Gaussian Process Predictive Posterior Function

[Bob_Carpenter]

@avehtari

It’s possible that in order to keep the memory use of autodiff in control for the birthday problem we really need to present covariance functions as functions and call covariance matrix generation with that function, that is, we provide on function argument as for ODE solvers and algebraic solver.

We could do that, but I don’t see how it helps with memory.

@avehtari

Data is available and the source is also mentioned in BDA3.

Found it in the bibliographic note (helps to have the PDF!): http://www.mechanicalkern.com/static/birthdates-1968-1988.csv