#! /usr/bin/env python

import openturns as ot

ot.TESTPREAMBLE()

dim = 3
mean = ot.Point(dim)
sigma = ot.Point(dim)
R = ot.CorrelationMatrix(dim)
for i in range(dim):
    mean[i] = i + 0.5
    sigma[i] = 2 * i + 1.0
    for j in range(i):
        R[i, j] = 0.5 * (1.0 + i) / dim
distribution = ot.Normal(mean, sigma, R)
size = 10000
sample = distribution.getSample(size)
factory = ot.NormalFactory()
estimatedDistribution = factory.build(sample)
print("distribution=", repr(distribution))
print("Estimated distribution=", repr(estimatedDistribution))
estimatedDistribution = factory.build()
print("Default distribution=", estimatedDistribution)
estimatedDistribution = factory.build(distribution.getParameter())
print("Distribution from parameters=", estimatedDistribution)
estimatedNormal = factory.buildAsNormal(sample)
print("Normal          =", distribution)
print("Estimated normal=", estimatedNormal)
estimatedNormal = factory.buildAsNormal()
print("Default normal=", estimatedNormal)
estimatedNormal = factory.buildAsNormal(distribution.getParameter())
print("Normal from parameters=", estimatedNormal)
# 1D estimation
result = factory.buildEstimator(sample[:, 0])
estimatedDistribution = result.getDistribution()
print("Estimated distribution            =", repr(estimatedDistribution))
parameterDistribution = result.getParameterDistribution()
print("Parameter distribution            =", parameterDistribution)
# Robust estimation
factory = ot.NormalFactory(True)
estimatedDistribution = factory.build(sample)
print("distribution=", distribution)
print("Estimated distribution=", estimatedDistribution)