What are the major benefits of using turing over DifferentialEvolutionMCMC Package?

I have this question because whenever I look in forum question and issues its always related to Turing and rarely to DifferentialEvolutionMCMC which seems like people are mostly using Turing over DifferentialEvolutionMCMC. Can someone explain?

What is this DifferentialEvolutionMCMC package you’re talking about?

Here this one : GitHub - itsdfish/DifferentialEvolutionMCMC.jl: A Julia package for Differential Evolution MCMC

I have used same in r but here in Julia I was facing some issues and for that reason I reached out.

Turing is a general purpose ppl, so you can use it for just about any Bayesian problem. It’s also very well known in this community.

There package you linked looks like a one-person effort and hasn’t been updated in a year. I’ve never heard of it before, and I suspect many others haven’t either, which is why they wouldn’t bring it up.

If the package does what you need, then great! It’s often nice to use something pre built rather than implementing from scratch. If you only need minor things, consider opening issues (or better yet PRs!). Package authors usually like engagement from users.

1 Like

I have no issue in using Turing. As I was familiar with DifferentialEvolution in R so I thought it will be better to try that first. I have one main problem i.e. defining multiplicative error in likelihood (let it be either Turing or DE MCMC ) but I think I should post a fresh issue for that or is it ok to ask here? I had posted it before but got no replies.