trackstar 1.0.0a0.dev0

Description:

trackstar 1.0.0a0.dev0

TrackStar












TrackStar is a highly optimized, user-friendly library designed to compute
Bayesian likelihood estimates for arbitrary lines and curves when the data have
errors in every direction.
When combined with your favorite optimization routine (e.g. Markov chain Monte
Carlo, maximum a posteriori esimation), TrackStar provides blazing fast
best-fit parameters!
TrackStar users enjoy the following features:


Support for samples in which some quantities are not available for
every data vector.
For example, if θ1 and θ2 are
available for the whole sample, but only half of the data have
measurements of θ3, TrackStar will automatically use
θ3 where it is available.


Multi-threading with the
OpenMP library.
If these features are enabled when TrackStar is installed, simply
tell it how many threads you'd like it to use, and hit run!


Full control over the expected N-dimensional distribution of the data,
both due to selection effects and arising from model predictions.


Developers: To-Do Items


Finish implementing unit tests


Ensure the likelihood estimates is properly normalized when the data
vectors do not have the same dimensionality (i.e. when not all
quantities are measured for each datum).


Finish writing documentation: API reference, science documentation,
installation instructions, developers documentation, Changelog


Write example codes