cmc-obs

This library takes in a Snapshot object loaded by cmc-browser and computes various mock observations. In particular, this library will extract number density profiles, line-of-sight and both radial and tangential proper motion dispersion profiles and stellar mass functions.

This all starts with the Observations object:

import cmc_obs
k = cmc_obs.observations.Observations(snap)

which then handles the extraction of simulated observations which in most cases is as simple as calling a method:

# compute simulated Gaia proper motions
bin_centers, sigma_r, delta_sigma_r, sigma_t, delta_sigma_t, mean_mass = k.gaia_PMs()

Internally, filtering based on stellar types and magnitudes is done for each dataset to match real-world performance. For example, line-of-sight velocity dispersions are limited to bright giants while the Gaia proper motions cover a range of $13 < G< 19$.

To compute the dispersion profiles use Hamiltonian Monte Carlo, implemented in blackjax to sample from a Gaussian likelihood that is implemented in JAX which means that we don't lose all that much speed compared to the MLE approach.

The HST and LOS dispersion profiles and number density profiles are computed in a similar way.

bin_centers, sigma_r, delta_sigma_r, sigma_t, delta_sigma_t, mean_mass = k.hubble_PMs()
bin_centers, sigmas, delta_sigmas, mean_mass = k.LOS_dispersion()
bin_centers, number_density, delta_number_density, mean_mass = k.number_density()

The extraction of mass function data happens in a similar way, though it only operates in a single annulus at a time:

mass_edges, mass_function, delta_mass_function = k.mass_function(r_in=0, r_out=0.4)

This method will additionally compute the number density within the annulus and use that to calculate a reasonable limiting mass for the annulus. This replicates the crowding-based effects we see in the real data and is based on the performance of the mass function data available for 47 Tuc.

Finally, this library will handle wrangling the data into the format that GCfit expects for its ClusterFile object and will output the data in a fully formed GCfit datafile, populated with all the needed metadata to be used directly in the fitting.

See also

cmc-browser: A small library for managing a local grid of CMC models and loading the models as Snapshot objects which are needed for cmc-obs.