PastelBelem8
commented
5 years ago For solving this issue, we propose adding the information of the necessary variables to the Surrogate: variables_indices. By default, the surrogate will assume all variables are being used, which corresponds to the Colon (:) (we opt to provide the Colon default option to avoid having the user to specify the number of variables in the Surrogate constructor).
Advantages of the proposed solution:
- Exploits the current code (evaluation + correction) mechanisms.
- Allows implementing a more global sampling mechanism for all the surrogates in the context of SBO, as well as to maintain the simpler mechanisms(sampling mechanisms associated with each surrogate)
Disadvantages of the proposed solution:
- Another parameter
- (Potentially) harms the code readability/legibility
Due to the way we implemented surrogates, i.e., by associating the sampling routines to each surrogate. We incurred a new problem: when we use these surrogates in the context of SBO (Surrogate Based Optimization), they might be using a smaller subset of variables instead of all the variables, which will yield to a mismatch in the correction, creation and validation routines. Also, the create! routine is poorly conceived as it determines the number of variables based on the number of unscalers, which might now always be the case, i.e., we might sometimes use a smaller subset of unscalers.
This issue is priority and requires all the attention.