how to simultaneously optimize GP and external function

[pilchat]

Hi,

I am quite new to GPy, so please forgive my newbyness. I have been checking the tutorials but I could find only little help on what I say below.

I would like to fit a time series with a mean_function that I define. Is it possible to do it with the .optimize() method (or any built-in MC sampler)?

If not, I can define a function that first compute the mean_function and then computes the GP on the corresponding residuals. The problem with this approach is that it takes quite a long time for the GP model to recompute the likelihood at every update of the parameters (I have the feeling that each update is slower then the steps of .optimize(), but I haven't verified it yet).

Do you have any suggestion?

Thanks a lot