Confidence Limits on Simulation Averages 1.
Given a set of measurements {mi}, for some property M
There exists a distribution of values from which these measurements were sampled
We do not know, a priori, any details of this distribution
We wish to use our measurements {mi} to estimate the mean of the true distribution
Not surprisingly, the best estimate of the mean of the true distribution is given by the mean of the sample
We need to quantify our confidence in the value of this estimate for the mean
We must do this using only the sample data
{0.01, 0.1, 0.9, 0.06, 0.5, 0.3, 0.02}
<m> = (0.01+0.1+0.9+0.06+0.5+0.3+0.02)/7 = 0.27