[dune-fem] Fw: Fw: bug in alu or fem
Radcliffe, Alastair
A.Radcliffe at warwick.ac.uk
Sat Apr 11 14:02:27 CEST 2015
Sorry ... the message history got chopped ...
Alastair
________________________________________
From: Dedner, Andreas
Sent: Friday, April 10, 2015 11:49 PM
To: Radcliffe, Alastair
Subject: Re: Fw: bug in alu or fem
Hi Alastair.
I can't figure out anything at the moment.
You should perhaps report the bug to the dune-fem mailinglist with that
dgf file.
Andreas
On 10/04/15 19:15, Radcliffe, Alastair wrote:
> Hi,
>
> 'it' was a non-convergence of the solver, and the fem solvers didn't work either i think ... and of course, there are very few dune preconditioners that work in parallel to try ...
>
> the only way i've found to get it to 'solve' is to set the required residual to something e+3 or so ... then at least you can get paraview output to check it has "ranked" (ie: divided) the geometry up nicely ...
>
> i was hoping howto example 04 would do as a "minimal working example" for debugging ... but is it an alu or a dune-fem bug, so i can post it on the correct bug list ?
>
> Ta,
>
> Alastair
>
> ps: see you monday?, tuesday?, wednesday ? ...
>
> ________________________________________
> From: Dedner, Andreas
> Sent: Friday, April 10, 2015 4:18 PM
> To: Radcliffe, Alastair
> Subject: Re: Fw: bug in alu or fem
>
> Hi Alastair.
> Yes I can conform it - whatever 'it' was in your case. You really have
> to learn to write
> meaningful and helpful error reports! In my case the program got stuck -
> but is that the
> same behaviour you saw? Also: what sort of tests did you do? For
> example, did you see if
> there is an endless loop in the solver (setting the verbosity flag to
> 1)? I did do that and the
> solver then printed "nan" for the residual. It starts off ok but then
> goes bad. Did you test a
> different solver, i.e., change the flag in Makefile.am to not use istl.
> Anything else you tried/
> Best
> Andreas
>
>
>
> On 08/04/15 12:18, Radcliffe, Alastair wrote:
>> mpiexec -np 2 main
Hi Alastair.
I can't figure out anything at the moment.
You should perhaps report the bug to the dune-fem mailinglist with that
dgf file.
Andreas
Dedner, Andreas
Fri 4/10/2015 11:02 PM
To:
Radcliffe, Alastair;
Hi.
So the behaviour we are seeing is the sane, I'm inclined to think its a
dune-fem bug but I can't
think of anything causing it. I've never seen anything like it with the
different grids I tested -
and with the interval block grid it is working fine. We'll need to check
the DIrichlet constraints and
which are set I guess. Btw. if you want parallel preconditioners in fem
you need to install petsc.
They have great solvers.
We can meet Monday - I have a few meetings with phd students already but
I'll come by your
office.
Best
Andreas
Radcliffe, Alastair
Fri 4/10/2015 7:15 PM
Sent Items
> Hi,
>
> 'it' was a non-convergence of the solver, and the fem solvers didn't work either i think ... and of course, there are very few dune preconditioners that work in parallel to try ...
>
> the only way i've found to get it to 'solve' is to set the required residual to something e+3 or so ... then at least you can get paraview output to check it has "ranked" (ie: divided) the geometry up nicely ...
>
> i was hoping howto example 04 would do as a "minimal working example" for debugging ... but is it an alu or a dune-fem bug, so i can post it on the correct bug list ?
>
> Ta,
>
> Alastair
>
> ps: see you monday?, tuesday?, wednesday ? ...
Dedner, Andreas
Fri 4/10/2015 4:19 PM
> Hi Alastair.
> Yes I can conform it - whatever 'it' was in your case. You really have
> to learn to write
> meaningful and helpful error reports! In my case the program got stuck -
> but is that the
> same behaviour you saw? Also: what sort of tests did you do? For
> example, did you see if
> there is an endless loop in the solver (setting the verbosity flag to
> 1)? I did do that and the
> solver then printed "nan" for the residual. It starts off ok but then
> goes bad. Did you test a
> different solver, i.e., change the flag in Makefile.am to not use istl.
> Anything else you tried/
> Best
> Andreas
>
>
>
> On 08/04/15 12:18, Radcliffe, Alastair wrote:
>> mpiexec -np 2 main
Radcliffe, Alastair
Wed 4/8/2015 12:18 PM
Sent Items
Did you manage to confirm the bug ? ... or are you busy packing ?
