Problems with saving results to file using trilinos

Asked by Karen Støverud on 2013-01-25

I am running code based on cbc.block, which use Trilinos as linear algebra backend.
Recently I tried to run the code in parallell, but I cannot write to file and I get the following error:

*** Error: Unable to update ghost values.
*** Reason: Not implemented by current linear algebra backend.
*** Where: This error was encountered inside GenericVector.h.

Would be so happy if anyone responded to my first question ever on this or any similar list :o)

Karen

Question information

Language:
English Edit question
Status:
Answered
For:
DOLFIN Edit question
Assignee:
No assignee Edit question
Last query:
2013-01-25
Last reply:
2013-01-31
Johan Hake (johan-hake) said : #1

update_ghost_values is implemented in EpetraVector so that should not be
the problem. Are you using another linear algebra backend?

If you are able to write a simple script that reproduces the error it
would be easier to find what is failing.

Johan

On 01/25/2013 10:45 AM, Karen Støverud wrote:
> New question #220121 on DOLFIN:
> https://answers.launchpad.net/dolfin/+question/220121
>
> I am running code based on cbc.block, which use Trilinos as linear algebra backend.
> Recently I tried to run the code in parallell, but I cannot write to file and I get the following error:
>
> *** Error: Unable to update ghost values.
> *** Reason: Not implemented by current linear algebra backend.
> *** Where: This error was encountered inside GenericVector.h.
>
>
> Would be so happy if anyone responded to my first question ever on this or any similar list :o)
>
> Karen
>

Joachim Haga (jobh) said : #2

This is a simple example that fails in parallel (with any linear algebra backend). It looks like update... isn't implemented in the Vector wrapper.

from dolfin import *
m = UnitSquareMesh(2,2)
V = FunctionSpace(m, 'CG', 1)
v = TestFunction(V)
b = assemble(v*dx)
u = Function(V, b)
f = File('u.pvd')
f << u

Johan Hake (johan-hake) said : #3

Thanks for reporting.

This should now be fixed in trunk.

Johan

On 01/30/2013 10:31 PM, Joachim Haga wrote:
> Question #220121 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/220121
>
> Joachim Haga posted a new comment:
> This is a simple example that fails in parallel (with any linear algebra
> backend). It looks like update... isn't implemented in the Vector
> wrapper.
>
> from dolfin import *
> m = UnitSquareMesh(2,2)
> V = FunctionSpace(m, 'CG', 1)
> v = TestFunction(V)
> b = assemble(v*dx)
> u = Function(V, b)
> f = File('u.pvd')
> f << u
>

Joachim Haga (jobh) said : #4

Thanks Johan! I suggest also 1.1, since it's a straightforward bug.
Den 30. jan. 2013 23:11 skrev "Johan Hake" <
<email address hidden>> følgende:

> Question #220121 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/220121
>
> Johan Hake proposed the following answer:
> Thanks for reporting.
>
> This should now be fixed in trunk.
>
> Johan
>
> On 01/30/2013 10:31 PM, Joachim Haga wrote:
> > Question #220121 on DOLFIN changed:
> > https://answers.launchpad.net/dolfin/+question/220121
> >
> > Joachim Haga posted a new comment:
> > This is a simple example that fails in parallel (with any linear algebra
> > backend). It looks like update... isn't implemented in the Vector
> > wrapper.
> >
> > from dolfin import *
> > m = UnitSquareMesh(2,2)
> > V = FunctionSpace(m, 'CG', 1)
> > v = TestFunction(V)
> > b = assemble(v*dx)
> > u = Function(V, b)
> > f = File('u.pvd')
> > f << u
> >
>
> --
> You received this question notification because you are a member of
> DOLFIN Team, which is an answer contact for DOLFIN.
>

Johan Hake (johan-hake) said : #5

Then I guess you do not get the commit emails, because that was done
yesterday ;)

Johan

On 01/31/2013 08:51 AM, Joachim Haga wrote:
> Question #220121 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/220121
>
> Joachim Haga proposed the following answer:
> Thanks Johan! I suggest also 1.1, since it's a straightforward bug.
> Den 30. jan. 2013 23:11 skrev "Johan Hake" <
> <email address hidden>> følgende:
>
>> Question #220121 on DOLFIN changed:
>> https://answers.launchpad.net/dolfin/+question/220121
>>
>> Johan Hake proposed the following answer:
>> Thanks for reporting.
>>
>> This should now be fixed in trunk.
>>
>> Johan
>>
>> On 01/30/2013 10:31 PM, Joachim Haga wrote:
>>> Question #220121 on DOLFIN changed:
>>> https://answers.launchpad.net/dolfin/+question/220121
>>>
>>> Joachim Haga posted a new comment:
>>> This is a simple example that fails in parallel (with any linear algebra
>>> backend). It looks like update... isn't implemented in the Vector
>>> wrapper.
>>>
>>> from dolfin import *
>>> m = UnitSquareMesh(2,2)
>>> V = FunctionSpace(m, 'CG', 1)
>>> v = TestFunction(V)
>>> b = assemble(v*dx)
>>> u = Function(V, b)
>>> f = File('u.pvd')
>>> f << u
>>>
>>
>> --
>> You received this question notification because you are a member of
>> DOLFIN Team, which is an answer contact for DOLFIN.
>>
>

Joachim Haga (jobh) said : #6

Lovely! No, I'm just an innocent bystander ;)

On 31 January 2013 09:01, Johan Hake
<email address hidden>wrote:

> Question #220121 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/220121
>
> Johan Hake proposed the following answer:
> Then I guess you do not get the commit emails, because that was done
> yesterday ;)
>
> Johan
>
> On 01/31/2013 08:51 AM, Joachim Haga wrote:
> > Question #220121 on DOLFIN changed:
> > https://answers.launchpad.net/dolfin/+question/220121
> >
> > Joachim Haga proposed the following answer:
> > Thanks Johan! I suggest also 1.1, since it's a straightforward bug.
> > Den 30. jan. 2013 23:11 skrev "Johan Hake" <
> > <email address hidden>> følgende:
> >
> >> Question #220121 on DOLFIN changed:
> >> https://answers.launchpad.net/dolfin/+question/220121
> >>
> >> Johan Hake proposed the following answer:
> >> Thanks for reporting.
> >>
> >> This should now be fixed in trunk.
> >>
> >> Johan
> >>
> >> On 01/30/2013 10:31 PM, Joachim Haga wrote:
> >>> Question #220121 on DOLFIN changed:
> >>> https://answers.launchpad.net/dolfin/+question/220121
> >>>
> >>> Joachim Haga posted a new comment:
> >>> This is a simple example that fails in parallel (with any linear
> algebra
> >>> backend). It looks like update... isn't implemented in the Vector
> >>> wrapper.
> >>>
> >>> from dolfin import *
> >>> m = UnitSquareMesh(2,2)
> >>> V = FunctionSpace(m, 'CG', 1)
> >>> v = TestFunction(V)
> >>> b = assemble(v*dx)
> >>> u = Function(V, b)
> >>> f = File('u.pvd')
> >>> f << u
> >>>
> >>
> >> --
> >> You received this question notification because you are a member of
> >> DOLFIN Team, which is an answer contact for DOLFIN.
> >>
> >
>
> --
> You received this question notification because you are a member of
> DOLFIN Team, which is an answer contact for DOLFIN.
>

Can you help with this problem?

Provide an answer of your own, or ask Karen Støverud for more information if necessary.

To post a message you must log in.