What do you mean by 1?1/m1?1/m ?
Even if interpreted as (1?1/m)(1?1/m) your probabilities don't add up to 1. Is this your intention?
If you chose n^2/m^2 ones and and put them into your matrix at randomly chosen off-diagonal positions the probability of finding a one should be 1/m^2.
In this case you should arrive at n^2 - n^2/m^2 zeros, the probability of which is then 1 - 1/m^2.
