Latin Hypercubes

LH stands for Latin Hypercube. An OLH is computed from an LH, so OLH stands for Optimized Latin Hypercube. The ideal OLH situation is when the entire sample space is sampled evenly, i.e. divided into equally-probable subspaces (with the same density). In this case, we should speak about an Orthogonal Latin Hypercube.

However, finding such samples is a very difficult combinatorial problem for which only reasonable approximations can be found in practice, at the end of an optimization procedure. This is exactly what we do in Simcenter Amesim and why we speak about Optimized Latin Hypercubes.