Error of the Bubnov–Galerkin method for elliptic boundary-value problems

In this paper the author, developing one of his own methods and making use of more refined Markov inequalities, obtains estimates for the error of the Bubnov–Galerkin method for the derivatives of the solutions of non-self-adjoint elliptic boundary-value problems. Regarding the order of accuracy, they coincide with the best known estimates of the Ritz method for the solutions of positivedefinite problems.