STK_MATERNCOV32_ANISO computes the anisotropic Matern covariance with nu=3/2

 CALL: K = stk_materncov32_aniso (PARAM, X, Y)

   computes the covariance matrix K between the sets of locations X and Y,
   using the anisotropic Matern covariance function with nu=3/2 and
   parameters PARAM. The output matrix K has size NX x NY, where NX is the
   number of rows in X and NY the number of rows in Y. The vector of
   parameters must have DIM + 1 elements, where DIM is the common number
   of columns of X and Y:

     * PARAM(1) = log (SIGMA ^ 2), where SIGMA is the standard deviation,

     * PARAM(1+i) = - log (RHO(i)), where RHO(i) is the range parameter
       for the ith dimension.

 CALL: dK = stk_materncov32_aniso (PARAM, X, Y, DIFF)

   computes the derivative of the covariance matrix with respect to
   PARAM(DIFF) if DIFF~= -1, or the covariance matrix itself if DIFF is
   equal to -1 (in which case this is equivalent to stk_materncov_aniso
   (PARAM, X, Y)).

 CALL: K = stk_materncov32_aniso (PARAM, X, Y, DIFF, PAIRWISE)

   computes the covariance vector (or a derivative of it if DIFF > 0)
   between the sets of locations X and Y.  The output K is a vector of
   length N, where N is the common number of rows of X and Y.