<div dir="ltr"><div dir="ltr">On Tue, Apr 25, 2023 at 5:38 AM Karthikeyan Chockalingam - STFC UKRI <<a href="mailto:karthikeyan.chockalingam@stfc.ac.uk">karthikeyan.chockalingam@stfc.ac.uk</a>> wrote:<br></div><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="msg-6717559652267183367">





<div lang="EN-GB" style="overflow-wrap: break-word;">
<div class="m_-6717559652267183367WordSection1">
<p class="MsoNormal"><span style="font-size:11pt">Thank you for your response. I will come back to the error later (for now I have it working with
</span>-pc_fieldsplit_detect_saddle_point.)</p>
<p class="MsoNormal"><u></u> <u></u></p>
<p class="MsoNormal"><span style="font-size:11pt">I would like to understand how I need to index while running in parallel.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">(Q1)<b> Index set (IS) for the ISLocalToGlobalMapping.</b><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="m_-6717559652267183367p2"><span class="m_-6717559652267183367apple-converted-space">     </span></p>
<p class="m_-6717559652267183367p1"><span class="m_-6717559652267183367apple-converted-space">    </span><span class="m_-6717559652267183367s1"><b>for</b></span> (<span class="m_-6717559652267183367s1"><b>int</b></span> i =
<span class="m_-6717559652267183367s2">0</span>; i < N; i++)</p>
<p class="m_-6717559652267183367p1"><span class="m_-6717559652267183367apple-converted-space">        </span>vec_col_index[i] = i;</p>
<p class="m_-6717559652267183367p1"><u></u> <u></u></p>
<p class="m_-6717559652267183367p1"><span class="m_-6717559652267183367apple-converted-space">    </span><span class="m_-6717559652267183367s1"><b>for</b></span> (<span class="m_-6717559652267183367s1"><b>int</b></span> i =
<span class="m_-6717559652267183367s2">0</span>; i < C; i++)</p>
<p class="m_-6717559652267183367p1"><span class="m_-6717559652267183367apple-converted-space">        </span>vec_shifted_row_index[i] = i + N;<span class="m_-6717559652267183367apple-converted-space">    </span></p>
<p class="m_-6717559652267183367p1"><u></u> <u></u></p>
<p class="m_-6717559652267183367p2"><span class="m_-6717559652267183367apple-converted-space">    </span></p>
<p class="m_-6717559652267183367p1"><span class="m_-6717559652267183367apple-converted-space">    </span><span class="m_-6717559652267183367s3">ISLocalToGlobalMapping</span> phi_mapping, lamda_mapping;</p>
<p class="m_-6717559652267183367p2"><span class="m_-6717559652267183367apple-converted-space">    </span></p>
<p class="m_-6717559652267183367p4"><span class="m_-6717559652267183367apple-converted-space">    </span>ISLocalToGlobalMappingCreate<span class="m_-6717559652267183367s5">(</span>PETSC_COMM_WORLD<span class="m_-6717559652267183367s5">,
</span><span class="m_-6717559652267183367s2">1</span><span class="m_-6717559652267183367s5">, C, vec_shifted_row_index, </span>
PETSC_COPY_VALUES<span class="m_-6717559652267183367s5">, &lamda_mapping);</span></p>
<p class="m_-6717559652267183367p4"><span class="m_-6717559652267183367apple-converted-space">    </span>ISLocalToGlobalMappingCreate<span class="m_-6717559652267183367s5">(</span>PETSC_COMM_WORLD<span class="m_-6717559652267183367s5">,
</span><span class="m_-6717559652267183367s2">1</span><span class="m_-6717559652267183367s5">, N, vec_col_index, </span>PETSC_COPY_VALUES<span class="m_-6717559652267183367s5">, &phi_mapping);</span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="m_-6717559652267183367MsoListParagraph" style="margin-left:54pt">
<u></u><span style="font-size:11pt"><span>(i)<span style="font:7pt "Times New Roman"">                 
</span></span></span><u></u><span style="font-size:11pt">I believe the above index set is created by all parallel processes – why then I would have to offset it with the first index of this process?<u></u><u></u></span></p>
<p class="MsoNormal" style="margin-left:54pt"><span style="font-size:11pt">Since the mapping will always hold true and we would only use a
<u>subset</u> of this index set while populating the submatrices. I can see that I might have an over-allocated index set than necessary but I can’t see why it can be wrong.</span></p></div></div></div></blockquote><div><br></div><div>Oh, N is the global size of your matrix. Then what your L2G map says above is "replicate this matrix on every process", which is not a scalable solution. A scalable solution has only part of the rows on each process, which is the default in PETSc. In my example, I just</div><div>put the owned rows/cols on each process. You could choose to have an overlap, so that some shared rows also map to the local</div><div>representation, but I did not do that in my example.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="msg-6717559652267183367"><div lang="EN-GB" style="overflow-wrap: break-word;"><div class="m_-6717559652267183367WordSection1"><p class="MsoNormal" style="margin-left:54pt"><span style="font-size:11pt">
<u></u><u></u></span></p>
