libmesh-users — General information for libmesh users

Re: [Libmesh-users] DG and boundary conditions

[Roy S. <roy...@ic...>]

On Tue, 1 May 2018, Renato Poli wrote:

> Are DirichletBoundary BCs expected to work in Discontinuous Galerkin
> (L2_LAGRANGE)?

Nope.  And you wouldn't want them to, would you?  I'm not a DG person
myself, but I was under the impression that you always want the
solution to be allowed to be discontinuous at *every* element
boundary, even the ones on the domain boundary, with Dirichlet
conditions instead enforced weakly via boundary penalties, to avoid
overconstraining your solution.

I'd love to let users be able to do "create a boundary condition
object and then forget" about it with DG as well as CG, but I don't
know if there's any simple way to make that happen; the boundary terms
are inherently formulation-dependent, aren't they?
---
Roy

Re: [Libmesh-users] Question regarding Subdivision Implementation in libMesh

[Roy S. <roy...@ic...>]

On Tue, 1 May 2018, Lee, Jae Ho wrote:

> Actually, this issue is now resolved. The problem was that the
> default linear solver was not converging. This issue can be resolved
> by using a stronger linear solver.

Thanks for the update!

You had me worried at first - I don't think any of the most active
developers are currently doing anything with subdivision elements,
which is exactly the sort of situation where regressions are a risk.
Good to hear that wasn't the case.

How far in the "stronger linear solver" direction did you have to go?
Stronger tolerance(s); stronger preconditioner; direct instead of
iterative?
---
Roy

Re: [Libmesh-users] Question regarding Subdivision Implementation in libMesh

[Lee, J. H. <jae...@li...>]

Actually, this issue is now resolved. The problem was that the default linear solver was not converging. This issue can be resolved by using a stronger linear solver.

Thank you,
Mike

> On Apr 25, 2018, at 2:56 PM, Lee, Jae Ho <jae...@li...> wrote:
> 
> I tried the subdivision example provided in libMesh, and I had some questions about the example that I wasn’t able to find an answer to by myself.
> 
> In the example (that follows Cirak’s paper from 2000), I see that there are the following steps before the subdivision mesh gets prepared:
> 
> // Quadrisect the mesh triangles a few times to obtain a
> 
> // finer mesh. Subdivision surface elements require the
> 
> // refinement data to be removed afterward.
> 
> MeshRefinement mesh_refinement(mesh);
> 
> mesh_refinement.uniformly_refine(3);
> 
> MeshTools::Modification::flatten(mesh);
> 
> 
> Here, what’s the exact effect of the mesh refinement? I just tried increasing the refinement to 5 or 6, and the solution started deviating quite a lot (by orders of magnitudes) from the analytical solution. I was just wondering if it’s due to just the nature of subdivision or due to the way libMesh works/handles it? Is there a reason 3 is chosen for refinement?
> 
> Thank you,
> Mike
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Libmesh-users mailing list
> Lib...@li...
> https://lists.sourceforge.net/lists/listinfo/libmesh-users

[Libmesh-users] DG and boundary conditions

[Renato P. <re...@gm...>]

Hi,

Are DirichletBoundary BCs expected to work in Discontinuous Galerkin
(L2_LAGRANGE)?

I'm working with interior penalties, but I cannot see the BCs being
applied in "dof_map.constrain_element_matrix_and_vector".

thanks,
Renato

[Libmesh-users] new libmeshdm fails with newtonrsls

[gmail <a.m...@gm...>]

Hi,

The new implementation of the libMeshDM unfortunately fails when using variational inequality solvers VINEWTONRSLS or VINEWTONSSLS (for example miscellaneous/ex7). This happens when PETSc tries to VecDuplicate the solution vector for lower bound in src/snes/impls/vi/vi.c: 383 ?!

I pulled the latest libMesh (48bd50d4a) this morning and compiled it with PETSc 3.8 (756c7f9f8).  I’m attaching my config log for libMesh.


Best,
Ata

Re: [Libmesh-users] SolverConfiguration?

