Thank you for the precisions. Well in my case I will never have a gradient so I have to make do with what I have 
. I’ll probably end up having more parameters too so we’ll see. I saw this other thread with someone sampling 200 parameters and using Metropolis. Hopefully I’ll never have that many…
When you say almost our only hope, what other hopes are there? SMC?
And for the PR, I don’t think I’ll be able to do it as of now, but maybe in the future when I understand more.