search

nmdickson / GCfit

Multimass (LIMEPY) globular cluster equilibrium models and fitting them to observations
https://gcfit.readthedocs.io
MIT License
3 stars 1 forks source link

Add testing framework #13

Open nmdickson opened 4 years ago

nmdickson commented 4 years ago

It's probably past time to add things like unit tests, just to make sure changes aren't breaking things like the likelihood functions or data reading or etc.

nmdickson commented 3 years ago

Two testing scripts, for the data loading, etc, and for the likelihoods, etc, make the most sense to me.

This will require creating a test dataset, i.e. a test cluster hdf resource. It's going to need to have all the usual trappings from any cluster, and be able to run through every likelihood, so I think for now I'll borrow the data from both M62 and 47Tuc, where they exist. If necessary (or expediant) I'll adjust the actual variable values as we go.

nmdickson commented 3 years ago

I've also placed the PM_tot and PM_ratio variables into the high_mass subgroup, as it's a good idea to test on these subgroups, but you cannot have a mix of variables inside and outside subgroups, (whenever a subgroup exists it removes the parent group, in data). I'm not sure when that decisions was made but it's how things stand for now so it's how we'll test as well.

nmdickson commented 3 years ago

Significant progress made in 357533d on data tests.

nmdickson commented 3 years ago

This should be bumped up in priority now, especially testing that the fitting scripts can run correctly, in order to easily check that new branches will function before they are PR'd.