
104 Statistical Computing in Nuclear Imaging
draws are not possible methods such as Markov chains discussed in Section 4.4
can be used.
Total error and sample error:
The histogram shown in Figure 4.1 at the top can be used to est imate the val-
ues of marginal distribution of p(f
1
) at th e centers of each bar of the histogram.
This is done by dividing t he number of samples obtained in ea ch histogram
bin by the total number of samples and the histogram width. Equation (4.10)
provides an estimate of the total error of su ch estimation. Figure 4.2 schema ti-
cally presen ts the extent of the total error. The bars are indicative of the square
root of