectory of the planet then. I also think it would be a bit strange if the curvature of the orbit was visible during the short duration that the planet passes in front of the star, so using a straight trajectory is likely better.
", "rawMarkdown": "I see. My impression was that the data in train_star_info.csv was too noisy to be useful, but I did not explicitly compute d_min and transit duration from this data.\n\nI guess Oleh was not referring to the trajectory of the planet then. I also think it would be a bit strange if the curvature of the orbit was visible during the short duration that the planet passes in front of the star, so using a straight trajectory is likely better." } ] } ] } ], "comments": [ { "id": 3294630, "author_name": "Oleh Kivernyk", "author_url": "", "post_date": "2025-09-26T13:07:31.397000", "content": "Nice visualization!
\nBut what about orbital parameters like P, i, e, etc which have quite significant impact on the transit curve? :)
I don't know. I did not use any of the parameters from train_star_info.csv in my solution. It would be relatively straightforward to change the trajectory of the planet to a polynomial instead of a straight line, but I did not have time to look into stuff like that either.
", "votes": 0, "replies": [] }, { "id": 3294772, "author_name": "CPMP", "author_url": "", "post_date": "2025-09-26T17:22:02.940000", "content": "i, and a are used to compute d,min. P and a are used to compute the transit duration.
\nBoth are captured by the parameters used in these visualizations.
", "votes": 1, "replies": [ { "id": 3294787, "author_name": "Thomas Dueholm Hansen", "author_url": "", "post_date": "2025-09-26T18:11:13.727000", "content": "I see. My impression was that the data in train_star_info.csv was too noisy to be useful, but I did not explicitly compute d_min and transit duration from this data.
\nI guess Oleh was not referring to the trajectory of the planet then. I also think it would be a bit strange if the curvature of the orbit was visible during the short duration that the planet passes in front of the star, so using a straight trajectory is likely better.
", "votes": 0, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3294628": "Before making my writeup, I wanted to create some fun animations of the effect of limb darkening on transits. I have done that now, and the notebook for creating such figures and animations [is shared here](https://www.kaggle.com/code/thomasdueholmhansen/ariel-2025-transit-visualizer).\n\nLimb darkening means that the intensity of the star is weaker toward the edge of the star than the center. Without limb darkening, the valley of the transit is very flat, because the observed light mainly depends on whether the planet is fully inside the boundary of the star or not. This is shown in the following animation:\n\n\n\nWith limb darkening, the observed light continues being reduced as the planet gets closer to the center of the star, because it then covers an area with higher intensity. Here is the same animation as above, but with a large degree of limb darkening:\n\n\n\nNote that the depth of the transit significantly deviates from the depth observed without limb darkening. There are other properties that can cause significant deviations, such as the planet not getting close to the center of the star:\n\n\n\nDepending on the range of parameters used by the organizers, there may therefore be other significant outliers in the data than just those caused by clipped transits.\n\nI have described the parameterization of my model in [the shared notebook](https://www.kaggle.com/code/thomasdueholmhansen/ariel-2025-transit-visualizer), and you can try out some other settings if you would like. Here are some still images from a range of other transits:\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n", "3294630": "Nice visualization!\nBut what about orbital parameters like P, i, e, etc which have quite significant impact on the transit curve? :) " } }