Why is the Cholesky decomposition preferred to sqrtm in filtering algorithms?

237 Views Asked by At

Many nonlinear filtering algorithms like UKF and CKF make use of Cholesky decomposition. Whay can't we go for $sqrtm(P)$, such that $SS=P$, rather than $SS^{T}=P$? I have seen many answers stating that Cholesky reduces computational time and is light on storage space. Are there any other reasons?