Dear all,
I am a new user of sfepy that was attracted to this FE code because of the homogenization engine.
I am trying to solve a microproblem using the homogenization engine of sfepy. In my case the microsopic variable is not a vector -like $u^1_i$- but a tensor $\omega^1_{ij}$.
As I understand sfepy provides the CorrDimDim class for storing the resulting microfield for the differnt unitary loads (pis). This class works fine if the unknown is a vector, but in my case it is a second order tensor how can I implement a Corr2Dim2Dim class or even more general CorrNDimMDim class for storing the values of more general fields?
Thank you in advance for your help,
Hi Alexandros,
(sorry for late reply, the message got buried...)
On 18/10/2024 11:21, Alexandros STATHAS via SfePy wrote:
Dear all,
I am a new user of sfepy that was attracted to this FE code because of the homogenization engine.
I am trying to solve a microproblem using the homogenization engine of sfepy. In my case the microsopic variable is not a vector -like $u^1_i$- but a tensor $\omega^1_{ij}$.
As I understand sfepy provides the CorrDimDim class for storing the resulting microfield for the differnt unitary loads (pis). This class works fine if the unknown is a vector, but in my case it is a second order tensor how can I implement a Corr2Dim2Dim class or even more general CorrNDimMDim class for storing the values of more general fields?
Maybe you do not need to implement a new class - what about using the generic CorrNN, and map your two dimensions to a single dimension? (something like I = dim * i + j).
Otherwise have a look at CorrNN (and CoefMN) in sfepy/homogenization/coefs_base.py - there is no user-facing documentation of the internals besides the code.
Cheers, r.
participants (2)
-
Alexandros STATHAS -
Robert Cimrman