[petsc-dev] I think equispaced nodes are in the wrong place

[Matthew Knepley]

On Mon, Jul 7, 2025 at 3:53 PM Toby Isaac <toby.isaac at gmail.com> wrote:

>
> The problem is that the node families come in two flavors: with endpoints
> and without endpoints.  When you left the node family as `DEFAULT`, it knew
> to choose "with endpoints" for continuous nodes, but by overriding the node
> family it didn't.  I made an MR and fixed that logic.  You can fix it
> before the MR gets merged from the command line with
> "-potential_petscdualspace_lagrange_node_endpoints 1"
>

Ah, thanks man.

   Matt


> On Mon, Jul 7, 2025 at 3:04 PM Matthew Knepley <knepley at gmail.com> wrote:
>
>> Try running this:
>>
>> make -f ./gmakefile test search="snes_tutorials-ex13_2d_p1_conv"
>> EXTRA_OPTIONS="-potential_petscspace_degree 3
>> -potential_petscdualspace_lagrange_node_type equispaced -petscds_view
>> ::ascii_info_detail"
>>
>> You see that
>>
>> # >           Dual basis vector 0
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (-0.333333, -0.333333) +1
>> # >           Dual basis vector 1
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (-0.25, -1) +1
>> # >           Dual basis vector 2
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (+0.25, -1) +1
>> # >           Dual basis vector 3
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (+0.25, -0.25) +1
>> # >           Dual basis vector 4
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (-0.25, +0.25) +1
>> # >           Dual basis vector 5
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (-1, +0.25) +1
>> # >           Dual basis vector 6
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (-1, -0.25) +1
>> # >           Dual basis vector 7
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (-1, -1) +1
>> # >           Dual basis vector 8
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (+1, -1) +1
>> # >           Dual basis vector 9
>> # >               Quadrature on a unknown of order -1 on 1 points (dim 2)
>> # >               p0 (-1, +1) +1
>>
>> but I think those should be +- 1/3, not 1/4. What is going on?
>>
>>   Thanks,
>>
>>      Matt
>> --
>> What most experimenters take for granted before they begin their
>> experiments is infinitely more interesting than any results to which their
>> experiments lead.
>> -- Norbert Wiener
>>
>> https://urldefense.us/v3/__https://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!bTyJShLisRzn1ylWPd1k-85-nIGObFzbLGZuzs9Xp3GfwVjtPvPwD3BiZIDF6bCcPHGdr8JdCHqiHw6dPaNq$ 
>> <https://urldefense.us/v3/__http://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!bTyJShLisRzn1ylWPd1k-85-nIGObFzbLGZuzs9Xp3GfwVjtPvPwD3BiZIDF6bCcPHGdr8JdCHqiH5-YL7WD$ >
>>
>

-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener

https://urldefense.us/v3/__https://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!bTyJShLisRzn1ylWPd1k-85-nIGObFzbLGZuzs9Xp3GfwVjtPvPwD3BiZIDF6bCcPHGdr8JdCHqiHw6dPaNq$  <https://urldefense.us/v3/__http://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!bTyJShLisRzn1ylWPd1k-85-nIGObFzbLGZuzs9Xp3GfwVjtPvPwD3BiZIDF6bCcPHGdr8JdCHqiH5-YL7WD$ >
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20250707/096f0b42/attachment-0001.html>