Dear Sergei,

How does the ConstructCategoryMatrix command work when the likelihood function uses more than one category variable? This is not clear from the documentation.

regards,
Konrad

Dear Konrad,

If you have N category variables with C_1, ... C_N rate classes each, ConstructCategoryMatrix will generate a matrix with C_1 x C_2 ... x C_N rows and a column for each site. You can index into the column as you would into a normal multi-dimensional array in C, but using explicit indices. For example, for two category variables, to get the value for class i in variable 1 and class j in variable 2, you would use the index i*C_2 + j (assuming i and j are 0-based). For 3 variables:
i*C_2*C_3 + j*C_3 + k etc.

To determine the order for the category variables for indexing, you can call the function GetInformation (order_ID, lf_ID), and have the names of category variables stored in the matrix referred to by order_ID.

Cheers,
Sergei