Utilities

bayesian_quadrature.util.find_good_parameters(logpdf, x0, method, ntry=10)

bayesian_quadrature.util.hlines(ax, y, **kwargs)

bayesian_quadrature.util.improve_conditioning(gp)

bayesian_quadrature.util.improve_tail_covariance(gp)

bayesian_quadrature.util.set_scientific(ax, low, high, axis=None)

Set the axes or axis specified by axis to use scientific notation for ticklabels, if the value is <10**low or >10**high.

Parameters :

ax : axis object

The matplotlib axis object to use

low : int

Lower exponent bound for non-scientific notation

high : int

Upper exponent bound for non-scientific notation

axis : str (default=None)

Which axis to format (‘x’, ‘y’, or None for both)

bayesian_quadrature.util.slice_sample(logpdf, niter, w, xval, nburn=1, freq=1)

Draws samples from ‘logpdf’, optionally starting from ‘xval’. The pdf should return log values.

Parameters :

logpdf : function

Target distribution. logpdf(xval) should return ln(Pr(xval))

niter : int

Number of iterations to run

w : np.ndarray

The step by which to adjust the window size.

xval : numpy.ndarray

The initial starting value.

nburn : int (default 1)

Number of samples to skip at the beginning

freq : int (default 1)

How often to record samples

bayesian_quadrature.util.vlines(ax, x, **kwargs)