|
From: Michel T. <ta...@lp...> - 2017-01-18 15:00:19
|
Le 18/01/2017 à 15:46, Michel Talon a écrit : > First b line > 0.1185185185185185 0 0.5189863547758284 0.5061314903420167 −0.18 > 0.03636363636363636 > I don't know where these numbers appear, the last one is in truc_error. > > second b line > 0.1157407407407407 0 0.5489278752436647 0.5353313840155945 <-- 46 bfloat > -0.2 0 In fact i have seen in https://en.wikipedia.org/wiki/Runge%E2%80%93Kutta_methods that to estimate the error one substracts the two b lines. But then 0.1185185185185185-0.1157407407407407=0.0027777777777.. 0.5189863547758284-0.5489278752436647=- 0.0299415204678.. 0.5061314903420167-0.5353313840155945=- 0.02919989367357789 etc. which are all the numbers appearing in trunc_errors. So i think one can safely say that rkf45.mac is correct. -- Michel Talon |