Re: [deal.II] Integrate difference isn't picking on the

[Daniel Arndt]

Abbas,

Try following the steps at
https://github.com/dealii/dealii/wiki/Contributing.

Best,
Daniel

On Thu, Jul 27, 2023 at 1:24 PM Abbas Ballout 
wrote:

> Lucas
> It's not the linker but I am guessing it's when I edit something deep
> within the library.h files.
> Since everything is linked to it many files are being rebuilt
> I didn't know about other linker so thanks for the tip. I am sure I will
> do that in the future.
>
> Anyone,
> This is my first pull request ever and I don't know exactly how to
> approach this.
> Details:
> I extended the VectorFunctionFromTensorFunction. to return the
> gradient of a function,
> I ended up editing the files: function.h and function.templates.h. but I
> ended up copying from other files.
>
> I am getting the right H1 convergence in my example code so I know what I
> am doing works.
> How do I proceed?
>
> Best,
> Abbas
>
>
> On Saturday, July 22, 2023 at 6:53:24 PM UTC+2 lucasm...@gmail.com wrote:
>
>> I haven't tried this with deal.II yet, but I've recently started using
>> the mold linker for programs I write which have many compilation units
>> (link to GitHub here: https://github.com/rui314/mold). If you're only
>> changing one line then probably only one compilation unit is being
>> recompiled, so the bulk of the wait time is probably being taken by the
>> linker.
>>
>> Note that, to get a significant speedup, you'll have to recompile with
>> all of your cores so the linking can be parallelized. I've found that using
>> all cores to compile sometimes causes make to get killed because it uses
>> too much memory, but if most things are compiled and you're just on the
>> linking stage you don't really run into this.
>>
>> Alternatively, I think that some other folks use gold linker or the lld
>> linker (in case you can't get mold to work).
>>
>> - Lucas
>>
>>
>> On Sat, Jul 22, 2023, 11:37 AM Abbas Ballout 
>> wrote:
>>
>>> If I change a single line in deal I'll have to wait a good amount of
>>> time for it to compile.
>>> Do you all go through this? All I have is a 4 cores and 16 GBs of RAM on
>>> my Laptop.
>>>
>>> On Monday, July 17, 2023 at 9:48:00 AM UTC+2 Abbas Ballout wrote:
>>>
 YES. Would take me a while but I ll do it.
 Be prepared for many questions.

 On Monday, July 17, 2023 at 2:44:01 AM UTC+2 Wolfgang Bangerth wrote:

> On 7/16/23 07:22, Abbas Ballout wrote:
> > I get this error when I call the
> exact_solution_vector_function.gradient(p1):
> > !ERROR Message:
> > /An error occurred in line <135> of file
> >
> 
> in function
> > dealii::Tensor<1, dim, Number> dealii::Function > RangeNumberType>::gradient(const dealii::Point&, unsigned int)
> const
> > [with int dim = 2; RangeNumberType = double]
> > The violated condition was:
> > false
> > Additional information:
> > You (or a place in the library) are trying to call a function
> that is
> > declared as a virtual function in a base class but that has not
> been
> > overridden in your derived class./
> > /
> > /
> > Calling value  works fine but not gradient even though I have
> overridden both
> > functions.
>
> You do in your own class, but VectorFunctionFromTensorFunction does
> not.
> That's not by design, but probably simply because nobody has ever had
> a need
> for this. Would you like to try and write a patch that adds
> VectorFunctionFromTensorFunction::gradient()? You could model it on
> VectorFunctionFromTensorFunction::value().
>
> Best
> Wolfgang
>
> --
> 
>
> Wolfgang Bangerth email: bang...@colostate.edu
> www: http://www.math.colostate.edu/~bangerth/
>
>
> --
>>> The deal.II project is located at http://www.dealii.org/
>>> For mailing list/forum options, see
>>> https://groups.google.com/d/forum/dealii?hl=en
>>> ---
>>> You received this message because you are subscribed to the Google
>>> Groups "deal.II User Group" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to dealii+un...@googlegroups.com.
>>> To view this discussion on the web visit
>>> https://groups.google.com/d/msgid/dealii/15f0b1e7-d5ea-49ed-b58f-ea10fae84d52n%40googlegroups.com
>>> 
>>> .
>>>
>> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see
> https://groups.google.com/d/forum/dealii?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "deal.II User Group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to dealii+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> 

Re: [deal.II] Integrate difference isn't picking on the

[Abbas Ballout]

Lucas 
It's not the linker but I am guessing it's when I edit something deep 
within the library.h files. 
Since everything is linked to it many files are being rebuilt   
I didn't know about other linker so thanks for the tip. I am sure I will do 
that in the future. 

