9. Random sampling

lsqfitgp.raniter(mean, cov, n=None, eps=None)

Take random samples from a multivariate Gaussian.

This generator mimics the interface of gvar.raniter, but takes as input the mean and covariance separately instead of a collection of gvars.

Parameters:

meanscalar, array, or dictionary of scalars/arrays

The mean of the Gaussian distribution.

covscalar, array, or dictionary of scalars/arrays

The covariance matrix. If mean is a dictionary, cov must be a dictionary with pair of keys from mean as keys.

nint, optional

The maximum number of iterations. Default unlimited.

epsfloat, optional

Used to correct the eigenvalues of the covariance matrix to handle non-positivity due to roundoff, relative to the largest eigenvalue. Default is number of variables times floating point epsilon.

Yields:

sampscalar, array, or dictionary of scalars/arrays

The random sample in the same format of mean.

Examples

>>> mean = {'a': np.arange(3)}
>>> cov = {('a', 'a'): np.eye(3)}
>>> for sample in lgp.raniter(mean, cov, 3):
>>>     print(sample)

lsqfitgp.sample(*args, **kw)

Shortcut for next(raniter(*args, **kw)).