<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Mon, Mar 7, 2016 at 10:28 AM, Lukasz Kaczmarczyk <span dir="ltr"><<a href="mailto:Lukasz.Kaczmarczyk@glasgow.ac.uk" target="_blank">Lukasz.Kaczmarczyk@glasgow.ac.uk</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div style="word-wrap:break-word">
<div>Many thanks all for help. I started to implement function for DM. </div>
<div><br>
</div>
<div>I understand that minimal implementation is that for the DM i need to have, is to have DMCoarsen and in each level for all DMs, set operators</div>
<div>DMKSPSetComputeOperators and DMCreateInterpolation. Matrix matrix free P from DMCreateInterpolation have to have operators for mult and mult_traspose. Is that is all? </div></div></blockquote><div><br></div><div>Yes, that should be it. It would be nice to have some example that does this if you would</div><div>be willing to contribute some version of your code.</div><div><br></div><div> Thanks,</div><div><br></div><div> Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div style="word-wrap:break-word">
<div>Kind regards,</div>
<div>Lukasz</div><div><div class="h5">
<div><br>
</div>
<div><br>
</div>
<br>
<div>
<blockquote type="cite">
<div>On 7 Mar 2016, at 15:55, Mark Adams <<a href="mailto:mfadams@lbl.gov" target="_blank">mfadams@lbl.gov</a>> wrote:</div>
<br>
<div>
<div dir="ltr">You can just set the coarse grid matrix/operator instead of using Galerkin. If you have a shell (matrix free) P then you will need to create and set this yourself. Our Galerkin requires a matrix P.</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Mon, Mar 7, 2016 at 9:32 AM, Lukasz Kaczmarczyk <span dir="ltr">
<<a href="mailto:Lukasz.Kaczmarczyk@glasgow.ac.uk" target="_blank">Lukasz.Kaczmarczyk@glasgow.ac.uk</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div>
<div><br>
> On 7 Mar 2016, at 14:21, Lawrence Mitchell <<a href="mailto:lawrence.mitchell@imperial.ac.uk" target="_blank">lawrence.mitchell@imperial.ac.uk</a>> wrote:<br>
><br>
> On 07/03/16 14:16, Lukasz Kaczmarczyk wrote:<br>
>><br>
>>> On 7 Mar 2016, at 13:50, Matthew Knepley <<a href="mailto:knepley@gmail.com" target="_blank">knepley@gmail.com</a><br>
>>> <mailto:<a href="mailto:knepley@gmail.com" target="_blank">knepley@gmail.com</a>>> wrote:<br>
>>><br>
>>> On Mon, Mar 7, 2016 at 6:58 AM, Lukasz Kaczmarczyk<br>
>>> <<a href="mailto:Lukasz.Kaczmarczyk@glasgow.ac.uk" target="_blank">Lukasz.Kaczmarczyk@glasgow.ac.uk</a><br>
>>> <mailto:<a href="mailto:Lukasz.Kaczmarczyk@glasgow.ac.uk" target="_blank">Lukasz.Kaczmarczyk@glasgow.ac.uk</a>>> wrote:<br>
>>><br>
>>> Hello,<br>
>>><br>
>>> I run multi-grid solver, with adaptivity, works well, however It<br>
>>> is some space for improving efficiency. I using hierarchical<br>
>>> approximation basis, for which<br>
>>> construction of interpolation operators is simple, it is simple<br>
>>> injection.<br>
>>><br>
>>> After each refinement level (increase of order of approximation<br>
>>> on some element) I rebuild multigrid pre-conditioner with<br>
>>> additional level. It is a way to add dynamically new levels<br>
>>> without need of rebuilding whole MG pre-conditioner.<br>
>>><br>
>>><br>
>>> That does not currently exist, however it would not be hard to add,<br>
>>> since the MG structure jsut consists of<br>
>>> arrays of pointers.<br>
>>><br>
>>><br>
>>> Looking at execution profile I noticed that 50%-60% of time is<br>
>>> spent on MatPtAP function during PCSetUP stage.<br>
>>><br>
>>><br>
>>> Which means you are using a Galerkin projection to define the coarse<br>
>>> operator. Do you have a direct way of defining<br>
>>> this operator (rediscretization)?<br>
>><br>
>> Matt,<br>
>><br>
>><br>
>> Thanks for swift response. You are right, I using Galerkin projection.<br>
>><br>
>> Yes, I have a way to get directly coarse operator, it is some sub<br>
>> matrix of whole matrix. I taking advantage here form hierarchical<br>
>> approximation.<br>
>><br>
>> I could reimplement PCSetUp_MG to set the MG structure directly, but<br>
>> this probably not good approach, since my implementation which will<br>
>> work with current petsc version could be incompatible which future<br>
>> changes in native MG data structures. The alternative option is to<br>
>> hack MatPtAP itself, and until petsc MG will use this, whatever<br>
>> changes you will make in MG in the future my code will work.<br>
><br>
> Why not provide a shell DM to the KSP that knows how to compute the<br>
> operators (and how to refine/coarsen and therefore<br>
> restrict/interpolate). Then there's no need to use Galerkin coarse<br>
> grid operators, and the KSP will just call back to your code to create<br>
> the appropriate matrices.<br>
<br>
</div>
</div>
Hello Lawrence,<br>
<br>
Thanks, it is good advice.<br>
I have already my DM shell, however I have not looked yet how make it in the context of MG. Now is probably time to do that.<br>
<br>
DM shell<br>
<a href="http://userweb.eng.gla.ac.uk/lukasz.kaczmarczyk/MoFem/html/group__dm.html" rel="noreferrer" target="_blank">http://userweb.eng.gla.ac.uk/lukasz.kaczmarczyk/MoFem/html/group__dm.html</a><br>
<br>
<br>
Regards,<br>
Lukasz</blockquote>
</div>
<br>
</div>
</div>
</blockquote>
</div>
<br>
</div></div></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature">What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener</div>
</div></div>