[Derek G. <fri...@gm...>]

Ok - I just thought it was kind of odd that it has some data members and
stuff - but then doesn't do anything by default.

We use the command-line options for PETSc... but we set them
programmatically.  Gives us the most flexibility because there are
thousands of them and they are always changing.

Thanks for the reply - I was just making sure I wasn't missing anything.

Derek

On Fri, Apr 27, 2018 at 7:21 PM David Knezevic <dav...@ak...>
wrote:

> On Fri, Apr 27, 2018 at 6:40 PM, Derek Gaston <fri...@gm...> wrote:
>
>> What is up with SolverConfiguration?  It's abstract... and I can't see a
>> single implementation of it anywhere in libMesh.  In addition - everything
>> is public.
>>
>> Shouldn't there be a few implementations of this (like for PETSc, etc.)?
>>
>> What am I missing?
>>
>
>
> You're not missing anything, there are no subclasses of
> SolverConfiguration in libMesh itself. It is demonstrated in
> systems_of_equations_ex6, though.
>
> Personally, I use it a lot in app codes. I find it to be a good way to
> control solver options programmatically, i.e. without relying on command
> line options. You can override configure_solver and basically write
> whatever PETSc code you want in there to set the solver options.
>
> We could certainly add SolverConfiguration implementations to libMesh, but
> the main point was to allow users to customize solver options themselves,
> so I guess no one had a need to add any specific cases to libMesh so far.
>
> David
>

Re: [Libmesh-users] SolverConfiguration?

[David K. <dav...@ak...>]

On Fri, Apr 27, 2018 at 6:40 PM, Derek Gaston <fri...@gm...> wrote:

> What is up with SolverConfiguration?  It's abstract... and I can't see a
> single implementation of it anywhere in libMesh.  In addition - everything
> is public.
>
> Shouldn't there be a few implementations of this (like for PETSc, etc.)?
>
> What am I missing?
>


You're not missing anything, there are no subclasses of SolverConfiguration
in libMesh itself. It is demonstrated in systems_of_equations_ex6, though.

Personally, I use it a lot in app codes. I find it to be a good way to
control solver options programmatically, i.e. without relying on command
line options. You can override configure_solver and basically write
whatever PETSc code you want in there to set the solver options.

We could certainly add SolverConfiguration implementations to libMesh, but
the main point was to allow users to customize solver options themselves,
so I guess no one had a need to add any specific cases to libMesh so far.

David

[Libmesh-users] SolverConfiguration?

[Derek G. <fri...@gm...>]

What is up with SolverConfiguration?  It's abstract... and I can't see a
single implementation of it anywhere in libMesh.  In addition - everything
is public.

Shouldn't there be a few implementations of this (like for PETSc, etc.)?

What am I missing?

Derek

Re: [Libmesh-users] [RB] Question about "output assembly"

[David K. <dav...@ak...>]

On Fri, Apr 27, 2018 at 6:09 AM, <ss...@pu...> wrote:

> Hello, all.
>
>
>
> I have a question about "output assembly."
>
>
>
> To estimate the error, I know that we need to define the "output assembly"
> like "A0 assembly" and "F0 assembly" in the RB code.
>
> However, the RB example 5 does not have this code, so I do not know how the
> error is handled in the RB example 5.
>
>
>
> If we do not define the "output assembly," does the libMesh automatically
> perform the compliant problem?, i.e., "F0 assembly" = "output assembly?"
>



RB example 5 does not use an output. If you want to see how to define an
output, refer to RB example 1, for example.

Outputs are not required, and they are certainly not required for the error
estimation. The standard error estimation gives an estimate of the error in
the solution (based on the dual norm of the residual) and if an output is
defined, its error bound is derived from the solution's error bound.

David

[Libmesh-users] [RB] Question about "output assembly"

[<ss...@pu...>]

Hello, all.

 

I have a question about "output assembly."

 

To estimate the error, I know that we need to define the "output assembly"
like "A0 assembly" and "F0 assembly" in the RB code.

