plot_gauss_parameters.py

import matplotlib.mlab as mlab
import numpy as np
import pylab as P


def plot_gauss_parameters(mu, covar, colorstr, delta=.1):
    '''
    %PLOT_GAUSS:  plot_gauss_parameters(mu, covar,xaxis,yaxis,colorstr)
    %
    %  Python function to plot the covariance of a 2-dimensional Gaussian
    %  model as a "3-sigma" covariance ellipse  
    %
    %  INPUTS: 
    %   mu: the d-dimensional mean vector of a Gaussian model
    %   covar: d x d matrix: the d x d covariance matrix of a Gaussian model
    %   colorstr: string defining the color of the ellipse plotted (e.g., 'r')
    '''

    # make grid
    x = np.arange(mu[0] - 3. * np.sqrt(covar[0, 0]), mu[0] + 3. * np.sqrt(covar[0, 0]), delta)
    y = np.arange(mu[1] - 3. * np.sqrt(covar[1, 1]), mu[1] + 3. * np.sqrt(covar[1, 1]), delta)
    X, Y = np.meshgrid(x, y)

    # get pdf values
    Z = mlab.bivariate_normal(X, Y, np.sqrt(covar[0, 0]), np.sqrt(covar[1, 1]), mu[0], mu[1], sigmaxy=covar[0, 1])

    P.contour(X, Y, Z, colors=colorstr, linewidths=1)

# plot_gauss_parameters(3+np.zeros((2,)), .5*np.eye(2), 'r')
# P.show()