For example in Non Linear computer simulations dealing with instabilities. By 'denser near zero' I mean that 0+10^{-30}=10^{-30} and 1+10^{-30}=1. Thus, the systematic error on computed real numbers depends on where you are in [-1,1], if you use normalized parameters.

I changed the title as it was not clear: I am not dealing with random number generator.

More Nathalie Olivi-Tran's questions See All
Similar questions and discussions