EDIT: When running the code locally, I didn't get three nice distinct posterior distributions for all the parameters as the Turing.jl tutorial got. This means that the retrodiction either is very good or pretty bad. I need to probably enforce more random starting points for the samplers.
This PR adds the lynx-hare model, that is, m16.5 from the 2nd edition.
Locally, this code works and my outputs are similar to https://turing.ml/dev/tutorials/10-bayesiandiffeq/#the-lotka-volterra-model. I'll merge the PR as soon as the CI passes.
EDIT: When running the code locally, I didn't get three nice distinct posterior distributions for all the parameters as the Turing.jl tutorial got. This means that the retrodiction either is very good or pretty bad. I need to probably enforce more random starting points for the samplers.