It is true that mathematically is an issue. But, probabilistically, generating a is equivalent to generating a in that 0 and 1 have a zero probability to occur. From a software point of view, good random generators cannot return those extreme values….

]]>On page 39 of Monte Carlo Statistical Methods, for , the set of is . (1 can also be a trouble in some cases)

So either you extend the domain of to and or as on wikipedia

My remark is not very interesting but in definition 2.3 the distribution function is defined on . So it does not look very clean.

Best Regards

]]>– x[i,j]=x[i+n*(j-1)] instead of x[i+p*(j-1)] on page 10

– instead of on page 44

]]>