<p class="m_-6717559652267183367MsoListParagraph" style="margin-left:54pt">
<u></u><span style="font-size:11pt"><span>(ii)<span style="font:7pt "Times New Roman"">               
</span></span></span><u></u><span style="font-size:11pt">What would be the first index of this process anyway, since at this point I haven’t declared any parallel sub-matrix or sub-vector?</span></p></div></div></div></blockquote><div><br></div><div>Either you know the parallel layout of your matrix, or if you let PETSc decide for you, then this function determines it</div><div><br></div><div>  <a href="https://petsc.org/main/manualpages/Sys/PetscSplitOwnership/">https://petsc.org/main/manualpages/Sys/PetscSplitOwnership/</a></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="msg-6717559652267183367"><div lang="EN-GB" style="overflow-wrap: break-word;"><div class="m_-6717559652267183367WordSection1"><p class="MsoNormal"></p>
<p class="MsoNormal"><span style="font-size:11pt">(Q2)<b> Index set (IS) while creating the submatrix</b><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="m_-6717559652267183367p1"><span class="m_-6717559652267183367apple-converted-space">   </span><span class="m_-6717559652267183367s1"><b>for</b></span> (<span class="m_-6717559652267183367s1"><b>int</b></span> i =
<span class="m_-6717559652267183367s2">0</span>; i < C; i++)</p>
<p class="m_-6717559652267183367p1"><span class="m_-6717559652267183367apple-converted-space">       </span>vec_row_index[i] = i;</p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">  
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">Mat</span><span style="font-size:8.5pt;font-family:Menlo;color:black"> SubP;<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">   </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">IS</span><span style="font-size:8.5pt;font-family:Menlo;color:black"> row_isx,
 col_isx;<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">  <u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">   ISCreateGeneral(PETSC_COMM_WORLD, C, vec_row_index, PETSC_COPY_VALUES, &row_isx);<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">   ISCreateGeneral(PETSC_COMM_WORLD, N, vec_col_index, PETSC_COPY_VALUES, &col_isx);<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">   MatSetOption(</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">A</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">[level],
 MAT_NEW_NONZERO_ALLOCATION_ERR, PETSC_FALSE);<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">    <u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">   </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">MatSetLocalToGlobalMapping</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">A</span><span style="font-size:8.5pt;font-family:Menlo;color:black">,
 lamda_mapping,  phi_mapping);<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">    <u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">   </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">MatGetLocalSubMatrix</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">A</span><span style="font-size:8.5pt;font-family:Menlo;color:black">,
 row_isx, col_isx, &SubP);<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">I believe I will have to request a submatrix SubP of global size C x N on all processes – right? Or do you think I can still break row_isx and col_isx into parallel parts?</span></p></div></div></div></blockquote><div><br></div><div>The submatrix is local to this process, "LocalSubmatrix", and thus should have a local number of rows and columns. I have chosen the</div><div>number of owned rows/cols since I did not need an overlap.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="msg-6717559652267183367"><div lang="EN-GB" style="overflow-wrap: break-word;"><div class="m_-6717559652267183367WordSection1">
<p class="MsoNormal"><span style="font-size:11pt">(Q3)<b> Populating the submatrix<u></u><u></u></b></span></p>
<p class="MsoNormal"><b><span style="font-size:11pt"><u></u> <u></u></span></b></p>
<p class="MsoNormal" style="background:white"><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">    const</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black">
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">PetscInt</span><span style="font-size:8.5pt;font-family:Menlo;color:black"> *nindices;<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">   
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">PetscInt</span><span style="font-size:8.5pt;font-family:Menlo;color:black"> count;<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">   
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">ISGetIndices</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">isout</span><span style="font-size:8.5pt;font-family:Menlo;color:black">[level],
 &nindices);<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">        //isout is the global index of the nodes to be constrained.
