On Fri, 11 Jun 2010, Yujie wrote:
> Thanks, Roy.
>
> You mean like
> std::vector* > subA1(N);
Yes
> for i: 1: N
> {
> tempA1(A(i));
> subA1[i]=&tempA1;
> }?
No
http://en.wikipedia.org/wiki/Dangling_pointer
---
Roy
Thanks, Roy.
You mean like
std::vector* > subA1(N);
for i: 1: N
{
tempA1(A(i));
subA1[i]=&tempA1;
}?
Regards,
Yujie
On Fri, Jun 11, 2010 at 4:42 PM, Roy Stogner wrote:
>
> On Fri, 11 Jun 2010, Yujie wrote:
>
>> However, from DenseSubMatrix class, I must initialize subA1[i],
>> subA2[i] usi
On Fri, 11 Jun 2010, Yujie wrote:
> However, from DenseSubMatrix class, I must initialize subA1[i],
> subA2[i] using A(i). Before that, I want to "new" subA1 and subA2
> array with N dimension. It looks there is not corresponding
> constructor function for it to my understanding. Therefore, I can
Dear LibMesh Developers,
I am sorry for this simple question.
I am using DenseMatrix and DenseSubMatrix. However, I have several
DenseMatrix. I would like to get them using
"std::vector > A(N);"
For each A(i), I have two DenseSubMatrix for it, that is subA1[i], subA2[i].
However, from DenseSubMa
On Fri, Jun 11, 2010 at 11:11 AM, Lorenzo Botti wrote:
> It turns out that the problem was due to PETSc and not to libMesh.
> If I build libMesh without PETSc support I have no compatibility issues.
>
> Actually I'm not able to run PETSc examples compiled on the xeon front-end
> on opteron nodes.
It turns out that the problem was due to PETSc and not to libMesh.
If I build libMesh without PETSc support I have no compatibility issues.
Actually I'm not able to run PETSc examples compiled on the xeon front-end
on opteron nodes.
I just get "Illegal Instruction"
Are there any flags for the com
Just to be completely clear: 4.5 takes longer to COMPILE a program than 4.2
does (twice as long in some instances). As Cody mentioned the speed of the
executables created by 4.5 is comparable to those created with 4.2. Only
the compile time is different.
Derek
On Fri, Jun 11, 2010 at 8:27 AM, C
One more thing to note here: The current binaries of GCC (4.5.x) available at
hpc.sourceforge.net are quite a bit slower than the earlier version 4.x (where
x < 5) compilers. Derek and I discovered this recently and reverted our
compilers back to version 4.2. We haven't really looked into it