Anyone, 
This is my first pull request ever and I don't know exactly how to approach 
this. 
Details: 
I extended the VectorFunctionFromTensorFunction. to return the 
gradient of a function, 
I ended up editing the files: function.h and function.templates.h. but I 
ended up copying from other files. 

I am getting the right H1 convergence in my example code so I know what I 
am doing works. 
How do I proceed? 

Best,
Abbas  
 

On Saturday, July 22, 2023 at 6:53:24 PM UTC+2 lucasm...@gmail.com wrote:

> I haven't tried this with deal.II yet, but I've recently started using the 
> mold linker for programs I write which have many compilation units (link to 
> GitHub here: https://github.com/rui314/mold). If you're only changing one 
> line then probably only one compilation unit is being recompiled, so the 
> bulk of the wait time is probably being taken by the linker. 
>
> Note that, to get a significant speedup, you'll have to recompile with all 
> of your cores so the linking can be parallelized. I've found that using all 
> cores to compile sometimes causes make to get killed because it uses too 
> much memory, but if most things are compiled and you're just on the linking 
> stage you don't really run into this. 
>
> Alternatively, I think that some other folks use gold linker or the lld 
> linker (in case you can't get mold to work). 
>
> - Lucas
>
>
> On Sat, Jul 22, 2023, 11:37 AM Abbas Ballout  wrote:
>
>> If I change a single line in deal I'll have to wait a good amount of time 
>> for it to compile. 
>> Do you all go through this? All I have is a 4 cores and 16 GBs of RAM on 
>> my Laptop.  
>>
>> On Monday, July 17, 2023 at 9:48:00 AM UTC+2 Abbas Ballout wrote:
>>
>>> YES. Would take me a while but I ll do it. 
>>> Be prepared for many questions. 
>>>
>>> On Monday, July 17, 2023 at 2:44:01 AM UTC+2 Wolfgang Bangerth wrote:
>>>
 On 7/16/23 07:22, Abbas Ballout wrote: 
 > I get this error when I call the 
 exact_solution_vector_function.gradient(p1): 
 > !ERROR Message: 
 > /An error occurred in line <135> of file 
 > 
 
  
 in function 
 > dealii::Tensor<1, dim, Number> dealii::Function>>> > RangeNumberType>::gradient(const dealii::Point&, unsigned int) 
 const 
 > [with int dim = 2; RangeNumberType = double] 
 > The violated condition was: 
 > false 
 > Additional information: 
 > You (or a place in the library) are trying to call a function 
 that is 
 > declared as a virtual function in a base class but that has not 
 been 
 > overridden in your derived class./ 
 > / 
 > / 
 > Calling value  works fine but not gradient even though I have 
 overridden both 
 > functions. 

 You do in your own class, but VectorFunctionFromTensorFunction does 
 not. 
 That's not by design, but probably simply because nobody has ever had a 
 need 
 for this. Would you like to try and write a patch that adds 
 VectorFunctionFromTensorFunction::gradient()? You could model it on 
 VectorFunctionFromTensorFunction::value(). 

 Best 
 Wolfgang 

 -- 
  

 Wolfgang Bangerth email: bang...@colostate.edu 
 www: http://www.math.colostate.edu/~bangerth/ 


 -- 
>> The deal.II project is located at http://www.dealii.org/
>> For mailing list/forum options, see 
>> https://groups.google.com/d/forum/dealii?hl=en
>> --- 
>> You received this message because you are subscribed to the Google Groups 
>> "deal.II User Group" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to dealii+un...@googlegroups.com.
>> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/dealii/15f0b1e7-d5ea-49ed-b58f-ea10fae84d52n%40googlegroups.com
>>  
>> 
>> .
>>
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/8ba758f4-06dc-4664-a715-72a7e60ff7bdn%40googlegroups.com.

Re: [deal.II] Integrate difference isn't picking on the

[Lucas Myers]

I haven't tried this with deal.II yet, but I've recently started using the
mold linker for programs I write which have many compilation units (link to
GitHub here: https://github.com/rui314/mold). If you're only changing one
line then probably only one compilation unit is being recompiled, so the
bulk of the wait time is probably being taken by the linker.

