All notable changes to this project will be documented in this file. The format is based on Keep a Changelog and this project adheres to Semantic Versioning.
- inside the RVmodel constructor can now write
auto data = get_data()
instead ofauto data = Data::get_instance()
, for readability
- ability to analyse data from multiple instruments!!
- new
kima-template
script to create a directory with all files necessary for a run - new
get_timespan()
method of theData
class - new
load
/save
methods ofKimaResults
to load and save the model as a pickle file - ctrl+c copy to all figures
- histogram of slope when present
- save log likelihoods in the file posterior_sample_info.txt
- read log likelihoods; new function to get the highest likelihood posterior sample
- kima tips
- add "pickle" option to
kima-showresults
- new helpful analysis functions
- fail when kima_model_setup.txt is not present
- corner plot limits are adjusted if limits of the eta3 prior changed
save_setup()
function to save the current model settings for later analysis- New example with data of 51 Peg
- Getting started guide, going over the 51 Peg example
- READMEs for each example
- New arguments for the
kima-showresults
script:rv
,planets
,orbital
,gp
,extra
,diagnostic
. Numbered arguments for specific plots still work - Titles and more labels to the DNest4 diagnostic plots and other plots
- The API for the RVModel constructor was simplified. One can now do
RVmodel::RVmodel():fix(false),npmax(1)
or even justRVmodel::RVmodel()
KimaResults
now reads a config file (created bysave_setup
) to get the model settings; much less error-prone than parsing the cpp files- Changed the url for the Eigen submodule: https://github.com/eigenteam/eigen-git-mirror.git
- Easier to add new examples to the makefile
- Improve
kima-checkpriors
script - Move utility functions from
display.py
toutils.py
- Changelog!
- New methods
data.get_RV_std
anddata.get_RV_var
for standard deviation and variance of the RVs - Reference time for slope is now middle of times t[0]+0.5*timespan
Thus in models with trend, Vsys is the systemic velocity a this time. A prior for Vsys between RVmin and RVmax is therefore appropriate.
- The prior for the fiber offset is now Gaussian(15, 3), based on a fit to the "standard" stars of Lo Curto et al. (2015)
- Nasty bug in the perturb method. if(slope) should be if(trend)