Ta,
Alastair
Radcliffe, Alastair
Thu 4/2/2015 2:10 PM
Sent Items
I can't find another grid manager that will handle 3-d simplices, so i
don't know if there is a bug in alu or dune-fem, but the following
definitely is one:
In the fem-howto 04 example, set to ALUGRID_SIMPLEX 3 3
set precon to "none" in the parameter file, and use the mesh at the end
of this message.
in both serial and parallel mode it works fine if NO Dirichelt
conditions are present (the "cos" example), but if Dirichlet conditions
are there (the "sin" example) then the serial works (./main) but the
parallel (mpiexec -np 2 main) doesn't !!!
Please confirm.
And because i don't know if it is a bug in dune-fem or alu, i don't know
where to report it ...
Ta,
Alastair
DGF
VERTEX
-100.000e-02 0.000e-02 0.000e-02
-100.000e-02 0.000e-02 -100.000e-02
-100.000e-02 0.000e-02 100.000e-02
-100.000e-02 -100.000e-02 0.000e-02
-100.000e-02 100.000e-02 0.000e-02
-100.000e-02 -100.000e-02 -100.000e-02
-100.000e-02 -100.000e-02 100.000e-02
-100.000e-02 100.000e-02 -100.000e-02
-100.000e-02 100.000e-02 100.000e-02
-50.000e-02 -50.000e-02 -50.000e-02
-50.000e-02 -50.000e-02 50.000e-02
-50.000e-02 50.000e-02 -50.000e-02
-50.000e-02 50.000e-02 50.000e-02
0.000e-02 0.000e-02 0.000e-02
0.000e-02 0.000e-02 -100.000e-02
0.000e-02 0.000e-02 100.000e-02
0.000e-02 -100.000e-02 0.000e-02
0.000e-02 100.000e-02 0.000e-02
0.000e-02 -100.000e-02 -100.000e-02
0.000e-02 -100.000e-02 100.000e-02
0.000e-02 100.000e-02 -100.000e-02
0.000e-02 100.000e-02 100.000e-02
50.000e-02 -50.000e-02 -50.000e-02
50.000e-02 -50.000e-02 50.000e-02
50.000e-02 50.000e-02 -50.000e-02
50.000e-02 50.000e-02 50.000e-02
100.000e-02 0.000e-02 0.000e-02
100.000e-02 0.000e-02 -100.000e-02
100.000e-02 0.000e-02 100.000e-02
100.000e-02 -100.000e-02 0.000e-02
100.000e-02 100.000e-02 0.000e-02
100.000e-02 -100.000e-02 -100.000e-02
100.000e-02 -100.000e-02 100.000e-02
100.000e-02 100.000e-02 -100.000e-02
100.000e-02 100.000e-02 100.000e-02
#
SIMPLEX
9 5 18 1
9 18 14 1
9 5 18 16
9 5 16 3
9 18 14 16
9 14 13 16
9 14 1 13
9 1 0 13
9 1 5 3
9 1 0 3
9 0 3 16
9 0 16 13
10 3 16 0
10 16 13 0
10 3 16 19
10 3 19 6
10 16 13 19
10 13 15 19
10 13 0 15
10 0 2 15
10 0 3 6
10 0 2 6
10 2 6 19
10 2 19 15
11 1 14 7
11 14 20 7
11 1 14 13
11 1 13 0
11 14 20 13
11 20 17 13
11 20 7 17
11 7 4 17
11 7 1 0
11 7 4 0
11 4 0 13
11 4 13 17
12 0 13 4
12 13 17 4
12 0 13 15
12 0 15 2
12 13 17 15
12 17 21 15
12 17 4 21
12 4 8 21
12 4 0 2
12 4 8 2
12 8 2 15
12 8 15 21
22 18 31 14
22 31 27 14
22 18 31 29
22 18 29 16
22 31 27 29
22 27 26 29
22 27 14 26
22 14 13 26
22 14 18 16
22 14 13 16
22 13 16 29
22 13 29 26
23 16 29 13
23 29 26 13
23 16 29 32
23 16 32 19
23 29 26 32
23 26 28 32
23 26 13 28
23 13 15 28
23 13 16 19
23 13 15 19
23 15 19 32
23 15 32 28
24 14 27 20
24 27 33 20
24 14 27 26
24 14 26 13
24 27 33 26
24 33 30 26
24 33 20 30
24 20 17 30
24 20 14 13
24 20 17 13
24 17 13 26
24 17 26 30
25 13 26 17
25 26 30 17
25 13 26 28
25 13 28 15
25 26 30 28
25 30 34 28
25 30 17 34
25 17 21 34
25 17 13 15
25 17 21 15
25 21 15 28
25 21 28 34
#
BoundaryDomain
default 1 % all boundaries have id 1
#
GridParameter
name 3d-unit-cube % name of the grid
refinementedge arbitrary % refinement edge for bisectioning
overlap 1 % number of overlap cells
periodic % periodicity parameter for YaspGrid
closure none % no green closure on UGGrid
copies all % let UGGrid make copies
#
# unitcube3.dgf
More information about the dune-fem
mailing list