Inquiry about general algorithm for sensitivity analysis

Please login with a confirmed email address before reporting spam

Hello, I have a question about sensitivity analysis in Comsol. This is not a specific question about a specific simulation problem, but a general question about the calculation algorithm of the calculator embedded in the software.

If you read the introduction in the "Theory for the sensitivity interface" section of the reference manual for Comsol 5.5, it says that it is possible to calculate sensitivity for any objective function. It is surprising that sensitivity calculation is possible for "any objective function." I wonder how this is possible. Since Comsol is a physics simulation software, I thought that it is possible to calculate sensitivity only for objective functions that have physical meaning. Please explain this. Also, if there is a related paper, please recommend it.

In addition, even when the objective function is a physical scalar quantity, the calculation of terms such as round(Q)/round(u), round(L)/round(zeta), which are partial derivatives with respect to the design variables, seem to be determined by the objective function, the governing equation of the physical phenomenon L, and the design variable zeta. Since these are parts that are arbitrarily set by the user, I wonder if it is impossible to calculate them using general algorithms. I am curious about how general algorithms can be used for these calculations. Please explain this. Also, if there is a related paper, please recommend it.

Thank you in advance.


Reply

Please read the discussion forum rules before posting.

Please log in to post a reply.

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.