Note that, to get a significant speedup, you'll have to recompile with all
of your cores so the linking can be parallelized. I've found that using all
cores to compile sometimes causes make to get killed because it uses too
much memory, but if most things are compiled and you're just on the linking
stage you don't really run into this.

Alternatively, I think that some other folks use gold linker or the lld
linker (in case you can't get mold to work).

- Lucas


On Sat, Jul 22, 2023, 11:37 AM Abbas Ballout 
wrote:

> If I change a single line in deal I'll have to wait a good amount of time
> for it to compile.
> Do you all go through this? All I have is a 4 cores and 16 GBs of RAM on
> my Laptop.
>
> On Monday, July 17, 2023 at 9:48:00 AM UTC+2 Abbas Ballout wrote:
>
>> YES. Would take me a while but I ll do it.
>> Be prepared for many questions.
>>
>> On Monday, July 17, 2023 at 2:44:01 AM UTC+2 Wolfgang Bangerth wrote:
>>
>>> On 7/16/23 07:22, Abbas Ballout wrote:
>>> > I get this error when I call the
>>> exact_solution_vector_function.gradient(p1):
>>> > !ERROR Message:
>>> > /An error occurred in line <135> of file
>>> >
>>> 
>>> in function
>>> > dealii::Tensor<1, dim, Number> dealii::Function>> > RangeNumberType>::gradient(const dealii::Point&, unsigned int)
>>> const
>>> > [with int dim = 2; RangeNumberType = double]
>>> > The violated condition was:
>>> > false
>>> > Additional information:
>>> > You (or a place in the library) are trying to call a function that
>>> is
>>> > declared as a virtual function in a base class but that has not
>>> been
>>> > overridden in your derived class./
>>> > /
>>> > /
>>> > Calling value  works fine but not gradient even though I have
>>> overridden both
>>> > functions.
>>>
>>> You do in your own class, but VectorFunctionFromTensorFunction does not.
>>> That's not by design, but probably simply because nobody has ever had a
>>> need
>>> for this. Would you like to try and write a patch that adds
>>> VectorFunctionFromTensorFunction::gradient()? You could model it on
>>> VectorFunctionFromTensorFunction::value().
>>>
>>> Best
>>> Wolfgang
>>>
>>> --
>>> 
>>> Wolfgang Bangerth email: bang...@colostate.edu
>>> www: http://www.math.colostate.edu/~bangerth/
>>>
>>>
>>> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see
> https://groups.google.com/d/forum/dealii?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "deal.II User Group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to dealii+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/dealii/15f0b1e7-d5ea-49ed-b58f-ea10fae84d52n%40googlegroups.com
> 
> .
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/CALTw5sYqf2rOe664ZpMftqvMtQ79%3Dejt%3DzZA8Mw5GH5XJLYNcg%40mail.gmail.com.

Re: [deal.II] Integrate difference isn't picking on the

[Abbas Ballout]

If I change a single line in deal I'll have to wait a good amount of time 
for it to compile. 
Do you all go through this? All I have is a 4 cores and 16 GBs of RAM on my 
Laptop.  

On Monday, July 17, 2023 at 9:48:00 AM UTC+2 Abbas Ballout wrote:

> YES. Would take me a while but I ll do it. 
> Be prepared for many questions. 
>
> On Monday, July 17, 2023 at 2:44:01 AM UTC+2 Wolfgang Bangerth wrote:
>
>> On 7/16/23 07:22, Abbas Ballout wrote: 
>> > I get this error when I call the 
>> exact_solution_vector_function.gradient(p1): 
>> > !ERROR Message: 
>> > /An error occurred in line <135> of file 
>> > 
>> 
>>  
>> in function 
>> > dealii::Tensor<1, dim, Number> dealii::Function> > RangeNumberType>::gradient(const dealii::Point&, unsigned int) 
>> const 
>> > [with int dim = 2; RangeNumberType = double] 
>> > The violated condition was: 
>> > false 
>> > Additional information: 
>> > You (or a place in the library) are trying to call a function that 
>> is 
>> > declared as a virtual function in a base class but that has not 
>> been 
>> > overridden in your derived class./ 
>> > / 
>> > / 
>> > Calling value  works fine but not gradient even though I have 
>> overridden both 
>> > functions. 
>>
>> You do in your own class, but VectorFunctionFromTensorFunction does not. 
>> That's not by design, but probably simply because nobody has ever had a 
>> need 
>> for this. Would you like to try and write a patch that adds 
>> VectorFunctionFromTensorFunction::gradient()? You could model it on 
>> VectorFunctionFromTensorFunction::value(). 
>>
>> Best 
>> Wolfgang 
>>
>> -- 
>>  
>> Wolfgang Bangerth email: bang...@colostate.edu 
>> www: http://www.math.colostate.edu/~bangerth/ 
>>
>>
>>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/15f0b1e7-d5ea-49ed-b58f-ea10fae84d52n%40googlegroups.com.

