<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">Am 04.09.2012 16:20, schrieb Hong
      Zhang:<br>
    </div>
    <blockquote
cite="mid:CAGCphBsHwed=ga+KvjrwXz5BnNFxXNw8J_U2e16=GaRDw13Yug@mail.gmail.com"
      type="cite">Thomas:
      <div>Do you mean create a mpiaij matrix of order 2nx2n from two
        local nxn matrices?</div>
    </blockquote>
    Yes.<br>
    <blockquote
cite="mid:CAGCphBsHwed=ga+KvjrwXz5BnNFxXNw8J_U2e16=GaRDw13Yug@mail.gmail.com"
      type="cite">
      <div>You might have to use MatSetValues() to do this.</div>
    </blockquote>
    I think, it will be faster and more efficient to create a MatShell
    to wrap around the local matrices. If anybody has some other ideas,
    please let me know.<br>
    <br>
    Thomas<br>
    <br>
    <br>
    <blockquote
cite="mid:CAGCphBsHwed=ga+KvjrwXz5BnNFxXNw8J_U2e16=GaRDw13Yug@mail.gmail.com"
      type="cite">
      <div><br>
      </div>
      <div>Hong<br>
        <div class="gmail_quote">
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div text="#000000" bgcolor="#FFFFFF">
              <div>
                <div class="h5">
                  <blockquote type="cite"> <br>
                    <div class="gmail_quote">Thomas :<br>
                      <blockquote class="gmail_quote" style="margin:0 0
                        0 .8ex;border-left:1px #ccc
                        solid;padding-left:1ex">In my FETI-DP code, each
                        rank creates a SEQAIJ matrix that represents the
                        discretization of the interior domain. Just for
                        debugging, I would like to join these sequential
                        matrices to one global MPIAIJ matrix. This
                        matrix has no off diagonal nnzs and should be
                        stored corresponding to the ranks unknowns,
                        thus, first all rows of the first rank and so
                        on. What's the most efficient way to do this? Is
                        it possible to create this parallel matrix just
                        as a view of the sequential ones, so without
                        copying the data? Thanks for any advise.</blockquote>
                      <div><br>
                      </div>
                      <div><a moz-do-not-send="true"
href="http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Mat/MatCreateMPIAIJConcatenateSeqAIJ.html"
                          target="_blank">http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Mat/MatCreateMPIAIJConcatenateSeqAIJ.html</a></div>
                      <div><br>
                      </div>
                      <div>Note: entries in seqaij matrices are copied
                        into a mpiaij matrix without</div>
                      <div>inter-processor communication. Use petsc-3.3
                        for this function.</div>
                    </div>
                  </blockquote>
                  <br>
                </div>
              </div>
              The function does not do what I expect. For example, if we
              have two mpi task and each contains one local square
              matrix with n rows, I want to create a global square
              matrix with 2n rows. This function create a non-square
              matrix of size 2n x n.<span class="HOEnZb"><font
                  color="#888888"><br>
                  <br>
                  Thomas<br>
                </font></span></div>
          </blockquote>
        </div>
        <br>
      </div>
    </blockquote>
    <br>
  </body>
</html>