Mailman 3 September 2022 - SfePy

ANN: SfePy 2022.3
by Robert Cimrman 29 Sep '22

29 Sep '22

I am pleased to announce the release of SfePy 2022.3. Description ----------- SfePy (simple finite elements in Python) is a software for solving systems of coupled partial differential equations by finite element methods. It is distributed under the new BSD license. Home page: https://sfepy.org Mailing list: https://mail.python.org/mm3/mailman3/lists/sfepy.python.org/ Git (source) repository, issue tracker: https://github.com/sfepy/sfepy Highlights of this release -------------------------- - reorganized scripts - PyPI support - speed-up of linear elastodynamics problems - improved homogenization-based recovery of micro-scale solutions For full release notes see [1]. Cheers, Robert Cimrman [1] http://docs.sfepy.org/doc/release_notes.html#id1 --- Contributors to this release in alphabetical order: Robert Cimrman Yves Delley Vladimir Lukes

1 0

Exerting two or more loading conditions
by kshargh.ali＠gmail.com 19 Sep '22

19 Sep '22

Hi all, To mimic a uniaxial tensile loading, I used "displacement = EssentialBC('displacement', gamma2, {'u.0' : disp_fun})" command before while the other end is fixed. Now, let's say I want to simultaneously stretch it along the y direction (i.e. assume I define another region gamma3 on the top edge). I was wondering how I should modify this code. I tried something like " displacement = EssentialBC('displacement', gamma2, {'u.0' : disp_fun}, gamma3, {'u.1' : disp1_fun})" and different variation but I didn't have any success. Thus, I was wondering if you could please help me out with this. Best, Ali

2 1

PyPI support and reorganized scripts
by Robert Cimrman 12 Sep '22

12 Sep '22

Hi all, the next release (2022.3, due to the end of September) will contain several important user-oriented changes. Originally, the sfepy package was meant to be used mostly in-place in the source tree - the main scripts were directly there. In order to facilitate the installed use as well as to allow using pip, the scripts now reside in sfepy/scripts/ directory and new entry points are installed to the executable path. Short summary (<path> is the path to the sfepy directory relative to the users directory): python3 <path>/simple.py example.py -> sfepy-run example.py python3 <path>/resview.py example.py -> sfepy-view example.py python3 <path>/script/convert_mesh.py -d 3 mesh1.mesh mesh1.vtk -> sfepy-convert -d 3 mesh1.mesh mesh1.vtk python3 <path>/script/blockgen.py -> sfepy-mesh block python3 <path>/script/show_mesh_info.py cylinder.vtk -> sfepy-mesh info cylinder.vtk A pre-release version with the above changes is available on PyPI, you can try it with pip install sfepy (or "python3 -m pip install sfepy") It should install sfepy and all its required dependencies. Some optional solvers are not installed automatically (e.g. petsc4py or scikit-umfpack). The installation can be tested using python3 -c "import sfepy; sfepy.test()" Feedback welcome. Best regards, r.

1 0

sfepy-convert problem
by Christopher Walker 10 Sep '22

10 Sep '22

Hello, I am newbie to sfepy. I am using Windows 10 and python 3.9.1 (I also tried some earlier python versions to see if that helped). I was following the instructions for "Preprocessing: FreeCAD/OpenSCAD + Gmsh" that can be found here: https://sfepy.org/doc-devel/preprocessing.html I got as far as this command: sfepy-convert -d 3 screwdriver_handle.msh screwdriver_handle.vtk but the command was not recognised. However, the command sfepy-run is recognised I was wondering whether this was because I needed to compile the C extension modules (which I had not done). I tried executing the command python setup.py build_ext --inplace from the top level sfepy directory, which, on my system, I believe to be here (at least that is where I found the setp.py file): C:\Users\ChrisWalker\anaconda3\pkgs\sfepy-2022.2-py39h5d4886f_0\Lib\site-packages\sfepy but I get a load of error messages starting with: Appending sfepy.applications configuration to sfepy Ignoring attempt to set 'name' (from 'sfepy' to 'sfepy.applications') Appending sfepy.base configuration to sfepy Ignoring attempt to set 'name' (from 'sfepy' to 'sfepy.base') non-existing path in 'discrete\\common\\extmods': 'fmfield.c' non-existing path in 'discrete\\common\\extmods': 'refmaps.c' etc etc I expect I am making some schoolboy errors here, but I cannot spot it. Is anyone able to help ? Thanks Chris

2 3

Re: About the stiffness matrix
by jhchen 06 Sep '22

06 Sep '22

Hi Robert: I’ve been able to get the stiffness matrix under your help, but I still get several puzzles about the linear system components [1] (8): Is the linear system in the matrix formula: Kx=r ? I exported both K and r to matlab to used the right division to get x: x = r \ K. But this gave me a different result from variables = pb.solve() and x = variables.vec. Analysis showed that pb.solve() gave the appropriate result, what’s the difference between the two approaches? In my project, the measurements (knowns) are Φ on surface (∂Ω) and the unknown is the source distribution Q in attached ProblemDescription.docx. So I need to separate the vectors Φ and Q (nodal values of Φ(x) and Q(x)) explicitly in the matrix form, which means I need to get the matrix F: (FQ is the discretized form of ∫_Ω▒Q(x)s(x)dx in (2), where Q(x) is the source distribution, s(x) is the test function). How can I get this matrix? I found that the solver in SfePy was much faster than the right division in matlab, so I’m curious that which function does SfePy adopts to solve the linear system. Besides, my project requires the inversion of a sub-matrix of the stiffness matrix K, does SfePy offer such methods or which function do you recommend to invert such a banded sparse matrix?

2 1