<u></u><u></u></span></p>
<p class="MsoNormal"><b><span style="font-size:11pt"><u></u> <u></u></span></b></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">   
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">PetscScalar</span><span style="font-size:8.5pt;font-family:Menlo;color:black">   val =
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(28,0,207)">1</span><span style="font-size:8.5pt;font-family:Menlo;color:black">;<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">    <u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">   
</span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">for</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black"> (</span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">int</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black">
 irow=</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(28,0,207)">0</span><span style="font-size:8.5pt;font-family:Menlo;color:black">; irow < n; irow++)<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">    {<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">       
</span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">int</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black"> icol = nindices[irow];<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">       
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">MatSetValuesLocal</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(SubP,</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(28,0,207)">1</span><span style="font-size:8.5pt;font-family:Menlo;color:black">,&irow,</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(28,0,207)">1</span><span style="font-size:8.5pt;font-family:Menlo;color:black">,&icol,&val,</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">ADD_VALUES</span><span style="font-size:8.5pt;font-family:Menlo;color:black">);<u></u><u></u></span></p>
<p class="MsoNormal" style="background:white"><span style="font-size:8.5pt;font-family:Menlo;color:black">    }<u></u><u></u></span></p>
<p class="MsoNormal"><b><span style="font-size:11pt"><u></u> <u></u></span></b></p>
<p class="MsoNormal"><span style="font-size:11pt">I do think if I have the MatGetOwnerShipRange of SubP; I will be able to parallelize the loop. Is MatGetOwnerShipRange also available for submatrices as well?</span></p></div></div></div></blockquote><div><br></div><div>It is a local submatrix, so I would only run over local things (parallelization is implicit).</div><div><br></div><div>I think I have shown all these operations in the example.</div><div><br></div><div>  THanks,</div><div><br></div><div>     Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="msg-6717559652267183367"><div lang="EN-GB" style="overflow-wrap: break-word;"><div class="m_-6717559652267183367WordSection1">
<p class="MsoNormal"><span style="font-size:11pt">If I have to parallelize the index sets maybe I have to reorder the above steps, can you please make a suggestion?<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Kind regards,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Karthik.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
<div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(181,196,223);padding:3pt 0cm 0cm">
<p class="MsoNormal" style="margin-bottom:12pt"><b><span style="font-size:12pt;color:black">From:
</span></b><span style="font-size:12pt;color:black">Matthew Knepley <<a href="mailto:knepley@gmail.com" target="_blank">knepley@gmail.com</a>><br>
<b>Date: </b>Monday, 24 April 2023 at 19:24<br>
<b>To: </b>Chockalingam, Karthikeyan (STFC,DL,HC) <<a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank">karthikeyan.chockalingam@stfc.ac.uk</a>><br>
<b>Cc: </b><a href="mailto:petsc-users@mcs.anl.gov" target="_blank">petsc-users@mcs.anl.gov</a> <<a href="mailto:petsc-users@mcs.anl.gov" target="_blank">petsc-users@mcs.anl.gov</a>><br>
<b>Subject: </b>Re: [petsc-users] Setting up a matrix for Lagrange multiplier<u></u><u></u></span></p>
</div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">On Mon, Apr 24, 2023 at 1:58 PM Karthikeyan Chockalingam - STFC UKRI <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:11pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:11pt">>
 wrote:<u></u><u></u></span></p>
</div>
<div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Changing the names to NULL produced the same error.
