抄録
In 1951, Gill proposed a new 4-stage 4th-order explicit Runge-Kutta method, which can diminish the round-off error in the explicit Runge-Kutta process. Moller applied the round-off error correction method to explicit Runge-Kutta process more easily. We will call the two correction methods, Gill and Moller methods, respectively. In this paper, we will compare the effectiveness of the two round-off error correction methods by numerical experiments. In addition, we will show the bad effect by optimization of compilers.
