I'm using Mathematica to find a closed-form solution for a conditional expectation. The distribution I am interested in is a mixture of two trivariate normals. In particular, I would like to know the conditional expectation of one random variable conditional on the other two being held fixed at some vector {a,b}.
The code below sets up the expectation, but I can't make it run. I believe the problem is in the part where I specify the distribution, as the other part works fine when I try it with other distributions. Any help is much appreciated!
muA = {muA1, muA2, muA3}
covA = {{varA1, covA12, covA13}, {covA12, varA2, covA23}, {covA13,
covA23, varA3}}
muB = {muB1, muB2, muB3}
covB = {{varB1, covB12, covB13}, {covB12, varB2, covB23}, {covB13,
covB23, varB3}}
Expectation[{x1} \[Conditioned] {x2, x3} == {"a", "b"}, {x1, x2,
x3} \[Distributed]
MixtureDistribution[{0.5, 0.5}, {MultinormalDistribution[muA, covA],
MultinormalDistribution[muB, covB]}]]