CONAN._classes.load_rvs.add_rvGP

CONAN._classes.load_rvs.add_rvGP#

CONAN._classes.load_rvs.add_rvGP(rv_list=None, par=['col0'], kernel=['mat32'], operation=[''], amplitude=[], lengthscale=[], h3=None, h4=None, h5=None, err_col=None, gp_pck='ce', verbose=True)#

Define GP hyperparameters for each RV. The first hyperparameter h1 is amplitude (standard deviation) in RV unit while the second h2 is lengthscale in unit of the desired column. h3 and h4 are the 3rd and 4th hyperparameters whose definitions depend on the choice of gp kernel. see titans-ge/CONAN

The priors can be defined in following ways:

  • fixed value as float or int, e.g amplitude = 2

  • normal prior as tuple of len 2, (mu, std) e.g. amplitude = (2, 1)

  • uniform as tuple of length 3, (min, start, max) e.g. amplitude = (1,2,5).

  • loguniform: add “LU” to a for loguniform e.g (1,2,5,”LU”)

  • truncnorm as tuple of length 4, (min, max, mu, std) e.g. q1 = (0,1,0.3,0.1)

If multiplying GP kernels together, the amplitudes are degenerate and only one amplitude is required, that of the first kernel. set the other amplitudes to –1 to disable them.

Parameters:

  • rv_list (str, list;) – list of rv files to add GP to. Default is None for no GP. if ‘all’ is given, GP is added to all rv files where gp use has been indicated in rv_baseline(). If ‘same’ is given, a global (same) GP is used for all indicated RV files in rv_baseline().

  • par (str, tuple, list;) – column of the input data to use as the GP independent variable. a list is expected if different columns are to be used for the rv files given in rv_list. To use 2 different kernels on a single rv file, give column name for each kernel as a tuple of length 2. e.g. rv_list=[‘rv1.dat’,’rv2.dat’], par = [(‘col0’,’col0’),’col3’] to use col0 for both kernels of rv1, and col3 for rv2.

  • kernel –

    kernel to use for the GP.

    • if George package, kernel must be in [‘mat32’, ‘mat52’, ‘exp’, ‘cos’, ‘expsq’,’exps2’,’qp’,’rquad’]

    • if celerite package, kernel must in [‘mat32’, ‘exp’, ‘cos’, ‘sho’,’qp_ce’]

    • if spleaf package, kernel must be in [‘mat32’, ‘mat52’, ‘exp’, ‘cos’, ‘sho’, ‘expsq’, ‘exps2’, ‘qp’, ‘qp_sc’, ‘qp_mp’]

operationstr, tuple, list;

operation to combine kernels. Must be one of [“+”,”*”]. Default is “” for no combination.

amplitudefloat, tuple, list;

amplitude of the GP kernel. Must be list of int/float or tuple of length 2/3/4

lengthscalefloat, tuple, list;

lengthscale of the GP kernel. Must be int/float or tuple of length 2/3/4

h3float, tuple, list;

3rd hyperparameter of the GP kernel. Must be list of int/float or tuple of length 2/3/4

h4float, tuple, list;

4th hyperparameter of the GP kernel. Must be list of int/float or tuple of length 2/3/4

h5float, tuple, list;

amplitude of the derivative of the GP kernel.

err_col: tuple, list;

column of the uncertainties for the gp axis. only relevant for spleaf multidimensional GP

gp_pckstr, list;

package to use for the GP. Must be one of [“ge”,”ce”,”sp”]. Default is “ce” for celerite.

verbosebool;

print output. Default is True.

_rvGP_dict#

dictionary of rv GP parameters

Type:

dict;

_sameRVgp#

namespace object to indicate if same GP is used for all RVs

Type:

SimpleNamespace;

_allRVgp#

whether all RVs have GP

Type:

bool;

_useGPrv#

list of ‘n’, ‘ce’,’ge’,’sp’ for each light curve indicating if a GP is to be fitted.

Type:

list;

_gp_rvs#

list of rvs with GP fitting

Type:

list;

Contents