Many players in the field of engineering simulation software are investing on digital twinning and reduced order modeling techniques, meaning that the field bears potential.

I was wondering if among the various subflavours - POD, reduced basis... - there is one in particular that is spreading more than the others, or that on paper sounds more promising, and understand the reasons behind this predominance (possibly with an eye to diffusion in industry).

I am especially interested in structural/mechanical engineering, but hearing about other areas would be interesting as well.

  • 1
    $\begingroup$ POD-ROMs have been used very successfully for prediction and control on diffusion-dominated problems. Most fixed-basis methods have a lot of trouble representing transport phenomena. $\endgroup$
    – whpowell96
    Commented Nov 6, 2023 at 19:40
  • 3
    $\begingroup$ Everyone has their own favorite method. This seems difficult to answer without getting into opinion-based stuff. $\endgroup$ Commented Nov 7, 2023 at 4:04
  • 3
    $\begingroup$ @WolfgangBangerth I agree that the answer to this question is highly opinion-based, but it would still be interesting to hear some answers. The question also mentions the application to industrial settings where sometimes other methods are used, e.g. academia tends to prefer intrusive ROMs (POD, RBM, PGD) due to its accuracy, whereas industry sometimes rather uses non-intrusive ROMs (POD+NN, Operator Inference) since it is easier to implement and does not require access to the system matrix. $\endgroup$ Commented Nov 8, 2023 at 17:12
  • 1
    $\begingroup$ I agree that the question and implications are interesting, but it may be to broad for this forum. Any reasonably complete answer to this question would boil down to providing a list of survey papers or an even longer list of standard papers unless more specificity is given. $\endgroup$
    – whpowell96
    Commented Nov 9, 2023 at 15:20


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.