chopping chains in results class

[semaphoreP]

I don't think we currently have the ability to read in posteriors from MCMC that are saved as a HDF5 file, and chop chains in the results class. This can be useful when we identify we have further burn-in that needs to be removed but don't want to rerun the whole MCMC.

[tomasstolker]

Great tool! I just did my first orbit fit :).

I have a question a bit related to the suggested enhancement by @semaphoreP.

What is the correct way to reshape the post array from Results backward to a shape in which the walkers have their own dimension?

I am guessing reshape(n_walkers, n_steps, 8) but I wasn't completely sure. Thanks!

[sblunt]

Awesome, thanks for using orbitize! @tomasstolker

Yup, what you suggested should work. Just remember that the total_orbits input to MCMC.run_sampler is actually num_steps_per_walker x num_walkers, so in your reshape command, n_steps should be the total_orbits input divided by num_walkers.

[semaphoreP]

This should be higher priority!

[semaphoreP]

This feature is in @manduhmia's plot_all_orbits branch. We should merge that branch in soon!