I am interested in discussing a question related to RNNPP here.

As we know, RNNPP is used to generate pseudorandom numbers from a nonhomogeneous Poisson process (NHPP). It is stated clearly that the algorithm is nonhomogeneous Poission process thinning method.

I am quite intersted in the related example 1, where the rate lamda(t) = 0.6342*exp(0.001427*t). Actually it is easier to consider this example, as the coefficient before t is positive, i.e. 0.001427.

My question is "If this coefficient is negative, i.e. -0.001427, can we still use this algorithm to simulation NHPP model?".

As far as I am concerned, Lewis did not differetiate these two conditions with respect to thinning algorithm. He did mention it in the gap statistics method. I am very happy to know the anwer of my question.

Thanks a lot!