Jeff Hiller
COMSOL Employee
Please login with a confirmed email address before reporting spam
Posted:
3 years ago
27.09.2021, 10:38 GMT-4
Updated:
3 years ago
27.09.2021, 13:32 GMT-4
Hello Gopinath,
Short answer: No, not particularly, and in fact it could go the other way.
Long answer: Occasionally though rarely, a sequence of steps taken in one version of the GUI to create a model results in a file that converges in that version of the software, while the same sequence of steps taken in the GUI of another version of the software results in a model that does not converge in that other version. When that happens, it can be a reflexion of a change we've introduced at some point between those two versions in what the settings for meshers or solvers do when a particular step in the GUI is chosen. Obviously, we try to minimize how often we introduce such changes and try to make sure they have minimal unintented impact on your work (and we document them in e.g. the Release Notes). But, however small and rare those changes may be, if a model just barely converges in one version, it is possible that it may barely not converge in another; therefore if it happens to one of your models, it invites double-checking how robust the convergence is in the version that does converge (by performing a mesh refinement study, of studying the sensitivity of the model to other model inputs, for instance).
Best,
Jeff
-------------------
Jeff Hiller
Hello Gopinath,
Short answer: No, not particularly, and in fact it could go the other way.
Long answer: Occasionally though rarely, a sequence of steps taken in one version of the GUI to create a model results in a file that converges in that version of the software, while the same sequence of steps taken in the GUI of another version of the software results in a model that does not converge in that other version. When that happens, it can be a reflexion of a change we've introduced at some point between those two versions in what the settings for meshers or solvers do when a particular step in the GUI is chosen. Obviously, we try to minimize how often we introduce such changes and try to make sure they have minimal unintented impact on your work (and we document them in e.g. the Release Notes). But, however small and rare those changes may be, if a model just barely converges in one version, it is possible that it may barely not converge in another; therefore if it happens to one of your models, it invites double-checking how robust the convergence is in the version that does converge (by performing a mesh refinement study, of studying the sensitivity of the model to other model inputs, for instance).
Best,
Jeff