Is there a standard way to get a more precise floating point number?