<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">This is just to make the code simpler.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Don’t I have to give the fields a name?<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">They get the default name, which is the numbering you used.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">The problem solution didn’t change from one time step to the next.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[1;31m[0]PETSC ERROR: --------------------- Error Message --------------------------------------------------------------</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0;39m[0;49m[0]PETSC ERROR: No support for this operation for this object type</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: Unsupported viewer gmres</b></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">You are not completely catching the error, because it should print out the entire options database on termination.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Here it says you gave "gmres" as a Viewer, which is incorrect, but I cannot see all the options you used.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  Thanks,<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">     Matt<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: WARNING! There are option(s) set that were not used! Could be the program crashed before they were used or a spelling mistake, etc!</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: Option left: name:-fieldsplit_1_ksp_type value: preonly source: code</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: Option left: name:-fieldsplit_1_ksp_view value: gmres source: code</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: Option left: name:-fieldsplit_1_pc_type value: lu source: code</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: Option left: name:-options_left (no value) source: code</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: See </b><a href="https://petsc.org/release/faq/" target="_blank"><b>https://petsc.org/release/faq/</b></a><b> for trouble shooting.</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: Petsc Development GIT revision: v3.18.4-529-g995ec06f92<span class="m_-6717559652267183367m-526573182846331489apple-converted-space"> 
</span>GIT Date: 2023-02-03 18:41:48 +0000</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: /Users/karthikeyan.chockalingam/AMReX/amrFEM/build/Debug/amrFEM on a<span class="m_-6717559652267183367m-526573182846331489apple-converted-space"> 
</span>named HC20210312 by karthikeyan.chockalingam Mon Apr 24 18:51:25 2023</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: Configure options --with-debugging=0 --prefix=/Users/karthikeyan.chockalingam/AMReX/petsc --download-fblaslapack=yes --download-scalapack=yes --download-mumps=yes --with-hypre-dir=/Users/karthikeyan.chockalingam/AMReX/hypre/src/hypre</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: #1 PetscOptionsGetViewer() at /Users/karthikeyan.chockalingam/AMReX/SRC_PKG/petsc/src/sys/classes/viewer/interface/viewreg.c:309</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: #2 KSPSetFromOptions() at /Users/karthikeyan.chockalingam/AMReX/SRC_PKG/petsc/src/ksp/ksp/interface/itcl.c:522</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: #3 PCSetUp_FieldSplit() at /Users/karthikeyan.chockalingam/AMReX/SRC_PKG/petsc/src/ksp/pc/impls/fieldsplit/fieldsplit.c:1054</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: #4 PCSetUp() at /Users/karthikeyan.chockalingam/AMReX/SRC_PKG/petsc/src/ksp/pc/interface/precon.c:994</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: #5 KSPSetUp() at /Users/karthikeyan.chockalingam/AMReX/SRC_PKG/petsc/src/ksp/ksp/interface/itfunc.c:405</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: #6 KSPSolve_Private() at /Users/karthikeyan.chockalingam/AMReX/SRC_PKG/petsc/src/ksp/ksp/interface/itfunc.c:824</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>[0]PETSC ERROR: #7 KSPSolve() at /Users/karthikeyan.chockalingam/AMReX/SRC_PKG/petsc/src/ksp/ksp/interface/itfunc.c:1070</b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><b>End of program<span class="m_-6717559652267183367m-526573182846331489apple-converted-space"> </span></b></p>
<p class="m_-6717559652267183367m-526573182846331489p1"><span class="m_-6717559652267183367m-526573182846331489apple-converted-space"><b> </b></span><b>solve time 0.03416619 seconds<span class="m_-6717559652267183367m-526573182846331489apple-converted-space"> </span></b></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">It will be a bit difficult for me to produce a stand-alone code.
