Andrew Johnson

Closed by #2822

> Should we be keeping the old signatures and just adding new ones? I thought that's what I was doing here (unless I'm misunderstanding). The old signatures are still present,...

> lgtm once the upstream tests pass Adding the early return with `fx = f(x);` actually help catch a bug with the existing hessian implementations. The other `hessian` functions already...

Also for information, one suggestion for these cases (credit to @alexmcqueen) was to use different outline vs fill colours for the points in question: 

This is still an issue with vscode 1.88 and both release and pre-release remote-ssh extension (macos arm64 host, windows arm64 remote). Downgrading to vscode v1.85 is the only thing that...

@WardBrian would you mind linking to where in the `cmdstanpy` source this extraction/conversion for initial values happens? I'm not at all familiar with the source to find it myself sorry!

Ahh that's really helpful thanks. So the behaviour of passing the results as inits: ``` inits=optimum.stan_variables() ``` This is only possible for `CmdStanMLE` and `CmdStanVB` objects, where there is only...

The implementation you linked is using the draws from the last iteration of sampling as the initial values for a new sampling run. The `cmdstanpy` implementation is just a consequence...

I was after a way of getting the dimensions of the unconstrained parameters, and `get_unconstrained_sizedtypes()` looks like the only option

Another option could be to parse the output of `unconstrained_param_names()`, but also open to other suggestions if there's something obvious I'm missing!