>>13556557Thanks. I'm already out of academia, it's something I'm researching for some type of AI bullshit I'm trying to develop as a startup.
Apparently these processes are called compound Cox processes, since the activator is a Cox process.
However, I can't find anything about defining an integral where the driving jump process is of this type.
I basically just want to know if I can program some Monte Carlo approximation akin to
So that
By an integration scheme similar to compound Poisson processes; I have no idea if this is "definable" or acceptable mathematically, because this may have some weird properties that makes everything diverge and not be defined or whatever.
The jump sizes and number are calculated just as defined above.
>>13559121Yes, that is what I meant.
I'm not a mathematician by training, so what I "know" about measure is just stuff I "learned" while trying to solve exercises in probability and stochastics books, so correct me if I'm wrong on what I'm about to write, it's probably complete bullshit but it will help explain my reasoning.
The process itself has a random intensity measure by the virtue of its intensity measure being a stochastic process itself, no?
TI don't see a way to escape the fact that the intensity measure has to be random, since I cannot just delete or make up jumps after I have already simulated a path of the process, it needs to be defined before I generate lambda because that will also take part in determining how many jumps will occur in the chosen time interval.
I am trying to get an integration scheme, is it not possible? I dn't really care about analytical manipulation. Will I be forced to try and get the characteristic function and then do some inverse FFT ?