Re: [deal.II] Integrate difference isn't picking on the

[Abbas Ballout]

YES. Would take me a while but I ll do it. 
Be prepared for many questions. 

On Monday, July 17, 2023 at 2:44:01 AM UTC+2 Wolfgang Bangerth wrote:

> On 7/16/23 07:22, Abbas Ballout wrote:
> > I get this error when I call the 
> exact_solution_vector_function.gradient(p1):
> > !ERROR Message:
> > /An error occurred in line <135> of file 
> > 
> 
>  
> in function
> > dealii::Tensor<1, dim, Number> dealii::Function > RangeNumberType>::gradient(const dealii::Point&, unsigned int) 
> const 
> > [with int dim = 2; RangeNumberType = double]
> > The violated condition was:
> > false
> > Additional information:
> > You (or a place in the library) are trying to call a function that is
> > declared as a virtual function in a base class but that has not been
> > overridden in your derived class./
> > /
> > /
> > Calling value  works fine but not gradient even though I have overridden 
> both 
> > functions.
>
> You do in your own class, but VectorFunctionFromTensorFunction does not. 
> That's not by design, but probably simply because nobody has ever had a 
> need 
> for this. Would you like to try and write a patch that adds 
> VectorFunctionFromTensorFunction::gradient()? You could model it on
> VectorFunctionFromTensorFunction::value().
>
> Best
> Wolfgang
>
> -- 
> 
> Wolfgang Bangerth email: bang...@colostate.edu
> www: http://www.math.colostate.edu/~bangerth/
>
>
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/c041d778-5abb-45c1-aa81-163e2a48d414n%40googlegroups.com.

Re: [deal.II] Integrate difference isn't picking on the

[Wolfgang Bangerth]

On 7/16/23 07:22, Abbas Ballout wrote:

I get this error when I call the exact_solution_vector_function.gradient(p1):
!ERROR Message:
/An error occurred in line <135> of file 
 in function
     dealii::Tensor<1, dim, Number> dealii::FunctionRangeNumberType>::gradient(const dealii::Point&, unsigned int) const 
[with int dim = 2; RangeNumberType = double]

The violated condition was:
     false
Additional information:
     You (or a place in the library) are trying to call a function that is
     declared as a virtual function in a base class but that has not been
     overridden in your derived class./
/
/
Calling value  works fine but not gradient even though I have overridden both 
functions.


You do in your own class, but VectorFunctionFromTensorFunction does not. 
That's not by design, but probably simply because nobody has ever had a need 
for this. Would you like to try and write a patch that adds 
VectorFunctionFromTensorFunction::gradient()? You could model it on

VectorFunctionFromTensorFunction::value().

Best
 Wolfgang

--

Wolfgang Bangerth  email: bange...@colostate.edu
   www: http://www.math.colostate.edu/~bangerth/


--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups "deal.II User Group" group.

To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/528f591f-ba68-0a94-d8d7-948665e1e119%40colostate.edu.

