Floating-point Precision errors

[amolgorithm]

There were often some errors in the precision of even simple floating-point results.
For example, the application evaluated (0.2)^2 as 0.04000...01, when the application could instead simply keep it as 0.04 .