Damon McDougall

Thanks for the explanation. It's a behaviour change but the behaviour right now doesn't map to the mathematics. I'm trying to convince myself it's an API change that isn't too...

This looks useful: http://dlib.net/optimization.html#find_min_box_constrained

It seems returning `INFINITY` for a point requested outside of the domain makes `gsl` give up and stop at the previous good point. This seems to be 'good enough' for...

Right now, to use the surrogate functionality in a likelihood class (or one of the specialised ones) the user has to do all of [this boilerplate](https://github.com/libqueso/queso/pull/351/files#diff-fe90a025468420a7fba4b10cb8a30344R95) in their likelihood's constructor...

Also this would be perfect for derivative calculations. All of that can be done automatically because the derivative and Hessian calculations are all analytical.

Actually, it'll be d (\mathbb{Q}) / d (\theta). This specialises when \mathbb{Q} = z.

@nicholasmalaya http://homepages.warwick.ac.uk/~masdr/JOURNALPUBS/stuart103.pdf

@briadam and I remember seeing some older pieces of code that check for 'disabled parameters' on the MCMC side. Let's see if we can revive some of that infrastructure to...

And I'll add an example too so I can point people to code.

Is there a `testmod.mod` file anywhere? Does the file actually exist? Does it exist and have the incorrect spelling or case?