store as two integers \((x,y)\) evaluate as \(x*2^y\) this is binary floating point this means you get inaccuracies
eg \((0.1+0.2-0.3)*10^{20}\) is not zero
alternative is decimal floating point store as \(x*10^y\)
The need to approximate real operations with pseudo real numbers. If we round small values to 0, then ln 0, x/0 break. This is underflow