You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Do we really want to return in any cases/platforms float32? (Changing this could break usage of cfunits.)
I believe that every 4 byte integer could be stored in a 8 byte float (double) without losing precision. Therefore it could be more accurate to use numpy.float64 in all cases. Further for an integer with more than 4 bytes a float32 is not sufficient. What do you think?