def scenarioEstimationQuadratic3(verbose=False,seed0=1,size=350):
# test a quadratic (two linear) PGBN
# Input parameters:
# - verbose, boolean (optional): print summary
# - seed0, integer (optional): seed to which the random number generator is set at the beginning
# - size, integer (optional): length of the simulated time series
print 'Scenario: Quadratic 3; three children; two mean maps are linear, the remaining one quadratic'
anom=grv.GaussianRandomVariable('t',[],[],1)
meas1=grv.GaussianRandomVariable('x',[anom],[mm.LinearMeanMap([0.3,0.1])],(0.02)**2)
meas2=grv.GaussianRandomVariable('y',[anom],[mm.LinearMeanMap([0.2,0.05])],(0.08)**2)
meas3=grv.GaussianRandomVariable('z',[anom],[mm.QuadraticMeanMap([0.25,0.06,-0.01])],(0.05)**2)
inputPGBN=gbn.PyramidGaussianBeliefNetwork(anom,[meas1,meas2,meas3])
return testEstimation.testEstimation(inputPGBN,seed0=seed0,size=size,verbose=verbose)
def scenarioEstimationLinear(verbose=False,seed0=1,size=350):
# test a linear PGBN
# Input parameters:
# - verbose, boolean (optional): print summary
# - seed0, integer (optional): seed to which the random number generator is set at the beginning
# - size, integer (optional): length of the simulated time series
print 'Scenario: Linear; three children; all mean maps are linear'
anom=grv.GaussianRandomVariable('t',[],[],1)
meas1=grv.GaussianRandomVariable('x',[anom],[mm.LinearMeanMap([-0.1,0.7])],3e-2)
meas2=grv.GaussianRandomVariable('y',[anom],[mm.LinearMeanMap([0.8,1.2])],1e-1)
meas3=grv.GaussianRandomVariable('z',[anom],[mm.LinearMeanMap([0.2,1.4])],6e-2)
inputPGBN=gbn.PyramidGaussianBeliefNetwork(anom,[meas1,meas2,meas3])
return testEstimation.testEstimation(inputPGBN,seed0=seed0,size=size,verbose=verbose)
def scenarioEstimationStudentQuadratic3(dof,verbose=False,seed0=1,size=350):
# test a quadratic (two linear) PGBN
# Input parameters:
# - dof, integer: degrees of freedom of Student distribution
# - verbose, boolean (optional): print summary
# - seed0, integer (optional): seed to which the random number generator is set at the beginning
# - size, integer (optional): length of the simulated time series
print 'Test estimation with Student distribution with %d degrees of freedom' %dof
print 'Scenario: Quadratic 3; three children; two mean maps are linear, the remaining one quadratic'
anom=grv.GaussianRandomVariable('t',[],[],1)
meas1=grv.GaussianRandomVariable('x',[anom],[mm.LinearMeanMap([-0.1,0.8])],1e-1)
meas2=grv.GaussianRandomVariable('y',[anom],[mm.LinearMeanMap([0.1,1.2,])],9e-2)
meas3=grv.GaussianRandomVariable('z',[anom],[mm.QuadraticMeanMap([0.2,1.1,-0.25])],3e-2)
inputPGBN=gbn.PyramidGaussianBeliefNetwork(anom,[meas1,meas2,meas3])
return testEstimation.testEstimation(inputPGBN,seed0=seed0,size=size,rvtype='Student',verbose=verbose,dof=dof)