<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Thank you for your help.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"><br>
Best,<br>
Karthik.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(181,196,223);padding:3pt 0cm 0cm">
<p class="MsoNormal" style="margin-bottom:12pt"><b><span style="font-size:12pt;color:black">From:
</span></b><span style="font-size:12pt;color:black">Matthew Knepley <</span><a href="mailto:knepley@gmail.com" target="_blank"><span style="font-size:12pt">knepley@gmail.com</span></a><span style="font-size:12pt;color:black">><br>
<b>Date: </b>Monday, 24 April 2023 at 18:42<br>
<b>To: </b>Chockalingam, Karthikeyan (STFC,DL,HC) <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:12pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:12pt;color:black">><br>
<b>Cc: </b></span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black"> <</span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black">><br>
<b>Subject: </b>Re: [petsc-users] Setting up a matrix for Lagrange multiplier</span><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">On Mon, Apr 24, 2023 at 1:37 PM Karthikeyan Chockalingam - STFC UKRI <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:11pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:11pt">>
 wrote:<u></u><u></u></span></p>
</div>
<div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Great to know there is an example now.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Yes, -pc_fieldsplit_defect_saddle_point worked.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">But I would like to understand what I did wrong (and why it didn’t work).<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Can you show the error? Or give something self-contained that I can run.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">After reading many posted posts, I needed to create PCFieldSplitSetIS to split the fields.<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Yes. I would give NULL for the name here.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">I set the first N indices to the field phi and the next C indices (starting from N) to the field lambda.<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">This looks fine. However, these are global indices, so this would not work in parallel. You would have to offset by<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">the first index on this process.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  Thanks,<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">     Matt<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Please tell me what I did wrong and how I can fix the lines which are now commented?<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">    PetscErrorCode ierr;</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">KSP</span><span style="font-size:8.5pt;font-family:Menlo;color:black">           ksp;</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">PC</span><span style="font-size:8.5pt;font-family:Menlo;color:black">            pc;</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black"> </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">    KSPCreate(PETSC_COMM_WORLD, &ksp);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">KSPSetType</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(ksp,
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(100,56,32)">KSPGMRES</span><span style="font-size:8.5pt;font-family:Menlo;color:black">);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">KSPSetOperators</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(ksp,
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">A</span><span style="font-size:8.5pt;font-family:Menlo;color:black">[level],
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">A</span><span style="font-size:8.5pt;font-family:Menlo;color:black">[level]);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    ierr = </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">KSPGetPC</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(ksp,&pc);</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(100,56,32)">CHKERRQ</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(ierr);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    ierr = </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">PCSetType</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(pc,</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(100,56,32)">PCFIELDSPLIT</span><span style="font-size:8.5pt;font-family:Menlo;color:black">);</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(100,56,32)">CHKERRQ</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(ierr);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"> </p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1">    ……..</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"> </p>
<p class="MsoNormal" style="background:white">
<b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">    for</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black"> (</span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">int</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black">
 i = </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(28,0,207)">0</span><span style="font-size:8.5pt;font-family:Menlo;color:black">; i < N; i++)</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">      vec_zero_field_index[i] = i;</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">PetscInt</span><span style="font-size:8.5pt;font-family:Menlo;color:black"> * vec_first_field_index =
</span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">NULL</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black">;</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(100,56,32)">PetscMalloc</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(n *
</span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">sizeof</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black">(</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(57,0,160)">PetscInt</span><span style="font-size:8.5pt;font-family:Menlo;color:black">),
 &vec_first_field_index);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">     </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">    </span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">for</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black"> (</span><b><span style="font-size:8.5pt;font-family:Menlo;color:rgb(155,35,147)">int</span></b><span style="font-size:8.5pt;font-family:Menlo;color:black">
 i = </span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(28,0,207)">0</span><span style="font-size:8.5pt;font-family:Menlo;color:black">; i < C; i++)</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:black">        vec_first_field_index[i] = N + i;</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"> </p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s1">    IS</span> zero_field_isx, first_field_isx;</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">CHKERRQ</span>(<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">ISCreateGeneral</span>(<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PETSC_COMM_WORLD</span>, N, vec_zero_field_index,
