Daniel,
in my script there is :
a = numerix.MAtake(b,c)
and c is masked : c represents the 6 adjacent cells of each cell in an
hexaedral meshing. c can be only defined for 3, 4 or 5 cells on the edge of
the meshing and consequently c takes masked values.
If you have an idea to by-pass the use of MAtake in this case, I am ok. For
example I can use 0 values of the indices instead of masked values but I
still have to take care to the real 0 indice of the first cell of the
meshing in the connectivity table.
Bests,
Joël
----- Original Message -----
From: "Daniel Wheeler" <[EMAIL PROTECTED]>
To: "Multiple recipients of list" <[email protected]>
Sent: Wednesday, November 28, 2007 4:52 PM
Subject: Re: Loss of RAM
On Nov 28, 2007 10:25 AM, joelvuillon <[EMAIL PROTECTED]> wrote:
in the same time, I will try the different paths you and Jonathan provide
:
importing MA : from fipy.tools.numerix import MA
discard the use of the old Numeric to avoid interference with numpy (I
use
numpy.__version__ '1.0.3' and you seem to have no problem with
numpy.__version__ '1.0.1')
replace the use of numerix.MAtake which may cause, to my opinion, the
main
troubles
We wrote MAtake as we thought MA.take didn't do what we wanted.
For example take the operation a = MA.take(b, c). I think I am correct in
saying that MA.take works as you expect if b is masked. However, if c
is masked the operation doesn't work although there doesn't
seem to be a good reason why it shouldn't. At least in your script, if c
is
not masked you don't have to use MAtake. (I may have mixed the ordering
of b and c)
Cheers
--
Daniel Wheeler
--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.503 / Virus Database: 269.16.9/1155 - Release Date:
27/11/2007 20:30
