Mailman 3 January 2012 - SfePy

Compilation error building sfepy
by Tom Aldcroft Jan. 18, 2012

Jan. 18, 2012

Hi - I'm trying to build SfePy on a CentOS-5 x86_64 machine. I'm using Python 2.7 which has the mandatory dependencies stated in the INSTALL file. I don't yet have all the optional dependencies but thought I would try to just get started with the simplest build strategy. The compiler is gcc 4.1.2. A full log of the output from "python setup.py build_ext --inplace" is in the following paste: http://paste.pocoo.org/show/536250/ The error is: ... compiling C sources C compiler: gcc -pthread -fno-strict-aliasing -fPIC -g -O2 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC compile options: '-D__SDIR__='"sfepy/fem/extmods"' -DSFEPY_PLATFORM=0 -D__SDIR__='"sfepy/fem/extmods"' -DSFEPY_PLATFORM=0 -Isfepy/fem/extmods -I/proj/sot/ska/dev/arch/x86_64-linux_CentOS-5/lib/python2.7/site-packages/numpy/core/include -I/proj/sot/ska/dev/arch/x86_64-linux_CentOS-5/include/python2.7 -c' extra options: '-g -O2' gcc: sfepy/fem/extmods/mappings.c sfepy/fem/extmods/mappings.c: In function \u2018__pyx_pf_5sfepy_3fem_7extmods_8mappings_14CVolumeMapping___cinit__\u2019: sfepy/fem/extmods/mappings.c:1116: error: incompatible types in assignment sfepy/fem/extmods/mappings.c:1190: error: incompatible types in assignment sfepy/fem/extmods/mappings.c:1264: error: incompatible types in assignment sfepy/fem/extmods/mappings.c: In function \u2018__pyx_pf_5sfepy_3fem_7extmods_8mappings_15CSurfaceMapping___cinit__\u2019: ... <AND MORE LIKE THIS> Any ideas on the source of this problem? I thought about trying to build on a more modern OS, but eventually I'll need to run this on CentOS-5. Thanks, Tom

4 3

linearization of results for visualization
by Robert Cimrman Jan. 17, 2012

Jan. 17, 2012

Hi, I have just merged in a branch implementing linearization of higher order solutions for the purposes of visualization. An example what it does can be found at [1]. The implementation was inspired by [2]. Check also tests/test_linearizer.py to see how it can be used, and its output for additional results. This development is just a preparation for implementing a hierarchical FE basis, as the current Lagrange basis is not really usable for polynomial orders >= 3, due to its bad conditioning properties. It's good to be able to see what one is implementing :) The current implementation is pure Python (mostly NumPy + one Python loop I could not avoid), so it's not very fast. However, the real bottleneck is saving the resulting (large) VTK file - we should think about how to speed up the MeshIO classes. A ready made solution with compatible license would be the best... r. [1] https://github.com/sfepy/sfepy/wiki/Linearizer-Examples [2] http://onlinelibrary.wiley.com/doi/10.1002/nme.1787/abstract

1 0

SfePy cmp Abaqus
by freevryheid Jan. 17, 2012

Jan. 17, 2012

I've uploaded an Abaqus report showing displacements, stresses and strains calculated at the nodes (extrapolation averaged). To compare I've included SfePy pb and mesh, that also generates a csv report. The stress/strain results differ - which is correct? Note I also calculate von mises (vms) that required the patch submitted earlier. a

5 13

gmsh mesh error
by Md. Golam Rashed Jan. 13, 2012

Jan. 13, 2012

what is the problem with this mesh?? its the primer mesh, except that it is the full section. and is this right in selecting the top and bottom point (center is at 0,0) - regions = { 'Omega' : ('all', {}), 'Bottom' : ('nodes in (x <= 0) & (y <= -75)', {}), 'Top' : ('nodes in (x <= 0) & (y <= 75)', {}), }

2 8

wilson.homeunix.com
by Andre Smit Jan. 12, 2012

Jan. 12, 2012

Anyone else getting the following message when posting to sfepy-devel. Just started about a week back or so. This is an automatically generated message from wilson.homeunix.com. Please don't reply to it. A technical issue was detected in your Email which cannot be repaired automatically. Your mail was delivered to the destination address nevertheless. For your convenience, the original email is attached. Thank you. A desciption of the issue(s) follows: ------------------------------ -------------------------------------- Your email contains HTML, which is marked with higher spam probablility by widely-deployed anti-spam programs, and which can cause display problems with various email clients. Please switch to "text messages" in your emails client. -- Andre

3 3

[PATCH] Add 2d von mises stress calcs
by Andre Smit Jan. 12, 2012

Jan. 12, 2012

--- sfepy/mechanics/tensors.py | 48 ++++++++++++++++++++++++++++--------------- 1 files changed, 31 insertions(+), 17 deletions(-) diff --git a/sfepy/mechanics/tensors.py b/sfepy/mechanics/tensors.py index 9aee18d..4974275 100644 --- a/sfepy/mechanics/tensors.py +++ b/sfepy/mechanics/tensors.py @@ -105,26 +105,40 @@ def get_von_mises_stress(stress, sym_storage=True): if sym_storage: dim = sym2dim(dim) - assert_(dim == 3) + if dim == 2: - if sym_storage: - s11 = stress[:,0] - s22 = stress[:,1] - s33 = stress[:,2] - s12 = stress[:,3] - s13 = stress[:,4] - s23 = stress[:,5] + if sym_storage: + s11 = stress[:,0] + s22 = stress[:,1] + s12 = stress[:,2] + + else: + s11 = stress[:,0,0] + s22 = stress[:,1,1] + s12 = stress[:,0,1] + vms = nm.sqrt(s11**2 - s11*s22 + s22**2 + 3*s12**2)[:,None] + else: - s11 = stress[:,0,0] - s22 = stress[:,1,1] - s33 = stress[:,2,2] - s12 = stress[:,0,1] - s13 = stress[:,0,2] - s23 = stress[:,1,2] - - vms = nm.sqrt(0.5 * ((s11 - s22)**2 + (s22 - s33)**2 + (s11 - s33)**2 - + 6.0 * (s12**2 + s13**2 + s23**2)))[:,None] + + if sym_storage: + s11 = stress[:,0] + s22 = stress[:,1] + s33 = stress[:,2] + s12 = stress[:,3] + s13 = stress[:,4] + s23 = stress[:,5] + + else: + s11 = stress[:,0,0] + s22 = stress[:,1,1] + s33 = stress[:,2,2] + s12 = stress[:,0,1] + s13 = stress[:,0,2] + s23 = stress[:,1,2] + + vms = nm.sqrt(0.5 * ((s11 - s22)**2 + (s22 - s33)**2 + (s11 - s33)**2 + + 6.0 * (s12**2 + s13**2 + s23**2)))[:,None] return vms -- 1.7.8.3

2 1

attach files
by freevryheid Jan. 12, 2012

Jan. 12, 2012

Have file attachments been disabled in this group?

2 2

gambit neutral file format
by nil...＠googlemail.com Jan. 11, 2012

Jan. 11, 2012

A description of the gambit neutral file format is available. See www.stanford.edu/class/me469b/handouts/gambit_write.pdf

2 1

mesh conversion gambit to Nastran
by nil...＠googlemail.com Jan. 11, 2012

Jan. 11, 2012

I would like to convert a mesh from gambit to Nastran. How can I do that with sfepy ? A small example would be appreciated. Thanks in advance. Nils

5 18

Getting displacements into postprocess data
by Torsten Bronger Jan. 11, 2012

Jan. 11, 2012

Hallöchen! I calculate the stress in my material with a postprocessor, i.e. def stress_strain(out, pb, state, extend=False): ev = pb.evaluate stress = ev("de_cauchy_stress.2.Omega(silicon.D, u)", mode="el_avg") out["cauchy_stress"] = Struct(name="output_data", mode="cell", data=stress, dofs=None) return out pb.save_state("linear_elasticity.vtk", vec, post_process_hook=stress_strain) view = Viewer("linear_elasticity.vtk") view(vector_mode="warp_norm", rel_scaling=1) This way, I get two images in the Mayavi window: the displacements and the stress. So far, so good. But how do I get the colouring from the stress into the distorted mesh? Currently, the stress visualisation uses the undistorted mesh. And the distorted (aka displaced) wireframe shows colours representing the displacement. I'd like to combine both: I'd like to have the node positions of the left-hand picture (displacements) and the colours of the right-hand picture (stress). How is this possible? Thank you! Tschö, Torsten. -- Torsten Bronger Jabber ID: torsten...(a)jabber.rwth-aachen.de or http://bronger-jmp.appspot.com

3 18