<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PETSC_COPY_VALUES</span>, &zero_field_isx));</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">CHKERRQ</span>(<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">ISCreateGeneral</span>(<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PETSC_COMM_WORLD</span>, C, vec_first_field_index,
<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PETSC_COPY_VALUES</span>, &first_field_isx));</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p2"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">  
</span></p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PCFieldSplitSetIS</span>(pc,<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s4">"0"</span>,zero_field_isx);<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space"> 
</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">CHKERRQ</span>(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PCFieldSplitSetIS</span>(pc,<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s4">"1"</span>,first_field_isx);<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space"> 
</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">CHKERRQ</span>(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p2"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">    </span></p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PetscOptionsSetValue</span>(<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s5"><b>NULL</b></span>,<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s4">"-ksp_type"</span>,
<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s4">"fgmres"</span>); <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">
CHKERRQ</span>(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PetscOptionsSetValue</span>(<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s5"><b>NULL</b></span>,<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s4">"-pc_type"</span>,
<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s4">"fieldsplit"</span>); <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">
CHKERRQ</span>(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"> </p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s6">/*</span></p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p3"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = PetscOptionsSetValue(NULL,"-fieldsplit_0_ksp_type", "gmres"); CHKERRQ(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p3"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = PetscOptionsSetValue(NULL,"-fieldsplit_0_pc_type", "jacobi"); CHKERRQ(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p3"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = PetscOptionsSetValue(NULL,"-fieldsplit_1_ksp_type", "preonly"); CHKERRQ(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p3"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = PetscOptionsSetValue(NULL,"-fieldsplit_1_pc_type", "lu"); CHKERRQ(ierr);*/</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p3"> </p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p4"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s7">ierr = </span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PetscOptionsSetValue</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s7">(</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s5"><b>NULL</b></span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s7">,
</span>"-pc_fieldsplit_detect_saddle_point"<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s7">,
</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s5"><b>NULL</b></span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s7">);</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">CHKERRQ</span><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s7">(ierr);</span></p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">PetscOptionsSetValue</span>(<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s5"><b>NULL</b></span>,
<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s4">"-options_left"</span>, <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s5">
<b>NULL</b></span>);<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">CHKERRQ</span>(ierr);</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080apple-converted-space">   
</span>ierr = <span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s3">KSPSetFromOptions</span>(ksp);
<span class="m_-6717559652267183367m-526573182846331489m7795183717874518080s2">CHKERRQ</span>(ierr);</p>
<p class="MsoNormal" style="background:white">
<span style="font-size:8.5pt;font-family:Menlo;color:rgb(108,54,169)">    KSPSolve</span><span style="font-size:8.5pt;font-family:Menlo;color:black">(ksp,
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">b</span><span style="font-size:8.5pt;font-family:Menlo;color:black">,
</span><span style="font-size:8.5pt;font-family:Menlo;color:rgb(50,109,116)">x</span><span style="font-size:8.5pt;font-family:Menlo;color:black">);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080p1"> </p>
<p class="MsoNormal"><span style="font-size:11pt">Best,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Karthik.<u></u><u></u></span></p>
<div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(181,196,223);padding:3pt 0cm 0cm">
<p class="MsoNormal" style="margin-bottom:12pt"><b><span style="font-size:12pt;color:black">From:
</span></b><span style="font-size:12pt;color:black">Matthew Knepley <</span><a href="mailto:knepley@gmail.com" target="_blank"><span style="font-size:12pt">knepley@gmail.com</span></a><span style="font-size:12pt;color:black">><br>
<b>Date: </b>Monday, 24 April 2023 at 17:57<br>
<b>To: </b>Chockalingam, Karthikeyan (STFC,DL,HC) <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:12pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:12pt;color:black">><br>
<b>Cc: </b></span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black"> <</span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black">><br>
<b>Subject: </b>Re: [petsc-users] Setting up a matrix for Lagrange multiplier</span><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">On Mon, Apr 24, 2023 at 10:22 AM Karthikeyan Chockalingam - STFC UKRI <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:11pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:11pt">>
 wrote:<u></u><u></u></span></p>
</div>
<div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Hello,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">I was able to construct the below K matrix (using submatrices P and P^T), which is of type MATAIJ
<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black">K = [A P^T</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black">       P   0]</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">and solved them using a direct solver.<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">I modified your example to create either AIJ or Nest matrices, but use the same assembly code:<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  </span><a href="https://gitlab.com/petsc/petsc/-/merge_requests/6368" target="_blank"><span style="font-size:11pt">https://gitlab.com/petsc/petsc/-/merge_requests/6368</span></a><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">However, I was reading online that this is a saddle point problem and I should be employing PCFIELDSPLIT.