However, the RB example 5 does not have this code, so I do not know how the
error is handled in the RB example 5.

 

If we do not define the "output assembly," does the libMesh automatically
perform the compliant problem?, i.e., "F0 assembly" = "output assembly?"

 

I look forward to your reply.

 

Thank you.

 

Regards,

SKang

[Libmesh-users] Question regarding Subdivision Implementation in libMesh

[Lee, J. H. <jae...@li...>]

I tried the subdivision example provided in libMesh, and I had some questions about the example that I wasn’t able to find an answer to by myself.

In the example (that follows Cirak’s paper from 2000), I see that there are the following steps before the subdivision mesh gets prepared:

// Quadrisect the mesh triangles a few times to obtain a

// finer mesh. Subdivision surface elements require the

// refinement data to be removed afterward.

MeshRefinement mesh_refinement(mesh);

mesh_refinement.uniformly_refine(3);

MeshTools::Modification::flatten(mesh);


Here, what’s the exact effect of the mesh refinement? I just tried increasing the refinement to 5 or 6, and the solution started deviating quite a lot (by orders of magnitudes) from the analytical solution. I was just wondering if it’s due to just the nature of subdivision or due to the way libMesh works/handles it? Is there a reason 3 is chosen for refinement?

Thank you,
Mike

Re: [Libmesh-users] Qusetion about Implmentation of Gmsh

[Roy S. <roy...@ic...>]

On Tue, 24 Apr 2018, 서승진 wrote:

>    ERROR: Finite element LAGRANGE on geometric element TRI3 only supports FEInterface::max_order =1, not fe_type.order = 2

TRI3 is a first-order geometric element, with only 3 nodes.

It sounds like you're trying to set up a second-order finite element
basis, which for LAGRANGE requires 6 nodes.

If that's the case, the easiest fix would be to call
mesh.all_second_order(), which would turn your 3-noded triangles into
6-noded triangles, before initializing your EquationSystems.
---
Roy

Re: [Libmesh-users] time varying Dirichlet Conditions

[Roy S. <roy...@ic...>]

On Sat, 21 Apr 2018, Manav Bhatia wrote:

>   I am trying to identify the best approach to handle time-varying
>   Dirichlet conditions with libMesh. It appears that
>   DirichletConstraints is used to create a set of DofConstraintRow
>   that is used after the nonlinear solves through
>   DofMap::enforce_constraints_exactly().

After the nonlinear solves, but also during the solve process when you
apply constraints to your system matrix and rhs at each linearized
step.

>   So, if time-varying DirichletConstraints are required, then the
>   dof constraints will have to be reinitialized at each time-step
>   with the DirichletCondition object returning the function values
>   for that time?

That's correct.

>    Is there another way to do this without have to reinit the
>    constraints?

Weakly enforce penalty or Nitsche boundary conditions?  But for strong
BCs in libMesh, reinitializing the constraint equations is the way to
go.

>    Presently I am not using AMR. Would reinit of
>    constraints at each time-step be the best approach if AMR is
>    used?

Right.  We split out a reinit_constraints() option recently to make it
possible to do this without unnecessarily reinitializing other parts
of the problem at the same time.

Be careful about the time setting of your system when you reinit
time-varying constraints!  Often users will keep their System time set
to the time at the beginning of a time step, but constraining a
solution for that time step should be done with the constraint values
at the *end* of the time step.
---
Roy

[Libmesh-users] time varying Dirichlet Conditions

[Manav B. <bha...@gm...>]

Hi, 

   I am trying to identify the best approach to handle time-varying Dirichlet conditions with libMesh. It appears that DirichletConstraints is used to create a set of DofConstraintRow that is used after the nonlinear solves through DofMap::enforce_constraints_exactly(). 

   So, if time-varying DirichletConstraints are required, then the dof constraints will have to be reinitialized at each time-step with the DirichletCondition object returning the function values for that time? 

    Is there another way to do this without have to reinit the constraints? Presently I am not using AMR. Would reinit of constraints at each time-step be the best approach if AMR is used?

