yes, floats are cheaper but i think this only really matters for heavy number crunching. probably it doesn't make much difference for passing some numbers around.

(in python there only are doubles. single precision floats don't exist.)

so i think doubles are fine but i also wouldn't mind the needed f that much.