<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Since I have one monolithic matrix K, I was not sure how to split the fields.<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">With this particular matrix, you can use<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  -pc_fieldsplit_detect_saddle_point<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">and it will split it automatically.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  Thanks,<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">     Matt<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Best regards,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Karthik.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(181,196,223);padding:3pt 0cm 0cm">
<p class="MsoNormal" style="margin-bottom:12pt"><b><span style="font-size:12pt;color:black">From:
</span></b><span style="font-size:12pt;color:black">Chockalingam, Karthikeyan (STFC,DL,HC) <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:12pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:12pt;color:black">><br>
<b>Date: </b>Wednesday, 19 April 2023 at 17:52<br>
<b>To: </b>Matthew Knepley <</span><a href="mailto:knepley@gmail.com" target="_blank"><span style="font-size:12pt">knepley@gmail.com</span></a><span style="font-size:12pt;color:black">><br>
<b>Cc: </b></span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black"> <</span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black">><br>
<b>Subject: </b>Re: [petsc-users] Setting up a matrix for Lagrange multiplier</span><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black"> I have declared the mapping </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black"> </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black">  ISLocalToGlobalMapping mapping;</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black">    ISLocalToGlobalMappingCreate(PETSC_COMM_WORLD, 1, n, nindices, PETSC_COPY_VALUES, &mapping);</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black"> </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black">But when I use MatSetValuesLocal(), how do I know the above mapping is employed because it is not one of the parameters passed to the function?</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black"> </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black">Thank you.</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black"> </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black">Kind regards,</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal" style="background:white">
<span style="font-size:11pt;color:black">Karthik.</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(181,196,223);padding:3pt 0cm 0cm">
<p class="MsoNormal" style="margin-bottom:12pt"><b><span style="font-size:12pt;color:black">From:
</span></b><span style="font-size:12pt;color:black">Matthew Knepley <</span><a href="mailto:knepley@gmail.com" target="_blank"><span style="font-size:12pt">knepley@gmail.com</span></a><span style="font-size:12pt;color:black">><br>
<b>Date: </b>Tuesday, 18 April 2023 at 16:21<br>
<b>To: </b>Chockalingam, Karthikeyan (STFC,DL,HC) <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:12pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:12pt;color:black">><br>
<b>Cc: </b></span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black"> <</span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black">><br>
<b>Subject: </b>Re: [petsc-users] Setting up a matrix for Lagrange multiplier</span><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">On Tue, Apr 18, 2023 at 11:16 AM Karthikeyan Chockalingam - STFC UKRI <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:11pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:11pt">>
 wrote:<u></u><u></u></span></p>
</div>
<div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080m-968874379040273580m-4128250693350498318p1">
Thank you for your response. I spend some time understanding how</p>
<p class="m_-6717559652267183367m-526573182846331489m7795183717874518080m-968874379040273580m-4128250693350498318p1">
MatSetValuesLocal and ISLocalToGlobalMappingCreate work.</p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">You can look at SNES ex28 where we do this with DMCOMPOSITE.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Q1) Will the matrix K be of type MATMPIAIJ or MATIS?<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black">K = [A P^T</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black">       P   0]</span><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">I assume MPIAIJ since IS is only used for Neumann-Neumann decompositions.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Q2) Can I use both MatSetValues() to MatSetValuesLocal() to populate K? Since I have already used MatSetValues() to construct A.<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">You can, and there would be no changes in serial if K is exactly the upper left block, but in parallel global indices would change.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Q3) What are the advantages of using MatSetValuesLocal()? Is it that I can construct P directly using local indies and map the entrees to the global
 index in K?<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">You have a monolithic K, so that you can use sparse direct solvers to check things. THis is impossible with separate storage.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Q4) I probably don’t have to construct an independent P matrix<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">You wouldn't in this case.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  Thanks,<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">     Matt<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Best regards,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Karthik.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(181,196,223);padding:3pt 0cm 0cm">
<p class="MsoNormal" style="margin-bottom:12pt"><b><span style="font-size:12pt;color:black">From:
</span></b><span style="font-size:12pt;color:black">Matthew Knepley <</span><a href="mailto:knepley@gmail.com" target="_blank"><span style="font-size:12pt">knepley@gmail.com</span></a><span style="font-size:12pt;color:black">><br>
<b>Date: </b>Tuesday, 18 April 2023 at 11:08<br>
<b>To: </b>Chockalingam, Karthikeyan (STFC,DL,HC) <</span><a href="mailto:karthikeyan.chockalingam@stfc.ac.uk" target="_blank"><span style="font-size:12pt">karthikeyan.chockalingam@stfc.ac.uk</span></a><span style="font-size:12pt;color:black">><br>
<b>Cc: </b></span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black"> <</span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:12pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:12pt;color:black">><br>
<b>Subject: </b>Re: [petsc-users] Setting up a matrix for Lagrange multiplier</span><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">On Tue, Apr 18, 2023 at 5:24 AM Karthikeyan Chockalingam - STFC UKRI via petsc-users <</span><a href="mailto:petsc-users@mcs.anl.gov" target="_blank"><span style="font-size:11pt">petsc-users@mcs.anl.gov</span></a><span style="font-size:11pt">>
 wrote:<u></u><u></u></span></p>
</div>
<div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt;color:black">Hello,</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black"> </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black">I'm solving a problem using the Lagrange multiplier, the matrix has the form</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black"> </span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black">K = [A P^T</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt;color:black">       P   0]</span><span style="font-size:11pt"><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">I am familiar with constructing K using MATMPIAIJ. However, I would like to know if had [A], can I augment it with [P], [P^T] and [0] of type MATMPIAIJ?
