Re-2: [Getdp] Warning from TransferInitSolution

[Christophe Geuzaine]

Matthias - The problem comes from the fact that TranferSolution will 
only work if the discrete quantities being transferred actually belong 
to the same FunctionSpace. (Patrick: any comments?)

What I would do to solve the problem (warning: I haven't actually tested 
this...) is define a single function space

FunctionSpace {
   { Name Hgrad_T; Type Form0;
     BasisFunction {
       { Name sn; NameOfCoef Tn; Function BF_Node; Support Vol_The;
         Entity NodesOf[All]; }
     }
     Constraint {
       { NameOfCoef Tn; EntityType NodesOf ; NameOfConstraint T; }
     }
   }
}

and define the constraint as follows:

Constraint {
   { Name T ;
     Case {
       { Region hot; Type Assign; Value 150.; }
       { Region cold; Type Assign; Value 5.; }
       { Region Vol_The; Type InitFromResolution ; NameOfResolution ini;}
     }
   }
}

The TransferSolution operation in the "ini" formulation will then 
transfer all the "non-assigned" DoFs as initial conditions in the "dyn" 
formulation.

I will try to have an more in-depth look at your example as soon as I 
have some spare time, but it's pretty hard these days :-)

Best,

Christophe


m.fenner at gmx.net wrote:
> Dear Christophe,
> 
> I had some time to visit my wire / TransferSolution problem again. I tried to follow your suggestion using TransferSolution but still got no result. 
> 
> What I tried is the following:
> 
> I DO NOT solve anything in the dynamical system but just "InitSolution [Tdyn]; SaveSolution [Tdyn] ;" (This starts the pre-resolution "ini" which gives the correct result when run alone.) After that I PostProcess and PostOperate. Then I'd expect to find the solution I got from the pre resolution via TransferSolution. Is that right or have I already messed up the solution by now? Some problem with my syntax e.g.?
> 
> I tried this with different Constraints:
> a) just setting the Dirichlet constraints (if I understood your suggestion correctly)
> b) AssignFromResolution
> c) InitFromResolution
> 
> but never retrieved my pre-resolution result (see the attached files). But I am still not sure if the pre-resolution actually transfers the data to the system Tdyn - again the question: Where do I loose the information?
> 
> When I try to actually solve the dynamical system I get:
> a) a rather weird result
> b) "Warning   : Generated system is of dimension zero" and T=0 everywhere
> c) the same result as a)
> 
> There is a work around (for 1-d Systems at least):
> 
> Export the data from the pre-resolution, and use the Init Constraint with the linear interpolation of these data. This gives a quite plausible result which actually answers my initial physical question. But a more general solution is desireable, isn't it? Unfortunately I could not get in touch with Ahmed Rassili yet to consult him on his experience.
> 
> As far as know there is no working example with TransferSolution in the mailing list archives, too. I shall attach the *.pro, *.geo, and *.dat files I used - just in case you find some time to look at this problem.
> 
> Best regards
> 
> Matthias
> 
> P.S. thanks for the hint concerning the magnetic flux in a sphere. It seems the texbooks on general physics are far too sloppy in their treatment of this stuff. You have to read the specialized literature.
> 
> Matthias
> 
>                           
> ---------------------------------------------------------------------
>      Matthias A. Fenner
> 
>      email: m.fenner at gmx.net                     Hugo-Eckener-Str. 94
>      fon: +49 6131 622 10 60                     D-55122 Mainz
>                                    
> ---------------------------------------------------------------------
> 
> 
> 
> -------- Original Message --------
> Subject: Re: [Getdp] Warning from TransferInitSolution (08-Sep-2004 5:32)
> From:    geuzaine at acm.caltech.edu
> To:      m.fenner at omicron.de
> 
> 
>>m.fenner at gmx.net wrote:
>>
>>>Dear Christophe,
>>>
>>>I want to solve the nonlinear time-dependent problem of a wire with
>>>ends held at different temperatures and an electric current switched
>>>on at a certain time. So I solve the 1-d Laplace equation first
>>>(resolution ini, system T) and then transfer this solution as an
>>>initial condition to the full system (resolution dyn, system Tdyn).
>>>ini works allright by itself. But when I try "TransferInitSolution T"
>>>I get a "GetDP: Warning: Operation: ???". The dynamical simulation
>>>then starts with T~10^-307.
>>>
>>>Where do I loose the information of the pre-resolution?
>>
>>Hi Matthias - I just checked in the code and we actually never 
>>implemented "TransferInitSolution". But I think you should still be able 
>>to do what you want by just using "TransferSolution", followed by the 
>>regular "InitSolution" on the other system. (At least that's how I 
>>remember doing it a loooong time ago.)
>>
>>I'm CC'ing Ahmed Rassili on this one: I think he's still using something 
>>similar in his simulations.
>>
>>Best,
>>
>>Christophe
>>
>>PS: Thanks for the examples in the Wiki! (I've added a couple of comments.)
>>
>>-- 
>>Christophe Geuzaine
>>Applied and Computational Mathematics, Caltech
>>geuzaine at acm.caltech.edu - http://geuz.org
>>
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> getdp mailing list
> getdp at geuz.org
> http://www.geuz.org/mailman/listinfo/getdp


-- 
Christophe Geuzaine
Applied and Computational Mathematics, Caltech
geuzaine at acm.caltech.edu - http://geuz.org