For that type of calculations, you do *not* want floating point numbers.
You should use one or more of the Fixed-Point data types (WWFixed, DWFixed,
etc.) and the various functions of the graphics library.
Depending on how much pain you're interested in (and/or how good/fast you
really need everything to be) you can look to implementing the math
functions that you need in a way which is tailored to your uses. Embedded
Systems magazine has had a number of articles/series on this type of thing
written by P.J. Plauger, Jack Crenshaw, and others. Good discrete
mathematics books are good resources but they often get into weird shit
rather than really useful things. :-)
Hope this helps,
John