[Getdp] Magnetostatic 3D

[Alain]

Olivier Castany wrote:

>Private message :
>-----------------
>
>Hello,
>
>you posted a message on the GetDP mailing list a few hours ago.
>
>It seems that you had an interesting private discussing with Bernhard 
>Kubicek. 
>
>I am also interested by a working source code with tree-gauge. If you 
>have a working example, could you consider sending it on the mailing 
>list (or putting it in the wiki) ?
>
>Regards,
>
>O.C.
>
>  
>
>>Kubicek Bernhard wrote:
>>
>>    
>>
>>>Hello Alain,
>>>
>>>after fooling around a bit with your files (and finding your 
>>>dual-constraint in the formulation quite useful for my stuff aswell 
>>>!) I managed to calculate a (probably) useful magnet field. However, 
>>>you cannot expect to get a physically meaningful vector-potential due 
>>>to the limitations of the tree-gauge, only the derived magnetic field 
>>>is meaningful.
>>>
>>> 
>>>
>>>      
>>>
>>Thank you for this clarification of the vector-potential problem which 
>>was puzzling. If I understand well, the best way to visualize the 
>>magnetic field lines is by using the streamlines plugin?
>>
>>    
>>
>>>See e.g. the following attached paper: Ticar, I.; Biro, O.; Preis, 
>>>K.: Vector potential expanded by edge basis functions associated with 
>>>loops on finite element facets. IEEE transactions on magnetics 38 
>>>(2002) , S. 437-440
>>>
>>>What I did, basically, to make your files work somehow is to make the 
>>>mesh finer, change nbfill in solverpar to 20 (I tend to use something 
>>>between the max/average bandwith of the inital matrix), and uncomment 
>>>the output to the mag. field.
>>>
>>>Personally, I would expand the calculation region further along the 
>>>axis of the core.
>>>
>>>Now, if I only could make my own calculations work (electric current 
>>>based magnetic vec. potential in 3d on a _hexahedral_ grid)....
>>>
>>>Very nice greetings,
>>>Bernhard
>>>      
>>>
>                             
>
>  
>
Hello Olivier,
    My sample code was send to the getdp list on 03/30/2007 and it is 
called ring_magnet..........
and it is sort of working. All communications was done through the list 
as far as I know.
I hope this help.
Alain