Thanks,
Manav

Re: [Libmesh-users] problems with installation

[Michael P. <mpo...@pu...>]

Thank you.

I upgraded to slepc 3.8 and it is configured correctly now.

Michael.


On 04/16/2018 10:14 AM, John Peterson wrote:
>
>
> On Fri, Apr 13, 2018 at 5:13 PM, Michael Povolotskyi 
> <mpo...@pu... <mailto:mpo...@pu...>> wrote:
>
>
>     Dear libmesh developers, I have a hard time to configure libmesh
>     with SLEPC.
>
>     This my configuration command:
>
>     ./configure
>     PETSC_DIR=/depot/kildisha/apps/brown/nemo5/libs/petsc/build-real
>     SLEPC_DIR=/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real
>     TSC_ARCH=linux F77="mpif77" CC="mpicc" GCC="gcc" CXX="mpicxx"
>     --enable-vtk
>     --with-vtk-include=/depot/kildisha/apps/brown/vtk/8.0.1_gcc-5.2.0_openmpi-2.1.0/include/vtk-8.0
>     --with-vtk-lib=/depot/kildisha/apps/brown/vtk/8.0.1_gcc-5.2.0_openmpi-2.1.0/lib
>     --disable-tetgen --enable-triangle --enable-slepc
>     --disable-nemesis --disable-strict-lgpl --enable-parmesh
>     --enable-amr --enable-shared=yes --disable-glibcxx-debugging
>     --enable-tecplot ;
>
>
>
> It appears your SLEPc is not installed correctly, as this line states:
>
>     /depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/conf/slepc_variables:22:
>     /depot/kildisha/apps/brown/nemo5/libs/petsc/build-real/conf/variables:
>     No such file or directory
>
>
> That is, line 22 of the slepc_variables file refers to a (probably 
> non-existent) PETSc build in 
> /depot/kildisha/apps/brown/nemo5/libs/petsc/build-real.
>
> Is it possible you recently rebuilt PETSc and SLEPc and your SLEPC_DIR 
> environment variable still points to an old SLEPc?
>
>
>
>     Please, advise. The config.log is attached.
>
>
> Sorry, unforunately no attachments are allowed on this list.
>
> -- 
> John

Re: [Libmesh-users] problems with installation

[John P. <jwp...@gm...>]

On Fri, Apr 13, 2018 at 5:13 PM, Michael Povolotskyi <mpo...@pu...>
wrote:

>
> Dear libmesh developers, I have a hard time to configure libmesh with
> SLEPC.
>
> This my configuration command:
>
> ./configure PETSC_DIR=/depot/kildisha/apps/brown/nemo5/libs/petsc/build-real
> SLEPC_DIR=/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real
> TSC_ARCH=linux F77="mpif77" CC="mpicc" GCC="gcc" CXX="mpicxx" --enable-vtk
> --with-vtk-include=/depot/kildisha/apps/brown/vtk/8.0.1_gcc-
> 5.2.0_openmpi-2.1.0/include/vtk-8.0 --with-vtk-lib=/depot/kildisha
> /apps/brown/vtk/8.0.1_gcc-5.2.0_openmpi-2.1.0/lib --disable-tetgen
> --enable-triangle --enable-slepc --disable-nemesis --disable-strict-lgpl
> --enable-parmesh --enable-amr --enable-shared=yes
> --disable-glibcxx-debugging --enable-tecplot ;
>
>
>

It appears your SLEPc is not installed correctly, as this line states:

/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/conf/slepc_variables:22:
> /depot/kildisha/apps/brown/nemo5/libs/petsc/build-real/conf/variables: No
> such file or directory


That is, line 22 of the slepc_variables file refers to a (probably
non-existent) PETSc build in
/depot/kildisha/apps/brown/nemo5/libs/petsc/build-real.

Is it possible you recently rebuilt PETSc and SLEPc and your SLEPC_DIR
environment variable still points to an old SLEPc?




> Please, advise. The config.log is attached.
>

Sorry, unforunately no attachments are allowed on this list.

