On the maximum order of optimal B-convergence of Lobatto III C schemes (Q1096342)

A discretization method is said to be optimally B-convergent with order p, if for all initial value problems \(y'(t)=f(t,y(t)),\) \(t>0\), \(y(0)=y_ 0\), with \((f(t,y)-f(t,z),y-z)\leq m\| y-z\|^ 2\) the numerical solution \((y_ n)\) satisfies \((*)\quad \| y(t_ n)-y_ n\| \leq C(t_ n)h^ p\) where the function C only depends on the one-sided Lipschitz constant m and the derivatives of the solution. In this context, the paper completes previous results of the author on the B- convergence of implicit Runge-Kutta methods of Lobatto III C type. In fact, after proving in Numer. Math. 51, 229-235 (1987; Zbl 0626.65065) the existence of an error bound (*) with \(p=s-1\), for all s-stage methods with \(s\geq 3\) and \(m<0\), here the author shows that \(p=s-1\) is the highest possible order of all s-stage Lobatto III C methods with more than three states for strictly dissipative problems (i.e. \(m<0)\).