Question: Nested DOE for MFK

[jtneedels]

Hi,

I know a requirement of this AR1 MFK implementation is a nested DOE, where higher fidelity data is included in the lower fidelity dataset.

I was checking the MFK function and noticed it will accept non-nested datasets without throwing an error, but of course results in an incorrect procedure. I was curious - what is occurring, in terms of this coded implementation, when non-nested datasets are used? And is there a reason for not having an error when non-nested inputs are detected? I saw in Le Gratiet's thesis he describes a version of the algorithm not requiring nesting, but I need to study it further to understand the differences.

The SMT package is great, I've used it for MFK in my PhD (citing of course!) and I was curious if there are any ways to get more involved with developing/contributing to the code base?

Thank you so much!

[relf]

Hi, sorry for the late reply. No reason to not check nested inputs for MFK only it is not implemented (PR welcomed!). Speaking of contributing you can refer to the documentation on the topic.
Notice in the latest release 2.8, the addition of the new multi-fi MFCK model which does not require nested inputs.