Nx = xparticles.shape[0]
Ntheta = xparticles.shape[2]
result = zeros(3)
observations = zeros(Nx * Ntheta)
weightobs = zeros(Nx * Ntheta)
for j in range(Ntheta):
observations[(Nx * j):(Nx * (j+1))] = \
observationGenerator(xparticles[..., j], thetaparticles[:, j]).reshape(Nx)
weightobs[(Nx * j):(Nx * (j + 1))] = repeat(thetaweights[j],
repeats=Nx)
observations = power(observations, 2)
weightobs = weightobs / sum(weightobs)
obsmean = average(observations, weights=weightobs)
ind = argsort(observations)
observations = observations[ind]
weightobs = weightobs[ind]
cumweightobs = cumsum(weightobs)
quantile5 = observations[searchsorted(cumweightobs, 0.05)]
quantile95 = observations[searchsorted(cumweightobs, 0.95)]
result[0] = obsmean
result[1] = quantile5
result[2] = quantile95
return result
modelx.addPredictionList([{
"function": predictionSquaredObservations,
"dimension": 3,
"name": "squaredobs"
}])
modelx.addObsPrediction()
def predictionSquaredObservations(xparticles, thetaweights, thetaparticles, t):
Nx = xparticles.shape[0]
Ntheta = xparticles.shape[2]
result = zeros(3)
observations = zeros(Nx * Ntheta)
weightobs = zeros(Nx * Ntheta)
for j in range(Ntheta):
observations[(Nx * j):(Nx * (j+1))] = \
observationGenerator(xparticles[..., j], thetaparticles[:, j]).reshape(Nx)
weightobs[(Nx * j):(Nx * (j+1))] = repeat(thetaweights[j], repeats = Nx)
observations = power(observations, 2)
weightobs = weightobs / sum(weightobs)
obsmean = average(observations, weights = weightobs)
ind = argsort(observations)
observations = observations[ind]
weightobs = weightobs[ind]
cumweightobs = cumsum(weightobs)
quantile5 = observations[searchsorted(cumweightobs, 0.05)]
quantile95 = observations[searchsorted(cumweightobs, 0.95)]
result[0] = obsmean
result[1] = quantile5
result[2] = quantile95
return result
modelx.addPredictionList([{"function": predictionSquaredObservations, "dimension": 3, "name": "squaredobs"}])
