Edgar J. Kaiser
Certified Consultant
Please login with a confirmed email address before reporting spam
Posted:
3 years ago
2021年7月28日 GMT-4 16:16
Hi,
this may work with a pointwise constraint.
Another option may be to set up a maximum operator maxop1 in the nonlocal couplings and use maxop1(vanMises) in a global contraint.
Cheers
Edgar
-------------------
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Hi,
this may work with a pointwise constraint.
Another option may be to set up a maximum operator maxop1 in the nonlocal couplings and use maxop1(vanMises) in a global contraint.
Cheers
Edgar
Please login with a confirmed email address before reporting spam
Posted:
3 years ago
2021年7月30日 GMT-4 05:16
Hello Edgar,
Thanks for the reply. I tried both the options you suggested.
I found "point-sum inequality constraint" to be working very nicely in general for all kinds of constraint but solver is not following "Global inequality constraint". I mean the results are not as expected when I used global constraint. I applied it the way you suggested, from non-linear coupling by choosing maximum operator maxop.
Do you have any other sugestions for applying a global constraint?
Hello Edgar,
Thanks for the reply. I tried both the options you suggested.
I found "point-sum inequality constraint" to be working very nicely in general for all kinds of constraint but solver is not following "Global inequality constraint". I mean the results are not as expected when I used global constraint. I applied it the way you suggested, from non-linear coupling by choosing maximum operator maxop.
Do you have any other sugestions for applying a global constraint?
Please login with a confirmed email address before reporting spam
Posted:
3 years ago
2022年1月28日 GMT-5 03:58
maxop is not differentiable, you need to make an integral over the domain of the VM stress to the power p. This makes a continuous p-norm, which is an approximation of the max function for p --> infinity.
maxop is not differentiable, you need to make an integral over the domain of the VM stress to the power p. This makes a continuous p-norm, which is an approximation of the max function for p --> infinity.