Hello
I am am using FiPy with Python 3.4 - I think the conversion went
well enough but I do have some questions.
I have successfully implemented a 1-dim concentration dependent
diffusion problem in fipy, where success was defined by reproducing
what other researchers had produced in a publication.
My diffusion coefficient looks like
Dx=Dpe_x*(a_x+((1.-a_x)/(b_x*C+g_x)))
where C is a CellVariable on a 1 dim grid and the other terms are
constants.
As mentioned this all worked well. However when I went to extend to
2 Dimension and include second diffusion coeff, Dz (C) , for
anisotropic diffusion, I am running into some issues.
first for constant diffusion coeff of Dx =1, Dz = 5 on 2 dim
rectangular grid (Grid2D)
When written as
TransientTerm() ==DiffusionTerm([[[1.,0],[0,5.]]])
the results behave like I would expect ( on a 2dim rectangular mesh)
But when written like
TransientTerm() == DiffusionTerm([[1.,5.]]), which I think the
manual said was equivalent ( under freq asked questions)
I get an assertion error, which I can post but before posting should
these two diffusion terms be equivalent?
Sticking with the complete matrix form of the diffusion Coefficient
for Dx and Dz ( both similar functions of the cellVariable C)
when I use this form
TransientTerm() == DiffusionTerm([[[Dx,0],[0,Dz]]])
I get the following traceback error
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-101-745650bc0cee> in<module>() 24
Dz=Dpe_z*(a_z+((1.-a_z)/(b_z*C+g_z)))
25
---> 26eqX = TransientTerm() ==
DiffusionTerm([[[Dx,0],[0,Dz]]])C:\Users\PhilB\Anaconda3\lib\site-packages\fipy-unknown-py3.4.egg\fipy\terms\abstractDiffusionTerm.py
in __init__(self, coeff, var) 62 from fipy.variables.variable import
Variable63 if not isinstance(self.nthCoeff, Variable):--->
64self.nthCoeff = Variable(value = self.nthCoeff)65 66 from
fipy.variables.cellVariable import
CellVariableC:\Users\PhilB\Anaconda3\lib\site-packages\fipy-unknown-py3.4.egg\fipy\variables\variable.py
in __init__(self, value, unit, array, name, cached) 109 unit = None110
--> 111self._setValueInternal(value=value, unit=unit, array=array)112
113 self._name =
nameC:\Users\PhilB\Anaconda3\lib\site-packages\fipy-unknown-py3.4.egg\fipy\variables\variable.py
in _setValueInternal(self, value, unit, array) 638 639 def
_setValueInternal(self, value, unit=None, array=None):--> 640self._value
= self._makeValue(value=value, unit=unit, array=array)641 642 def
_makeValue(self, value, unit=None,
array=None):C:\Users\PhilB\Anaconda3\lib\site-packages\fipy-unknown-py3.4.egg\fipy\variables\variable.py
in _makeValue(self, value, unit, array) 668 669 if unit is not None or
type(value) in [type(''), type(()), type([])]:--> 670value =
PF(value=value, unit=unit, array=array)671 elif array is not None:672
array[:] =
valueC:\Users\PhilB\Anaconda3\lib\site-packages\fipy-unknown-py3.4.egg\fipy\tools\dimensions\physicalField.py
in __init__(self, value, unit, array) 203 204 if type(value) in
[type([]),type< span class="ansiyellow" style="box-sizing: border-box;
color: rgb(196, 160, 0);">(())]:--> 205value = [PhysicalField(item,unit)
for item in value]206 if unit is None:207 unit =
value[0].unitC:\Users\PhilB\Anaconda3\lib\site-packages\fipy-unknown-py3.4.egg\fipy\tools\dimensions\physicalField.py
in <listcomp>(.0) 203 204 if type(value) in [type([]),type< span
class="ansiyellow" style="box-sizing: border-box; color: rgb(196, 160,
0);">(())]:--> 205value = [PhysicalField(item,unit) for item in
value]206 if unit is None:207 unit =
value[0].unitC:\Users\PhilB\Anaconda3\lib\site-packages\fipy-unknown-py3.4.egg\fipy\tools\dimensions\physicalField.py
in __init__(self, value, unit, array) 212 else:213 normalized +=
[item.inUnitsOf(unit).value]--> 214value = numerix.array(normalized)215
216 if unit is None:ValueError: setting an array element with a sequence.
Is there a more appropriate way of writing the 2Dim anisotropic
concentration dependent diff coeff?
Note if I use isotropic 2 dim concentration dependent diffusion the
results look correct ( in 2D)
TransientTerm() == DiffusionTerm(Dx)
Let me know if this challenge makes sense and if there is a solution
thanks
Phil
ps
I see questions posted on this list, Gist and stack overflow - is there
a preference for posting ?
_______________________________________________
fipy mailing list
[email protected]
http://www.ctcms.nist.gov/fipy
[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
