https://github.com/johannesbuchner/extinctionevents
Bayesian Recurring Event analysis
https://github.com/johannesbuchner/extinctionevents
Last synced: 7 months ago
JSON representation
Bayesian Recurring Event analysis
- Host: GitHub
- URL: https://github.com/johannesbuchner/extinctionevents
- Owner: JohannesBuchner
- Created: 2015-10-21T14:40:13.000Z (almost 10 years ago)
- Default Branch: master
- Last Pushed: 2015-10-23T17:19:32.000Z (almost 10 years ago)
- Last Synced: 2025-01-23T20:52:00.416Z (9 months ago)
- Language: Python
- Size: 852 KB
- Stars: 0
- Watchers: 4
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.rst
Awesome Lists containing this project
README
Recurring Event analysis
----------------------------
An attempt at a Bayesian re-analysis of Rampino&Caldeira 2015
* Article: Rampino&Caldeira 2015 http://mnras.oxfordjournals.org/content/454/4/3480.abstract
* Discussion thread: https://www.facebook.com/groups/astro.r/permalink/889900187772654/
Model for the event probability
---------------------------------
Code: extinction.py
A flat distribution plus a Gaussian which is repeated.
Parameters:
* s Strength of the Gaussian (0-1, remainder is in the flat distribution)
* w Width/Duration [in Myrs]
* p Period [in Myrs]
* phase Phase [0-1]
Interpretation
----------------
1. If a significant recurring signal is in the data, s should by > 0 and p should be constrained.
2. If the data is uniformly random (not periodic), s should be 0 and all signal parameters (s, w, p) should be unconstrained.
Analysis of Mock data
-----------------------
* Data: extinction_testdata.txt
* generated events at [ 15, 35, 55, 75, 95, 115, 135, 155, 175, 195, 215, 235] Myrs
* and 5 randomly placed events
* with uncertainty of 0.1 Myrs
* Data and posterior of the model: extinction_testdata.txt_gauss_predict.png
* Parameter posterior distributions: See file extinction_testdata.txt_gaussmarg.png
Conclusion: Signal detected/recovered.
Analysis of Real data
------------------------
* Data: extinction.txt
* from the article
* Data and posterior of the model: extinction.txt_gauss_predict.png
* Parameter posterior distributions: See file extinction.txt_gaussmarg.png
Conclusion: No signal detected. In particular the period posterior distribution is flat.