Is my construction almost-surely a conditional Poisson random variable?

35 Views Asked by At

Suppose a continuous random variable $$L \sim F_{L}$$ where $$Pr[L = 0] = 0$$ even though $$L = 0$$ exists for some countable subset of the outcome space.

I am considering the existence of a variable

$$C \sim \text{Poisson}(L)$$

which would not be defined if $L=0$.

We could just condition that $L\neq 0$: $$C\mid L\neq0 \sim \text{Poisson}(L\mid L\neq0).$$

But in practice this doesn't make a lick of difference. Metropolis-Hastings sampling from $$\text{Poisson}(L)$$ works just fine for various choices of $F_L$ due to the almost-never occurrence of $L=0$.

Should I say that $C$ is almost-surely a random variable which follows a conditional Poisson distribution?