On 06/02/10 15:50, osman wrote:
On Tue, 2010-06-01 at 13:04 +0200, Robert Cimrman wrote:
Do you still use numarray instead of numpy?
I have installed both. apt did not complain. I might even need numarray for one of my installed softwares, but can't remember which one. It looks like you are using "import numpy as nm" in stabilized_navier_stokes.py and I did not see that before :-(
It was not there before :-)
Now, I tried running schroedinger.py with
./schroedinger.py --mesh --2d
./schroedinger.py --oscillator
I got: Dimension: 2 sfepy: left over: ['tau', 'n_eigs', 'mesh', 'quadr', 'dim', '_filename', 'base_approx'] schroedinger: reading mesh (/home/osman/src/sfepy/tmp/mesh.vtk)... schroedinger: ...done in 0.57 s schroedinger: setting up domain edges... schroedinger: ...done in 0.14 s schroedinger: creating regions... schroedinger: Omega schroedinger: Surface schroedinger: ...done in 0.24 s schroedinger: equation "rhs": schroedinger: dw_mass_scalar.i1.Omega( v, Psi ) schroedinger: equation "lhs": schroedinger: dw_laplace.i1.Omega( m.val, v, Psi ) + dw_mass_scalar_w.i1.Omega( mat_v.V, v, Psi ) schroedinger: setting up dof connectivities... schroedinger: ...done in 0.00 s schroedinger: describing geometries... schroedinger: ...done in 0.01 s schroedinger: matrix shape: (7920, 7920) schroedinger: assembling matrix graph... schroedinger: ...done in 0.01 s schroedinger: matrix structural nonzeros: 55378 (8.83e-04% fill) schroedinger: updating materials... schroedinger: m schroedinger: mat_v schroedinger: ...done in 0.03 s schroedinger: updating variables... schroedinger: ...done schroedinger: assembling lhs... schroedinger: setting up dof connectivities... schroedinger: ...done in 0.00 s schroedinger: ...done in 0.05 s schroedinger: assembling rhs... schroedinger: setting up dof connectivities... schroedinger: ...done in 0.00 s schroedinger: ...done in 0.01 s schroedinger: computing resonance frequencies... schroedinger: loading... schroedinger: ...done schroedinger: solving... /home/osman/src/sfepy/sfepy/solvers/eigen.py:352: DeprecationWarning: PyArray_FromDimsAndDataAndDescr: use PyArray_NewFromDescr. strategy = conf.strategy ) /home/osman/src/sfepy/sfepy/solvers/eigen.py:352: DeprecationWarning: PyArray_As1D: use PyArray_AsCArray. strategy = conf.strategy ) /home/osman/src/sfepy/sfepy/solvers/eigen.py:352: DeprecationWarning: PyArray_FromDims: use PyArray_SimpleNew. strategy = conf.strategy ) schroedinger: number of converged eigenvalues: 7 schroedinger: ...done in 18.31 s schroedinger: ...done a=100.000000 Energies: n exact FEM error 0: 1.00000000 0.00070729 99.93% 1: 2.00000000 0.00079621 99.96% 2: 2.00000000 0.00158312 99.92% 3: 3.00000000 0.00174114 99.94% 4: 3.00000000 0.00318958 99.89% 5: 3.00000000 0.00327202 99.89% 6: 4.00000000 0.00664995 99.83% schroedinger: solution saved to mesh.vtk
Isn't this a bad result or am I missing something? There are 15K elements near the center I would have expected better convergence.
Yes, it's bad. It should report something like the output at [1] - so there were times it worked quite well. I suspect some problem with the pysparse eigenvalue solver used - maybe some parameters of the solver changed, but I am not sure - git bisect might help. The examples in examples/quantum were somewhat neglected recently - could you create a new Issue?
[1] http://code.google.com/p/sfepy/wiki/OldExamples
hydrogen results are much better but "well" , and "boron" are similar to the "oscillator". "dft" is not running : osman@destiny:~/src/sfepy$ ./schroedinger.py --dft Dimension: 2 sfepy: left over: ['tau', 'n_eigs', 'mesh', 'quadr', 'dim', 'n_electron', 'equations_vh', '_filename', 'base_approx'] schroedinger: reading mesh (tmp/mesh.vtk)... schroedinger: ...done in 0.57 s schroedinger: setting up domain edges... schroedinger: ...done in 0.15 s schroedinger: creating regions... schroedinger: Omega schroedinger: Surface schroedinger: ...done in 0.25 s Traceback (most recent call last): File "./schroedinger.py", line 737, in<module> main() File "./schroedinger.py", line 729, in main app = SchroedingerApp(conf, options, 'schroedinger:') File "./schroedinger.py", line 158, in __init__ init_equations = False ) File "/home/osman/src/sfepy/sfepy/applications/simple_app.py", line 50, in __init__ **kwargs ) File "/home/osman/src/sfepy/sfepy/fem/problemDef.py", line 79, in from_conf obj.set_regions(conf.regions, conf.materials, obj.functions) File "/home/osman/src/sfepy/sfepy/fem/problemDef.py", line 122, in set_regions materials = Materials.from_conf(conf_materials, functions) File "/home/osman/src/sfepy/sfepy/fem/materials.py", line 21, in from_conf mat = Material.from_conf(mc, functions) File "/home/osman/src/sfepy/sfepy/fem/materials.py", line 80, in from_conf function = functions[function] File "/home/osman/src/sfepy/sfepy/base/base.py", line 401, in __getitem__ raise IndexError(msg) IndexError: list.index(x): x not in list osman@destiny:~/src/sfepy$
Even better. I will check it out when I get some more spare time.
Best regards, r.