For the method order and the error constant detection was adopted an approximate method. In practice, there are a few number of methods of which order overpass 11. Suppose y₁ is the selected 'final' variable which approximate y(t₀+ph). Then

holds.

The method order is the largest i for which c_i=0.

We can find the first eleven coefficients if we compute the values of y₁ using the method with stepsize h, 2h,..., 11h, and we form a linear system in the unknowns c_i, i=1,...,11. We consider that c_i is zero if |c_i|< h.

Example for the method DIRK4.