Re: Recent post on stan-devel

[Bob Carpenter]

Thanks. That makes sense to me. It's what we already do with
"print", I think.

R isn't the only platform we're looking at, by the way. We
are using Andrew's wishes for R as a template for what we'll also
do in CmdStan and in PyStan. I don't think either CmdStan or PyStan
have well evolved notational conventions.

I'm guessing it'll bug Andrew because it's camel case ("logLik" instead
of "loglik" or "log_lik" or "log.lik"), and he wanted to keep our
interface consistent. It's hard to conform to all the conventions
going when they're inconsistent!

- Bob

On Jun 27, 2014, at 7:05 PM, Avraham Adler <avraha...@gmail.com> wrote:

> Good afternoon.
>
> Sorry, I cannot post to that list, but in reference to this post, it may be a reasonable idea to consider looking at naming it `logLik` and having a method, so that it more closely dovetails with what is in`R` already. Please see "Extract Log-Likelihood" from base R (stats package). Also, it would need to be very clear as to on which scale the IC are being returned: the standard (2*NLL + 2*elpd) or the one used in the waic paper for loo, which is not doubled.
>
> I copied "d...@mc-stan.org." hoping that Bob will see this too.
>
> Thanks,
>
> Avi

[Andrew Gelman]

Yeah, R is the worst, with a mix of camelcase, dots, singulars, plurals, etc. I’d prefer underbars to keep consistent within Stan…