Hierarchical Kriging with vector output

Hi all,
i am trying to set up a hierarchical kriging surrogate model which takes 4 inputs and gives a vector output of size 81.

I built my low fidelity surrogate model called myKriging_LF, from a design of experiment (DOE) using a low fidelity numerical solver.
Now i want to set up the hierarchical kriging model, from a DOE with a high fidelity solver.

The set up for the hierarchical kriging is like this :

mf_KrOptions.Type = ‘Metamodel’ ;
mf_KrOptions.MetaType = ‘Kriging’ ;
mf_KrOptions.ExpDesign.Sampling = ‘User’ ;

mf_KrOptions.ExpDesign.X = myKriging_HF.ExpDesign.X ;
mf_KrOptions.ExpDesign.Y = myKriging_HF.ExpDesign.Y ;
mf_KrOptions.EstimMethod = ‘ML’ ;

mf_KrOptions.Trend.Type = ‘custom’ ;
mf_KrOptions.Trend.CustomF = @(X) uq_evalModel(myKriging_LF,X) ;
mf_KrOptions.Scaling = false ;

Note that the ExpDesign.Y is of size 153 x 81 : i have 153 training points and 81 quantities of interest.
The myKriging_LF kriging model is outputting also the 81 quantities of interest.

When creating the model I get the following error :
> Unable to perform assignment because the size of the left side is 153-by-1 and the size of the right side is 153-by-81.
> Error in uq_Kriging_eval_F (line 34)
> F(:,ii) = f_ii(X);

Would you have any idea how to say to the model that the handle function should return 81 outputs for the trend ?
What am i doing wrong ?

I thank you kindly by advance for any help on this.

Have a nice day !

Hi @elieso,

I noticed that you mentioned

From the error message, it seems that your ExpDesign.X is of dimension 153-by-1 when it should be 153-by-4.
The rest of the code seems to be fine for me. However, I also noticed that you use your LF model with HF experimental design. Right?

Best :grin: