Mailman 3 May 2010 - SfePy

Trouble with surface traction function
by Logan Sorenson May 14, 2010

May 14, 2010

Hi all, I'm having problems with using a function to define a surface traction load. I attached my modified version of linear_elastic.py and the mesh (block.med). The function is define as follows: materials = { 'solid' : ('Omega', {'lam' : lam, 'mu' : mu}), 'traction' : ('Right', {'val' : [1.0e-6, 0., 0.]}), #'traction' :('Right', None, 'tractionLoad'), } def tractionLoad( ts, coors, region, ig=None, mode=None ): """ts = TimeStepper, coor = coordinates of field nodes in region.""" val = nm.array([[[1.0e-6, 0.0, 0.0]]]) val = nm.tile(val, (coors.shape[0], 1, 1)) print val.shape return {'val' : val} functions = { 'tractionLoad' : (tractionLoad,), } When I use the constant value, everything is fine, but when I switch to the function, I get the following output: <snip> sfepy: updating materials... sfepy: solid sfepy: traction (300, 1, 3) (1331, 1, 3) sfepy: ...done in 0.00 s sfepy: updating variables... sfepy: ...done /usr/lib/python2.5/site-packages/scipy/linsolve/__init__.py:4: DeprecationWarning: scipy.linsolve has moved to scipy.sparse.linalg.dsolve warn('scipy.linsolve has moved to scipy.sparse.linalg.dsolve', DeprecationWarning) sfepy: infs or nans in the residual: [ 0. 0. 0. ..., 0.03059925 0. 0. ] sfepy: False > /home/logan/projects/sfepy/sfepy/solvers/nls.py(197)__call__() 196 --> 197 if self.log is not None: 198 self.log(err, it) Is there a problem with the function definition? I looked at the ConstantFunction class in sfepy/fem/functions.py and I think I'm returning a similarly shaped array for the 'val' key in the returned dictionary in my function. Thanks! Logan -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

2 2

Correction to Windows install instructions
by Logan Sorenson May 14, 2010

May 14, 2010

Hi list, Just noticed a mistake in the Windows install instructions. The lines git svn clone http://svn.scipy.org/svn/scikits/trunk/umfpack umfpack-scikit should be all one line, i.e., git svn clone http://svn.scipy.org/svn/scikits/trunk/umfpack umfpack-scikit Hope it makes more sense. Let us know if you encounter any problems with the procedure. Thanks! Logan -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

2 4

Interesting article on Harmonic functions
by Logan Sorenson May 11, 2010

May 11, 2010

Hi list, I ran into this interesting article on harmonic functions and interpolation [1]. Not sure if it applies to anything in SfePy, but maybe an interesting read nonetheless. Cheers, Logan [1] http://www1.ams.org/samplings/feature-column/fcarc-harmonic -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

1 0

interruption while testing test_input_navier_stokes.py
by JiHun Roh May 11, 2010

May 11, 2010

Hi. I just installed pythonxy 2.6.5.1(the latest version). and additionally installed pyparsing, pysparse, also. When I ran runTests.py, the following error appeared. tests\test_input_navier_stokes.py Can't expand MemType 0: jcol 3341 So I tried to go into debug mode with the command, "python runTests.py --debug tests\test_input_navier_stokes.py" This is the message that I obtained. At the end of this message, python popped up an error message with interrupting its processing. I don't have any idea. What should I do. ----- <<< directory: tests, test files: 1 <<< tests\test_input_navier_stokes.py sfepy: left over: ['__builtins__', '_filename', '__file__', '__package__', 'TestInput', 'input_name' , '__name__', 'output_name', '__doc__'] sfepy: left over: ['utils', 'top_dir', 'cinc', '__package__', '__doc__', '__builtins__', '__file__', '__name__', 'cinc_name', 'verify_incompressibility', '_filename', 'op'] >>> test instance prepared (1 test(s)) ... solving ../examples/navier_stokes/navier_stokes.py... sfepy: reading mesh (c:\Python26\Lib\site-packages\sfepy\meshes\3d \elbow2.mesh)... sfepy: ...done in 0.16 s sfepy: warning: bad element orienation, trying to correct... sfepy: ...corrected sfepy: setting up domain edges... sfepy: ...done in 0.10 s sfepy: setting up domain faces... sfepy: ...done in 0.08 s sfepy: creating regions... sfepy: Omega sfepy: Outlet sfepy: Inlet sfepy: Walls sfepy: ...done in 0.21 s sfepy: equation "balance": sfepy: + div_grad.i2.Omega( fluid.viscosity, v, u ) + convect.i2.Omega( v, u ) - stokes.i1.Omega( v, p ) = 0 sfepy: equation "incompressibility": sfepy: stokes.i1.Omega( u, q ) = 0 sfepy: setting up dof connectivities... sfepy: ...done in 0.01 s sfepy: describing geometries... sfepy: ...done in 0.13 s sfepy: using solvers: nls: newton ls: ls sfepy: matrix shape: (29339, 29339) sfepy: assembling matrix graph... sfepy: ...done in 0.45 s sfepy: matrix structural nonzeros: 904488 (1.05e-03% fill) sfepy: updating materials... sfepy: fluid sfepy: ...done in 0.02 s sfepy: updating variables... sfepy: ...done C:\Python26\lib\site-packages\scipy\linsolve\__init__.py:4: DeprecationWarning: scipy.linsolve has m oved to scipy.sparse.linalg.dsolve warn('scipy.linsolve has moved to scipy.sparse.linalg.dsolve', DeprecationWarning) sfepy: nls: iter: 0, residual: 3.515775e-04 (rel: 1.000000e+00) Can't expand MemType 0: jcol 4042 ----- -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

4 12

ANN: SfePy 2010.2
by Robert Cimrman May 10, 2010

May 10, 2010

I am pleased to announce release 2010.2 of SfePy. Description ----------- SfePy (simple finite elements in Python) is a software for solving systems of coupled partial differential equations by the finite element method. The code is based on NumPy and SciPy packages. It is distributed under the new BSD license. Mailing lists, issue tracking, git repository: http://sfepy.org Home page: http://sfepy.kme.zcu.cz Documentation: http://docs.sfepy.org/doc Highlights of this release -------------------------- - significantly updated documentation - new wiki pages: - SfePy Primer [1] - How to use Salome for generating meshes [2] [1] http://code.google.com/p/sfepy/wiki/Primer [2] http://code.google.com/p/sfepy/wiki/ExampleUsingSalomeWithSfePy Major improvements ------------------ Apart from many bug-fixes, let us mention: - new mesh readers (MED (Salome, PythonOCC), Gambit NEU, UserMeshIO) - mechanics: - ElasticConstants class - conversion formulas for elastic constants - StressTransform class to convert various stress tensors - basic tensor transformations - new examples: - usage of functions to define various parameter - usage of probes - new tests and many new terms For more information on this release, see http://sfepy.googlecode.com/svn/web/releases/2010.2_RELEASE_NOTES.txt (full release notes, rather long). Best regards, Robert Cimrman and Contributors (*) (*) Contributors to this release (alphabetical order): Vladimír Lukeš, Andre Smit, Logan Sorenson, Zuzana Záhorová -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

1 0

2010.2 to be soon
by Robert Cimrman May 10, 2010

May 10, 2010

Hi all, It's over two months from 2010.1, so it's time to prepare a new release. We have made a pretty good progress too (both documentation, wiki pages, and code improvement and bug fixes). Now we need: - make all tests pass (the 2_4 5th order quadrature test fails now) - test all the top level scripts, as described in INSTALL (section Examples) - update INSTALL/README if necessary I am going write the release notes today, and also go through the steps above. My platform is Gentoo Linux (64-bit), so if you have another platform and a spare moments, try the code too, please, and report back any problems. Best regards, r. -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

3 27

Tutorial for using meshes generated with Salome in SfePy
by Logan Sorenson May 7, 2010

May 7, 2010

Hi All, I added a short tutorial on using meshes generated with Salome in SfePy at [1]. Please feel free to edit/comment as you want! Logan [1] http://code.google.com/p/sfepy/wiki/ExampleUsingSalomeWithSfePy P.S., thanks to Andre for the inspiration! -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

3 7

EuroSciPy 2010
by Robert Cimrman May 6, 2010

May 6, 2010

Hi all, I am going to EuroSciPy 2010 [1], where I would like to give a talk about our code. I am attaching the abstract below for comments (the submission deadline is Sunday May 16). My imagination is limited today, so the title is rather lame - enhancements welcome! Anybody is going to be there too? r. [1] https://www.euroscipy.org/conference/euroscipy2010 Title ----- SfePy - Introduction, Examples, and Plans Abstract -------- SfePy (simple finite elements in Python) is a software, distributed under the BSD license, for solving systems of coupled partial differential equations (PDEs) by the finite element method in 2D and 3D. It can be viewed both as black-box PDE solver, and a a Python package which can be used for building custom applications. The primary programming language used for SfePy development is Python while the time-demanding parts of the finite element computations are implemented in C, making thus SfePy sufficiently fast and efficient for our purposes. The code is based mainly on NumPy and SciPy packages, making use of their fast array-based functions and numerical algorithms. The code is multi-platform - it is known to work on flavors of Linux, Mac OS X and Windows. One of the key features from a user's point of view is that the definition of a problem is very similar to the weak formulation of the problem one would use on paper. This definition involves the equations, boundary conditions of various kinds (Dirichlet, Neumann, periodic, general linear combinations of degrees of freedom etc.), finite element approximations, and other necessary ingredients. Equations are built using terms, which correspond directly to the integral forms of weak formulation of a problem to be solved. A user writes this definition into a problem description file, which is a regular Python module. In the black-box usage, however, no real programming is involved, only setting several compulsory keywords. Examples are included with the package that illustrate the basic requirements. On the other hand, the whole power of Python and the available packages is at hand for advanced users. The code includes very simple meshing routines as well as functions to post process results and generate output plots. Alternatively, output from Sfepy can be pulled into visualization packages such as mayavi or paraview. So far, SfePy has been successfully employed for modelling in materials science, including, for example, multiscale biomechanical modelling (bone, muscle tissue with blood perfusion), computation of acoustic transmission coefficients across an interface of arbitrary microstructure geometry, thermoelastic analysis of car disc brakes, and other. In the presentation we will give the overview of SfePy capabilities, as well as some non-trivial examples. -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

2 2

new_core branch
by Robert Cimrman May 6, 2010

May 6, 2010

FYI: the new_core branch (aka "pure Python next generation SfePy" in the long run) is at [2]. See "Initial basic read support for MED file format (salome platform and pythonocc)" thread for background. Several things to notice: - it does not work - some functionality was already back-ported to master (e.g. the reference geometry elements) - it is now rebased w.r.t current master - errors due to e.g. new region selectors by element ids need fix - see (and run) tests/test_domain.py, tests/test_high_level.py to get a feel what was done and what not - only Domain and related classes were updated - the key motivation for all the changes was to remove element groups with the single exception being the (raw) Mesh. By element groups I mean basically the need to use two loops when accessing elements - one with a group id, the other with element id within that group - this complicates things that would otherwise be simple. The cost for the change is raw performance. - there are more things I would like to be back-ported to master soon, namely improvements of Struct, decoupling dof per node of a Variable from field shape, automatic order of Variables. The Plan is: - decouple high level API (see tests/test_high_level.py) from the numerical details of the discretization scheme used (discretization engine) as much as possible/practical - the basic engine would be pure Python: - easy to extend, experiment with - Cython acceleration possible (any Cython experts around?) - additional engines possible, namely femhub support (this surely defines some constraints to be compatible with adaptive discretization schemes) - the Big Question is how to do all this without having to re-implement all the terms [2] for each engine. We can hardly avoid re-implementing all of them for the new API, but that would be only once. - and of course, master branch needs to work/advance all the time. Thoughts/remarks/ideas welcome! r. [1] http://github.com/rc/sfepy/tree/new_core [2] http://docs.sfepy.org/doc-devel/terms_overview.html -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

