x | is parameter arguments to the function, other components are nan. |
x_index | is the variable index, on the tape, for the arguments to this function. This size of x_index is n, the number of arguments to this function. |
y_index | is the variable index, on the tape, for the results for this function. This size of y_index is m, the number of results for this function. |
for_jac_sparsity | On input, for j = 0, ... , n-1, the sparsity pattern with index x_index[j], is the forward Jacobian sparsity for the j-th argument to this atomic function. |
rev_jac_flag | This shows which variables affect the function we are computing the Hessian of. On input, for i = 0, ... , m-1, the rev_jac_flag[ y_index[i] ] is true if the Jacobian of function (we are computing sparsity for) is no-zero. Upon return, for j = 0, ... , n-1, rev_jac_flag [ x_index[j] ] as been adjusted to accound removing this atomic function. |
rev_hes_sparsity | This is the sparsity pattern for the Hessian. On input, for i = 0, ... , m-1, row y_index[i] is the reverse Hessian sparsity with one of the partials with respect to to y_index[i]. |