I have a problem concerning the order of the evaluation of the AceFEM parameters. My AceGen Element contains, among many others, an expression of the form

B (aa - bb/(cc Exp[dd w^2] + ff))

with B, aa, bb, cc, dd and ff being parameters and w being a degree of freedom with two entries per node. If I choose the wrong combination of values for the parameters, my problem diverges immediately even if the parameter B is equal to zero, which, in my opinion, makes the whole expression equal to zero. I already tried to set the term to zero "manually", by multiplying it directly (in the element) with zero and this does work.

So my question is, does AceFEM tries to evaluate the "inner" expression before multiplying it with the prefactor? This would explain the divergence.

Thank you in advance and best regards, Julia


1 Answer 1


For more in detail answer, can you provide some minimum working example of AceGen code and AceFEM output, values used etc? Otherwise I can only guess. Is the newton iteration output 10^50? Because that means some numbers in your element Residual/Tangents are likely indeterminate. If yes, how do the Local tangent and residual look like and their eigenvalues? If you multiply an expression with 0 in the element, then 0 will be directly exported so the "0 (aa - bb/(cc Exp[dd w^2] + ff))" is 0, otherwise, for example in the code BB*(something/0) is indeterminate even if BB==0. Can be that your parameters give some badly conditioned tangent matrix.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.