-- 
John

Re: [Libmesh-users] Nonlinear solver

[John P. <jwp...@gm...>]

On Sun, Apr 15, 2018 at 5:04 PM, Renato Poli <re...@gm...> wrote:

> Hi
>
> I am struggling to get the nonlinear solver to work.
>
> I am probably misconfiguring jacobian and residual matrices - maybe
> not properly setting boundary conditions, not sure. My testcase is a
> simple linear elasticity equation, which worked previously in the
> linear approach.
>
> Can someone perhaps point out the "simplest example" of the
> petsc_nonlinear_solver?
> Any piece of code would really help.
>

miscellaneous_ex3 is probably the simplest example demonstrating this.

-- 
John

[Libmesh-users] Nonlinear solver

[Renato P. <re...@gm...>]

Hi

I am struggling to get the nonlinear solver to work.

I am probably misconfiguring jacobian and residual matrices - maybe
not properly setting boundary conditions, not sure. My testcase is a
simple linear elasticity equation, which worked previously in the
linear approach.

Can someone perhaps point out the "simplest example" of the
petsc_nonlinear_solver?
Any piece of code would really help.

I went through the examples and the mailing list, without success.

Thanks in advance,
Renato

[Libmesh-users] problems with installation

[Michael P. <mpo...@pu...>]

Dear libmesh developers, I have a hard time to configure libmesh with 
SLEPC.

This my configuration command:

./configure 
PETSC_DIR=/depot/kildisha/apps/brown/nemo5/libs/petsc/build-real 
SLEPC_DIR=/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real 
TSC_ARCH=linux F77="mpif77" CC="mpicc" GCC="gcc" CXX="mpicxx" 
--enable-vtk 
--with-vtk-include=/depot/kildisha/apps/brown/vtk/8.0.1_gcc-5.2.0_openmpi-2.1.0/include/vtk-8.0 
--with-vtk-lib=/depot/kildisha/apps/brown/vtk/8.0.1_gcc-5.2.0_openmpi-2.1.0/lib 
--disable-tetgen --enable-triangle --enable-slepc --disable-nemesis 
--disable-strict-lgpl --enable-parmesh --enable-amr --enable-shared=yes 
--disable-glibcxx-debugging --enable-tecplot ;


The error message reads:

checking 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/include/slepcversion.h 
usability... yes
checking 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/include/slepcversion.h 
presence... yes
checking for 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/include/slepcversion.h... 
yes
checking 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/linux/include/slepcconf.h 
usability... yes
checking 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/linux/include/slepcconf.h 
presence... yes
checking for 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/linux/include/slepcconf.h... 
yes
<<< Querying SLEPc configuration from 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/conf/slepc_variables 
 >>>
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/conf/slepc_variables:22: 
/depot/kildisha/apps/brown/nemo5/libs/petsc/build-real/conf/variables: 
No such file or directory
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/conf/slepc_variables:23: 
/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/linux/conf/slepcvariables: 
No such file or directory
make: *** No rule to make target 
`/depot/kildisha/apps/brown/nemo5/libs/slepc/build-real/linux/conf/slepcvariables'. 
Stop.


Indeed, there is no file 
/depot/kildisha/apps/brown/nemo5/libs/petsc/build-real/conf/variables. 
It tried to look for the file 'variables', and this is what I found:

pwd

/depot/kildisha/apps/brown/nemo5/libs/petsc

brown-fe02>find -name "variables"
./conf/variables
./build-cplx/lib/petsc/conf/variables
./petsc-3.7.2/lib/petsc/conf/variables
./build-real/lib/petsc/conf/variables

Please, advise. The config.log is attached.

Michael.

Re: [Libmesh-users] About Libmesh guideline

[Roy S. <roy...@ic...>]

On Wed, 11 Apr 2018, 서승진 wrote:

> 1) In visualization, I used paraview and open out_2.e (in introduction_ex4).
> 
> The paraview displayed the white box. I think that this is mesh for the example.
> 
> However, paraview didn't displayed the solution distribution (colored box).
> 
> Is the problem is come from the absence of post-processing in the example?

