Finding and examinating the distribution of portfolio [mostly Maple question]

35 Views Asked by At

So that's more Maple related question but I'll give it a try. Assume we have portfolio $$S_N=X_1+\dots+X_N$$ Where $X\sim Poi(5)$ and $N\sim Bin(60,1/2)$. We want to study the distribution of $S_N$. One way to do this is to use the PGF of $X$ and $N$. Then the PGF of $S_N$ is given by $$P_{S_N}(t)=P_N(P_X(t))$$ And to get the distribution of $S_N$ we just need to expand the $P_{S_N}(t)$ in Taylor series. Now - since I'm new to using Maple I struggled with this a bit but eventually end up with a decent plot of this distribution.

My next problem is to generate a sample of size $50$ from that distribution. I've read the documentation of $Sample$ function but it didn't help at all. In $R$ for example the task is trvial since I have the array of the probabilities from the Taylor expansion I just type $Sample(Array,50)$ and that's it. Is there an analogous way to do that in Maple?

1

There are 1 best solutions below

0
On

By integrating out the binomial random variable in the joint distribution, we obtain the marginal of the sum: $$P(S_N=m)=P\bigg(\sum_{n=1}^NX_n=m\bigg)=\sum_{k=0}^MP\bigg(\sum_{n=1}^kX_n=m\bigg|N=k\bigg)P(N=k)$$ The sum of $k$ Poisson distributed , independent random variables with same parameter $\theta$ is Poisson with parameter $\theta k$. So $$P(S_N=m)=\sum_{k=0}^M\frac{(\theta k)^me^{-\theta k}}{m!}P(N=k)=\sum_{k=0}^M\frac{(\theta k)^me^{-\theta k}}{m!}\binom{M}{k}p^k(1-p)^{M-k}$$ To generate a sample from this one, generate a binomial RV $k$ and then generate $k$ Poisson distributed RVs and then sum them.