Re: [deal.II] Integrate difference isn't picking on the

[Abbas Ballout]

In hindsight I could've written something simpler. 
Here it is:


#include 
#include 

using namespace dealii;

template 
class Exact : public TensorFunction<1, dim>
{
public:
Exact() : TensorFunction<1, dim>(dim)
{}
virtual Tensor<1, dim> value(const Point ) const override;
virtual Tensor<2, dim> gradient(const Point ) const override;
};

template 
Tensor<1, dim> Exact::value(const Point ) const
{
return Tensor<1, dim>();
}

template 
Tensor<2, dim> Exact::gradient(const Point ) const
{
return Tensor<2, dim>();
}

int main()
{
const int dim = 2;
Exact exact_solution;
VectorFunctionFromTensorFunction 
exact_solution_vector_function(exact_solution, 
0, dim);

Point p1;
exact_solution_vector_function.value(p1);
exact_solution_vector_function.gradient(p1);

return 0;
}

I get this error when I call the exact_solution_vector_function.gradient
(p1):
!ERROR Message:







*An error occurred in line <135> of file 

 
in functiondealii::Tensor<1, dim, Number> dealii::Function::gradient(const dealii::Point&, unsigned int) const 
[with int dim = 2; RangeNumberType = double]The violated condition was: 
falseAdditional information: You (or a place in the library) are trying 
to call a function that isdeclared as a virtual function in a base 
class but that has not beenoverridden in your derived class.*

Calling value  works fine but not  gradient even though I have overridden 
both functions. 
On Friday, July 14, 2023 at 8:55:34 PM UTC+2 Wolfgang Bangerth wrote:

> On 7/14/23 09:34, Abbas Ballout wrote:
> > // THis throws exception
> > VectorTools::integrate_difference(dof_handler,
> > solution,
> > exact_solution_vector_function,
> > difference_per_cell,
> > quadrature_formula,
> > VectorTools::H1_seminorm);;
>
> Well, what's the error message? A good rule of thumb is that 50% debugging 
> consists of carefully reading what the error message says.
>
> Best
> W.
>
> -- 
> 
> Wolfgang Bangerth email: bang...@colostate.edu
> www: http://www.math.colostate.edu/~bangerth/
>
>
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/c1157723-d4a3-4c67-96f7-8eb98aa57fd5n%40googlegroups.com.

Re: [deal.II] Integrate difference isn't picking on the

[Wolfgang Bangerth]

On 7/14/23 09:34, Abbas Ballout wrote:

// THis throws exception
VectorTools::integrate_difference(dof_handler,
solution,
exact_solution_vector_function,
difference_per_cell,
quadrature_formula,
VectorTools::H1_seminorm);;


Well, what's the error message? A good rule of thumb is that 50% debugging 
consists of carefully reading what the error message says.


Best
 W.

--

Wolfgang Bangerth  email: bange...@colostate.edu
   www: http://www.math.colostate.edu/~bangerth/


--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups "deal.II User Group" group.

To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/2f8c9f5a-5592-42fd-779a-d592de4b336f%40colostate.edu.

[deal.II] Integrate difference isn't picking on the

[Abbas Ballout]

I am overriding the value and the gradient member functions in the 
TensorFunction class then similar to step 7 I am using compute_global_error to 
compute the L2: and H1 norms. I have to convert my TensorFunction to a 
vector valued function in order to do this using 
VectorFunctionFromTensorFunction.
I am getting an exception thrown when I ask for the H1 norm saying that the 
gradient operator is not overridden. 

Here is a sketch of the code:


class Exact : public TensorFunction<1, dim>
{
public:
Exact() : TensorFunction<1, dim>(dim)
{
}

virtual Tensor<1, dim> value(const Point ) const override;
virtual Tensor<2, dim> gradient(const Point ) const override;
};


Tensor<1, dim> Exact::value(const Point ) const
{
return Tensor<1, dim>;
}


Tensor<2, dim> Exact::gradient(const Point ) const
{
return Tensor<2, dim>;
}

int main()
{ 
//blablabla tri, dofs, fe..

 // declate then convert tensor function 
Exact exact_solution;
VectorFunctionFromTensorFunction exact_solution_vector_function(
exact_solution, 0, dim);

//L2 works fine
VectorTools::integrate_difference(dof_handler,
solution,
exact_solution_vector_function,
difference_per_cell,
quadrature_formula,
VectorTools::L2_norm);
// THis throws exception 
VectorTools::integrate_difference(dof_handler,
solution,
exact_solution_vector_function,
difference_per_cell,
quadrature_formula,
VectorTools::H1_seminorm);;


}

Doing something as simple as:
Point p1;
exact_solution_vector_function.value(p1); 
exact_solution_vector_function.gradient(p1); 
is raising the exception. 


I attached the code with the cmake below too.


-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/e7a9106b-ca2a-45b0-9d15-e74832198da4n%40googlegroups.com.
<>