Did you tell paraview to display the solution distribution?

When I first open out_2.e in paraview 5.0.1, it displays nothing.

When I then click on the Properties->Variables->u checkbox and hit
Apply, it displays a white box (the domain for the example).

When I then change from "vtkBlockColors" to "u" in the active variable
controls toolbar, it displays a box colored by the solution.

> I understood that Libmesh examples use Dirichlet condition, and the part to set boundary condition is;
> 
> std::set<boundary_id_type> boundary_ids;
> boundary_ids.insert(0);
> boundary_ids.insert(1);
> boundary_ids.insert(2);
> boundary_ids.insert(3);
> DirichletBoundary dirichlet_bc(boundary_ids, variables, exact_solution_object);
> system.get_dof_map().add_dirichlet_boundary(dirichlet_bc);

> However, I couldn't find the part to set the initial value of u_var and value of boundary condition for each boundary.ids...
> 
> How I can set the value of initial condition and boundary conditions?

"exact_solution_object" in the above code is the functor which returns
boundary condition values.

Initial condition values are typically set with
System::project_solution().  See transient_ex1.C for an example.
   
> 3) In mailing list, some people mentioned about the example which
> contain Neumann boundary condition with penalty method,

This sounds confused.  The penalty method sets a Robin boundary
condition which is closely related to a desired Dirichlet boundary
condition, it's not for Neumann boundary conditions.

> so I've searched the example in Libmesh, but I couldn't find. Is
> there no neumann B.C. example in Libmesh?

systems_of_equations_ex2.C, if you modify it to use the
set_stagnation_bcs() function, has a homogeneous Neumann outflow
condition. fem_system_ex4.C has a homogeneous Neumann condition on one
side.

But homogeneous Neumann in most finite element formulations means "do
nothing", so if you have a heterogeneous Neumann boundary you probably
still want an example... let's see.

miscellaneous_ex9.C uses a constant Neumann flux = 1 on one domain side.

systems_of_equations_ex4.C (and _ex5.C, and _ex6.C...) use a constant
Neumann traction on one side.

> Thanks for reading my email.
> 
> For your answer, I successfully installed the libmesh.

You're welcome.

You should probably join and send future emails to libmesh-users,
though; that way they might not languish without reply quite as long
when I'm busy.
---
Roy

Re: [Libmesh-users] Install Libmesh with PETSc-3.2

[Roy S. <roy...@ic...>]

On Mon, 9 Apr 2018, 서승진 wrote:

> I tried to reinstall the libmesh with PETSc 3.2 ver. (instead of PETSc 3.8.3 ver.)

What was wrong with the PETSc 3.8.3 installation?  I admit I'm on
PETSc 3.7.6 right now myself, but I believe our other developers were
keeping our support up to date with PETSc 3.8 last year.

PETSc 3.2 is new enough that we still support it in *theory* (I
believe we only removed support for PETSC 2.3.2 and earlier in
2016...) but in practice everyone is using newer versions and a
compatibility error with 3.2 might have slipped in at some point.

Which version of libMesh are you building?

> then, some error occur at Makefile. the error log is followed;
> 
>  
> 
> Makefile:15309: 'src/numerics/libmesh_dbg_la-sparse_matrix.lo' command failed for the target.

This tells us *which* file failed to build, but not *why*.  There
should be some compiler error message which precedes this message and
which has more information.
---
Roy

Re: [Libmesh-users] About Libmesh guideline

[Roy S. <roy...@ic...>]

On Sat, 7 Apr 2018, 서승진 wrote:

> I've tried to search informations in google and libmesh mailing list, but it is not sufficient yet.

Well, let me Cc: the list with my replies, and hopefully the next
person to search will have an easier time.
   
> 1) I will solve the non-linear 2D~3D Poisson problem with implementation of PETSc, then
> 
>      what kinds of configuration options (for optional packacge) should I used? (except PETSc)


For optional features I usually use --enable-everything (which turns
on a half dozen borderline useful options) but there's nothing in
those which you really *need*, and for other options we try to enable
a good default configuration without you specifying everything on the
command line.

