sdapy.gaussian_process.fit_gp

class sdapy.gaussian_process.fit_gp(x_data, y_data, yerr_data=1e-08, filters=None)

Fits data with gaussian process.

The package ‘george’ is used for the gaussian process fit.

Parameters

x_dataarray

Independent values.

y_dataarray

Dependent values.

yerr_dataarray, int

Dependent value errors.

filtersarray

If filters available, will convolve wavelengths to x_data to train gaussian process.

kernelstr, default ‘matern52’

Kernel to be used with the gaussian process. Possible choices are: ‘matern52’, ‘matern32’, ‘squaredexp’.

fix_scalebool

If fix default gaussian process param

gp_mean: str, default ‘mean’

Mean y_data function. Possible choices are: ‘mean’, ‘gaussian’, ‘bazin’, ‘villar’.

opt_routinestr,

Which technic to be used to realize optimization. Possible choices are: ‘minimize’, ‘mcmc’, ‘leastsq’.

nwalkersint

if mcmc adopted, set walker number

nsteps: int

if mcmc adopted, set step

nsteps_burnin: int

if mcmc adopted, set burnin step

clobber: bool

if gp already done, if redo it or not

Returns

Returns the interpolated independent and dependent values with the 1-sigma standard deviation.

Examples

gp = fit_gp(jd, flux, fluxerr, central_wavelength)

gp.train(gp_mean=’bazin’, opt_routine = ‘mcmc’)

gp_jd, gp_flux, gp_flux_errors, gp_ws = gp.predict()

gp.save_corner(saveplotas=’tmp.png’)

Methods

predict([x_pred, step, clobber, returnv])	output GP products
save_corner(figpath[, datadir, quantiles, ...])	generate corner plots

parse_filters
parse_wavelength
set_peak
train

__init__(x_data, y_data, yerr_data=1e-08, filters=None)

Methods

__init__(x_data, y_data[, yerr_data, filters])
parse_filters()
parse_wavelength(w)
predict([x_pred, step, clobber, returnv])	output GP products
save_corner(figpath[, datadir, quantiles, ...])	generate corner plots
set_peak([clobber])
train([kernel, fix_scale, gp_mean, ...])