Getting familiar with GPflowOpt

[Simone Lorenzi]

I would like to use GPflowOpt in Dynamic Contrast Enhanced MRI, where I have to fit an experimental curve on a 4-parameters model to extract the four parameters (i.e. f = f(t; a,b,c,d), where t is the time (I know the values of the function in each point that has been acquired) and a,b,c,d are the four parameters).

I was wondering how can I use GPflowOpt to do such analysis and how can I repeat the fitting on more that one curve (there is one curve for each pixel of the MRI image). In other words, I have to extract the four parameters for each pixel. Can you help me with that?

The last question: I read that GPflowOpt supports GPU computation. Does it mean that each optimization is computed on GPU or that I can parallelize the total number of optimizations? Is there a possibility to switch on/off GPU-computing?

Thank you for your attention and congratulations for having developed this useful project!