Does that mean I accept the draw if $log(\pi(x')) \geq log(\pi(x))$ ?

I'm aware of the non-normality and that the points should not lie on the line itself. I was rather wondering why the line is not 45°.

Thank you very much for your help! I might switch to normplot() instead to avoid confusion. Just to clarify, the qqplot shows that the sample has fatter tails than generated by a normal, doesn't it? EDIT: I found this in the qqplot function: "A reference line passing through the first and third quartiles is helpful for judging whether the points are linear."

For me it looks more like it's an OLS fit. But unfortunately one cannot set any relevant parameters in the qqplot().

Is there any way to obtain a 45° line as reference line? I thought this could be achieved by standardizing.

Thank you! I thought the red line is constructed so that you can see the deviations from the standard normal quantiles? When I use standardized data, the line should be y=x then, shouldn't it?

Thank you for the illustration! I kind of solved the problem with the negative mean. In a previous publication, the authors decided to block two variables together which led to complicated expressions for the mean. I "unblocked" these variables and get perfectly behaving means now. But thanks to you all for your helpful comments!

Wow, tbh I didn't consider it that way :D I am using the truncated normal as a posterior distribution, which requires a value larger than 0. Would you say it's reasonable to set it equal to the lower boundary (which is 0) in cases where the sampling method fails?

Sorry, I totally forgot about that post. My bad! I guess I will try the accept-reject method then. I was silently hoping that there exists some kind of library already. For some reason I cannot compile the one by Chopin.

I chose the proposal such that I get a rejection rate of roughly 40% per iteration.

Sorry to dig out this old post. But isn't the prior value (generated from the prior distribution) supposed to be irrelevant as long as there is a sufficient number of iterations?