|
From: Robert D. <rob...@gm...> - 2017-01-19 06:05:37
|
On 2017-01-18, Michel Talon <ta...@lp...> wrote: > Well all those rationals have been computed as double floats in > rkf45.mac which is a good thing so that everything is computed as double > float by contagion. Otherwise if the function is given in terms of > rationals maxima would compute evertything to rationals and it would be > terrible. Well, that's true, but it seems preferable to put in the constants as float(mmm/nnn) instead of a literal float value -- that makes it easier to check the values against published references. best Robert Dodier |