Supporting Information File (text S1)

in which f(·|·) denotes conditional density function. Maximizing (2) over Φ yields the maximum likelihood estimate Φ̂. Since Φ is a 20-dimensional vector and the functions pt (Φ) and q ∗ i,t(Φ) have to be computed recursively for 1 ≤ t ≤ n, direct maximization of (2) may be computationally expensive due to the curse of dimensionality. An alternative approach is to use the EM algorithm which exploits the much simpler structure of the log likelihood l(Φ|Sn) of the data {(yt,θt), 1 ≤ t ≤ n}: l(Φ|Sn) = − 1 2 n ∑