For optional packages, the only thing that's really important is PETSc
for parallel solves (or Trilinos, but I prefer PETSc) and SLEPc if you
intend to solve eigenproblems.

> 2) Is there no option "--download" instead of "--with" in configuration as like PETSc?

I'm afraid not.  For a couple packages we keep "contrib" directory
copies.

> 3) In example of introduction_ex4,  my computer displayed the different performance log, 

Sure, and that's expected and okay.  The PerfLog values will differ
depending on what computer runs the program and on what the libMesh
configuration is, and performance logging can be turned on or off
entirely at configuration time or run time.

> 4) I want to display the visualized result of introduction_ex4,
> which can be implmented with Paraview. How can I display the
> visualized result?

If your libMesh was configured with Exodus support, then
introduction_ex4 writes "out_2.e" in 2D and "out_3.e" in 3D, and you
can open either of those in Paraview.

Exodus is one of those packages we keep a "contrib" copy of, so it
should have been configured in unless you specifically disabled it.
---
Roy

Re: [Libmesh-users] Ways to modify order of Gaussian quadrature in reduced basis model

[吴家桦Gauvain <cau...@gm...>]

Thanks a lot for your prompt response and I figure out how to do it.

2018-04-10 1:10 GMT+08:00 David Knezevic <dav...@ak...>:

> On Mon, Apr 9, 2018 at 12:22 PM, 吴家桦Gauvain <cau...@gm...>
> wrote:
>
>> Hello,
>>
>> I've been working on reduced basis problems, in particular the
>> reduced_basis ex3. To improve the accuracy of integration of an
>> exponential
>> function, I want to set a high order for the Gaussian quadrature rule.
>> However, I found that the order of quadrature was decided by the FE_Type
>> object and I wonder if there is a way to allow the user to decide what
>> order rule is used for the integration. Thanks for your response.
>>
>
>
> The reduced basis code uses the FEMContext to define the quadrature rule.
> To vary the quadrature order that is used by the FEMContext you can either set
> the integer extra_quadrature_order in your System, or you can use
> the extra_quadrature_order argument to the FEMContext constructor.
>
> I suggest you refer to System and FEMContext for more info about this.
>
> David
>
>


-- 
*吴家桦 Gauvain*
*Mobile:13316300622*
*Email:g <gau...@fo...>auv...@gm...
<auv...@gm...>*
中山大学中法核工程与技术学院学生
Institut Franco-Chinois de l'Energie
Nucléaire, L'université Sun Yat-sen

Re: [Libmesh-users] Ways to modify order of Gaussian quadrature in reduced basis model

[David K. <dav...@ak...>]

On Mon, Apr 9, 2018 at 12:22 PM, 吴家桦Gauvain <cau...@gm...> wrote:

> Hello,
>
> I've been working on reduced basis problems, in particular the
> reduced_basis ex3. To improve the accuracy of integration of an exponential
> function, I want to set a high order for the Gaussian quadrature rule.
> However, I found that the order of quadrature was decided by the FE_Type
> object and I wonder if there is a way to allow the user to decide what
> order rule is used for the integration. Thanks for your response.
>


The reduced basis code uses the FEMContext to define the quadrature rule.
To vary the quadrature order that is used by the FEMContext you can either set
the integer extra_quadrature_order in your System, or you can use
the extra_quadrature_order argument to the FEMContext constructor.

I suggest you refer to System and FEMContext for more info about this.

David

[Libmesh-users] Ways to modify order of Gaussian quadrature in reduced basis model

[吴家桦Gauvain <cau...@gm...>]

Hello,

I've been working on reduced basis problems, in particular the
reduced_basis ex3. To improve the accuracy of integration of an exponential
function, I want to set a high order for the Gaussian quadrature rule.
However, I found that the order of quadrature was decided by the FE_Type
object and I wonder if there is a way to allow the user to decide what
order rule is used for the integration. Thanks for your response.

Regards,
Gauvain