Unfortunately, this work has started in May 2005 under fipy 1.1 and I
understand that you suggest to upgrade not to waste too much time on an old
version of the tool.
The error message I sent to you appear a long time after the beginning of
the calculation, after hundred of correct uses of MA.filled or
numerix.MAtake functions, at the time when no more memory is available for
the calculation due to the loss of RAM problem.
I can give you more informations on the code but the remaining problem is
the loss of RAM. Did you succeed in getting such a loss of RAM problem
running under windows XP the little script I enclosed in a precedent mail on
the subject ?
----- Original Message -----
From: "Jonathan Guyer" <[EMAIL PROTECTED]>
To: "Multiple recipients of list" <[email protected]>
Sent: Tuesday, November 27, 2007 10:44 PM
Subject: Re: Loss of RAM
On Nov 27, 2007, at 3:37 PM, joelvuillon wrote:
I use fipy 1.1 under windows XP with the following elements :
:
:
which are the ones you recommend.
FiPy 1.2 has been available and recommended since last February. We
suggest that you upgrade, along with the packages that FiPy requires.
In addition, the error message me and another user working with the
elasticity code found is :
Ef = MA.filled(mu_lame_sffc*gradXStf + lambda_lame_sffc*tr_gradXS_If -
MemoryError: can't allocate memory for array
This error is usually due to a misinterpretation in the shape of some of
your vectors, so when multiplying, for instance, a [1xN] vector by a
[1xN] vector, you are getting an [NxN] vector instead of a [1xN] vector.
Can you show us more of the code that leads up to the declaration of Ef?
--
No virus found in this incoming message.
Checked by AVG Free Edition. Version: 7.5.503 / Virus Database:
269.16.8/1153 - Release Date: 26/11/2007 21:08
