You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have read both the 2.0 and the original release papers looking for a way of building a multifidelity model in which the design space of the low fidelity model has less dimensions than the high fidelity space. This does not seem to be possible with the current Co-Kriging model. The new Hierarchical Design Spaces, which can tackle variable design spaces, seem to be designed only for single-fidelity models. Across the literature, I have only found the technique on design space mappings to deal with this case.
I can think of a very naive way of dealing with this, which is expanding the low-fid input vectors onto the high-fid space forcing the low fid points to be constant along the new dimensions, but I was wondering if I could find a better solution within the toolbox.
An aerospace example of my case would be a low fid surrogate of lift and drag coefficients that only depends on the angle of attack, and a high fid surrogate which accounts in addition for Reynolds and Mach.
Is there any tool in SMT for my issue? If there's not, is it in the roadmap?
Thanks again for your work!
The text was updated successfully, but these errors were encountered:
Hi, thanks for such an amazing toolbox!
I have read both the 2.0 and the original release papers looking for a way of building a multifidelity model in which the design space of the low fidelity model has less dimensions than the high fidelity space. This does not seem to be possible with the current Co-Kriging model. The new Hierarchical Design Spaces, which can tackle variable design spaces, seem to be designed only for single-fidelity models. Across the literature, I have only found the technique on design space mappings to deal with this case.
I can think of a very naive way of dealing with this, which is expanding the low-fid input vectors onto the high-fid space forcing the low fid points to be constant along the new dimensions, but I was wondering if I could find a better solution within the toolbox.
An aerospace example of my case would be a low fid surrogate of lift and drag coefficients that only depends on the angle of attack, and a high fid surrogate which accounts in addition for Reynolds and Mach.
Is there any tool in SMT for my issue? If there's not, is it in the roadmap?
Thanks again for your work!
The text was updated successfully, but these errors were encountered: