Hello Jonathan,

It turns out that my spatial discretization was too fine, so Python ran out
of memory. After decreasing my spatial discretization and trying the
PowerLawConvectionTerm that you suggested, I still get the "IndexError: id1
does not have the same size as b" error.

Best,
Edwin


On Thu, Mar 21, 2013 at 11:59 PM, Edwin Sze Lun Khoo <[email protected]>wrote:

> Hello Jonathan,
>
> I tried both PowerLawConvectionTerm(coeff=([[-alpha]]/C**2, ), var=C) and
> PowerLawConvectionTerm(coeff=([[-alpha]]/C.faceValue**2, ), var=C) and
> Python froze without giving any error messages. I am not sure what happened.
>
> Best,
> Edwin
>
>
> On Mon, Mar 18, 2013 at 8:38 AM, Jonathan Guyer <[email protected]> wrote:
>
>>
>> On Mar 15, 2013, at 5:08 PM, Edwin Sze Lun Khoo wrote:
>>
>> > The term I tried was PowerLawConvectionTerm(coeff=(-alpha/C**2, ),
>> var=C) but I got an error that said "IndexError: id1 does not have the same
>> size as b". The rest of the error log can be found at
>> http://pastebin.com/K4hy0GCh.
>>
>> The coefficient of a convection term must be rank-1 (a vector).
>>
>> A tuple of rank-0 Variables is not the same thing as a rank-1 Variable.
>> Since C is, by definition, scalar, alpha must be a vector, so you should be
>> able to write:
>>
>>  PowerLawConvectionTerm(coeff=([[-alpha]]/C**2, ), var=C)
>>
>> or, better yet,
>>
>>  PowerLawConvectionTerm(coeff=([[-alpha]]/C.faceValue**2, ), var=C)
>>
>>
>>
>> > I also tried declaring the term explicitly, but the solver could not
>> converge after when sharp gradients in C appeared.
>>
>> I don't find that altogether surprising. If you go the explicit route,
>> you may need to reduce your timesteps.
>>
>>
>> _______________________________________________
>> fipy mailing list
>> [email protected]
>> http://www.ctcms.nist.gov/fipy
>>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>>
>
>
_______________________________________________
fipy mailing list
[email protected]
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Reply via email to