From help-request at octave dot org Thu Feb 16 11:46:01 2006 Subject: Re: rand("state") From: Francesco Potorti` To: Mike Miller CC: Help-Octave List , Paul Kienzle Date: Thu, 16 Feb 2006 18:41:38 +0100 >Thinking about it a little more - the entire pseudorandom sequence has a >length of about 4*10^6001, so if you are making numerous sequences of >length 10^9 say, the probability that two or more are overlapping is >always extremely small (assuming that the starting point in the sequence >is always selected at random from all possible starting points). You just sent a calculation that can be used as the base to compute the probability of having non-overlapping sequences, given the number of sequences, their length, and the generator period, under the assumption of uniformly choosing a random starting point for each sequence. I did not check it, but yes, it can be seen as an instance of a birthday problem. Or else you can see it as this: have a generator of period P, from which you have already taken N non-overlapping sequences of length L. The probability that one more sequence of lenght L starting from a random point does not overlap with one of the already taken N sequences is equal to the probability that no taken sequence starts in an interval of length 2*L, that is, (1-N/P)^(2*L), under the assumption that N*L<