what Ar value i should set for resistance variation in Ar/sqrt(w*l)
what else should i parametrize to do a good simulation?
and what after that. i will have monte carlo results, if fails (huge variations) what should i change to achieve better results before using a special layout techniques.
basically the deviations of device parameters is settled by guys, which are responsible for models.
But if you haven't special files with defined deviations, try to set as more as possible deviation parameters to get a worst case simulation results