Configuration Settings#

Chronostar has many configurable settings. You can configure them via a config file or by calling each class’s configure class (static) method.

See Command-Line Interface for example config files.

Here we provide an exhaustive list of all configurable settings and their uses.

Components#

SpaceComponent#

reg_covar: float, default 1.e-6

A regularisation constant added to the diagonals of the covariance matrix

nthreads: int, default None

Number of OMP threads used by numpy matrix operations

SphereSpaceTimeComponent#

reg_covar: float, default 1.e-6

A regularisation constant added to the diagonals of the covariance matrix

minimize_method: str, default ‘Nelder-Mead’

The method used by scipy.optimize.minimize. Must be one of

  • ‘Nelder-Mead’ (recommended)

  • ‘Powell’ (not receommended)

nthreads: int, optional

Manually restrict how many threads openMP tries to use when executing optimized numpy functions

trace_orbit_func: Callable: f(start_loc, time), default trace_epicyclic_orbit()

A function that traces a position by time Myr. Positive time traces forward, negative time backwards

age_offset_interval: int, default 20

After how many calls to maximize() age offsets should be explored

stellar_uncertainties: bool, default True

Whether data covariance matrices are encoded in final 36 columns of input data X

resp_tol: float, default 1e-6

Only samples with a responsibility (membership probability) greater than resp_tol will be included in the evaluation of the loss function

max_agefloat, default 200

The upper bound on the age when maximising the component parameters

Mixtures#

ComponentMixture#

tol: float, default 1e-3

Used to determine convergence by sklearn’s EM algorithm. Convergence determined if “change” between EM iterations is less than tol, where change is the difference between the average log probability of each sample

reg_covar: float, default 1e-6

A regularization constant added to the diagonals of covariance matrices

max_iter: int, default 100

The maximum iterations for sklearn’s EM algorithm

n_init: int, default 1

(included only for sklearn API compatbility, ignored)

init_params: str, default ‘random’

The initialization approach used by sklearn if component parameters aren’t pre set. Must be one of

  • ‘init_resp’: responsibilites are taken from input

  • ‘kmeans’: responsibilities are initialized using kmeans.

  • ‘k-means++’: use the k-means++ method to initialize.

  • ‘random’: responsibilities are initialized randomly.

  • ‘random_from_data’: initial means are randomly selected data points.

random_state: int, default None

Controls the random seed given to the method chosen to initialize the parameters (see init_params). In addition, it controls the generation of random samples from the fitted distribution. Pass an int for reproducible output across multiple function calls.

warm_start: bool, default True

(leave True for correct interactions between self and self.sklmixture)

verbose: int, default 0

Whether to print sklearn statements:

  • 0: no output

  • 1: prints current initialization and each iteration step

  • 2: same as 1 but also prints log probability and execution time

verbose_interval: int, default 10

If verbose > 0, how many iterations between print statements

ICPools#

SimpleICPool#

max_components: int, default 100

The max components in an initial condition provided by SimpleICPool

GreedyCycleICP#

max_componentsint, default 100

The max components in an initial condition provided by SimpleICPool, configurable

index_from_frontbool, default True

Whether the target component index is a positive number (hence counting from front) or a negative number (hence counting from the back). Because the component list is modified in place, indexing from front results in any added component being split on the next iteration. Where as indexing from back skips the added component. For quicker, less accurate runs, set index_from_front=True. For slower, more accurate runs, set index_from_front=False.

Drivers#

Driver#

intermediate_dumps: bool, default True

Whether to write to file the results of mixture model fits

savedir: str, default ‘./result’

Path to the directory of where to store results

Runs#

The three command line tools for fitting ( fit-chronostar, fit-mixture, and fit-component ) have “run level” parameters.

nthreads: int, default 1

Provided for future high-level parallelism. Currently nothing is implemented, so leave this at 1

savedir: str, default “./result”

The output directory for the final results. This can be the same directory that ComponentMixture uses to store intermediate dumps.