by gwhite » Sun Mar 20, 2022 3:55 pm
Eureka! I finally solved the puzzle. I had to rewrite the code I was using for the Link-Barker data type to get it in the appropriate structure, but finally success.
The trick was to recognize that the code had to be written in terms of lamda, not f directly. Let f’ and phi’ be the parameters not corrected for the length of the time interval (L), and f and phi the parameters corrected for the time interval. So equating 2 expressions for lambda across an interval, f’ + phi’ = (f + phi)^L, with phi’^L = phi. However, there is no easy correction for f’ to f because this is a rate parameter, not a probability parameter. Anyway, given the above relationships, you can solve for f as f’ = (f + phi)^L – phi’. Even understanding the trick, I still couldn’t get the code just right without a lot of torturing, but eventually SUCCESS. New version on my web site where the Pradel data types and the Link-Barker data type give exactly the same results without losses on capture even with time intervals not equal to 1.
Gary