It’s an issue with floating point arithmetic in general. Any math library that uses floats will be affected the same way. A classic example is 0.1 + 0.2 = 0.30000000000000004
Perhaps adding a default significant figure (configurable perhaps) for rounding will help.