2 2

Re: distinguish running ipython shell (ipython vs. ipdb)
by Robert Cimrman May 4, 2010

May 4, 2010

Hi Fernando, On 05/04/10 05:04, Fernando Perez wrote: >> -------- Original Message -------- >> Subject: distinguish running ipython shell (ipython vs. ipdb) >> Date: Mon, 03 May 2010 10:50:18 +0200 >> From: Robert Cimrman<cimr...(a)ntc.zcu.cz> >> To: ipytho...(a)scipy.org >> >> Hi all, >> >> recently I have come up with a function that seemed to resolve the problem >> [1]. >> The solution, however, broke our IPython-based interactive sfepy shell >> (isfepy). (That is, in fact, a shameless copy of isympy). >> >> The problem is caused by the fact, that we try to reuse an already running >> ipython shell (e.g. when running isfepy from ipython). But now there is >> always >> a running ipython shell (initialized in the debug() function), that is not >> interactive and just exits. >> >> So my question: is there a way how to distinguish running ipython shell - >> how >> can I tell I am within ipython vs. ipdb? This would help, as we could have >> just >> reinitialized ipython in the latter case. Also, ipdb contains >> >> shell = IPShell(argv=['']) >> def debug(): >> ip = ipapi.get() >> def_colors = ip.options.colors >> Pdb(def_colors).set_trace(sys._getframe().f_back) >> >> so a shell is always started on import. Putting the 'shell = >> IPShell(argv=[''])' into the debug sub-function might help too(?). >> >> It is possible that the way isfepy imports symbols that should be available >> in >> the shell namespace is flawed. Any advice in this respect is more than >> welcome >> - for those interested, see [2], or specifically [3], [4]. > > Those apis in 0.9/0.10 are a bit messy, especially because > unfortunately we never really made the debugger component > embedding-safe. The debugger and the 'real' ipython are fairly mixed > up. I would definitely make sure you defer the initialization to > happen inside your debug() function, that will mitigate the problems. Ok, I did that and it worked well, thanks! I thought that trying ipdb first was a clever idea, as it will reflect automatically IPython API changes, but for now I just removed it - so maybe later. > For the long haul however, I hope when you get a chance you start > playing with the 0.11 series where we have the opportunity to actually > clean things up and build the right embedding apis with a proper > notion of components. I hope to transition the repos to github this > weekend and I think it will be best to make a 'tech preview' release > of 0.11 even if it has known limitations, so we start getting some > feedback on what works and what doesn't. I am forwarding this to our list too, so that isfepy users know what's going on (and try the 0.11). It's great you address those issues in a thourough way. Thanks again, r. >> [1] https://bugs.launchpad.net/bugs/412396 >> [2] http://github.com/rc/sfepy/tree/master >> [3] >> http://github.com/rc/sfepy/blob/d8c5ad8f48c15ecbb93feb5ffab4c834668e352c/is… >> [4] >> http://github.com/rc/sfepy/blob/d8c5ad8f48c15ecbb93feb5ffab4c834668e352c/sf… -- You received this message because you are subscribed to the Google Groups "sfepy-devel" group. To post to this group, send email to sfepy...(a)googlegroups.com. To unsubscribe from this group, send email to sfepy-devel...(a)googlegroups.com. For more options, visit this group at http://groups.google.com/group/sfepy-devel?hl=en.

1 0