Technolink Wrote:Use a long?
I see nothing wrong with working with the enormous numbers, there's many ways to still deal with them and get an accurate result.
I agree though, however confusing a normal distribution would be more accurate, and probably easier.
strafe+crit = 16 possiblities, all with different probabilities and results. It just flat out sucks.
Then again, you could always just do it arrow-by-arrow, then round up to the nearest 4th arrow, so if a 1 hit was 5 arrows, you'd know it would be a 2 hit with strafe.
A long must be converted to a double to get a right answer when multiplied with a double, else it'll only display long*(rounded down to integer double) which would be in probability "always" 0.
It's actually funny, how 0.1 + 0.1 != 0.2 in doubles. Calculation with doubles is a pain in the ass.
Binomial distribution is relevant for 2hkos. Edit: I'll add in this when I got time. I don't atm, in 1 hr or less. School.