 Likewise for vectors as well.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Can you please point me to the right resource, if it is a common operation in PETSc?<u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">You can do this at least 2 ways:<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  1) Assemble you submatrices directly into the larger matrix by constructing local-to-global maps for the emplacement. so that you do<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">      not change your assembly code, except to change MatSetValues() to MatSetValuesLocal(). This is usually preferable.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  2) Use MATNEST and VecNEST to put pointers to submatrices and subvectors directly in.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">  Thanks,<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">     Matt<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin:5pt 0cm 5pt 4.8pt">
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">Many thanks.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Kind regards,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt">Karthik.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
</div>
</div>
</blockquote>
</div>
<p class="MsoNormal"><span style="font-size:11pt"><br clear="all">
<u></u><u></u></span></p>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<p class="MsoNormal"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080m-968874379040273580m-4128250693350498318gmailsignatureprefix"><span style="font-size:11pt">--
</span></span><span style="font-size:11pt"><u></u><u></u></span></p>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">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<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank"><span style="font-size:11pt">https://www.cse.buffalo.edu/~knepley/</span></a><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
</div>
<p class="MsoNormal"><span style="font-size:11pt"><br clear="all">
<u></u><u></u></span></p>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<p class="MsoNormal"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080m-968874379040273580gmailsignatureprefix"><span style="font-size:11pt">--
</span></span><span style="font-size:11pt"><u></u><u></u></span></p>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">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<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank"><span style="font-size:11pt">https://www.cse.buffalo.edu/~knepley/</span></a><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
</div>
<p class="MsoNormal"><span style="font-size:11pt"><br clear="all">
<u></u><u></u></span></p>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<p class="MsoNormal"><span class="m_-6717559652267183367m-526573182846331489m7795183717874518080gmailsignatureprefix"><span style="font-size:11pt">--
</span></span><span style="font-size:11pt"><u></u><u></u></span></p>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">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<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank"><span style="font-size:11pt">https://www.cse.buffalo.edu/~knepley/</span></a><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
</div>
<p class="MsoNormal"><span style="font-size:11pt"><br clear="all">
<u></u><u></u></span></p>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<p class="MsoNormal"><span class="m_-6717559652267183367m-526573182846331489gmailsignatureprefix"><span style="font-size:11pt">--
</span></span><span style="font-size:11pt"><u></u><u></u></span></p>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">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<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank"><span style="font-size:11pt">https://www.cse.buffalo.edu/~knepley/</span></a><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
</div>
<p class="MsoNormal"><span style="font-size:11pt"><br clear="all">
<u></u><u></u></span></p>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<p class="MsoNormal"><span class="m_-6717559652267183367gmailsignatureprefix"><span style="font-size:11pt">--
</span></span><span style="font-size:11pt"><u></u><u></u></span></p>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11pt">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<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11pt"><u></u> <u></u></span></p>
</div>
<div>
<p class="MsoNormal"><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank"><span style="font-size:11pt">https://www.cse.buffalo.edu/~knepley/</span></a><span style="font-size:11pt"><u></u><u></u></span></p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>

</div></blockquote></div><br clear="all"><div><br></div><span class="gmail_signature_prefix">-- </span><br><div dir="ltr" class="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div>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><br></div><div><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank">https://www.cse.buffalo.edu/~knepley/</a><br></div></div></div></div></div></div></div></div>