<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">Anton:<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div bgcolor="#FFFFFF"><span class="gmail-"><blockquote type="cite">
      
      <div dir="ltr">
        <div class="gmail_extra">
          
          <div class="gmail_quote">If replacing superlu_dist with mumps,
            does your code work?</div>
        </div>
      </div>
    </blockquote></span>
    yes</div></blockquote><div> </div><div>You may use mumps in your code, or tests different options for superlu_dist:</div><div><br></div><div><div>  -mat_superlu_dist_equil: <TRUE> Equilibrate matrix (None)</div><div>  -mat_superlu_dist_rowperm <LargeDiag> Row permutation (choose one of) LargeDiag NATURAL (None)</div><div>  -mat_superlu_dist_colperm <METIS_AT_PLUS_A> Column permutation (choose one of) NATURAL MMD_AT_PLUS_A MMD_ATA METIS_AT_PLUS_A PARMETIS (None)</div><div>  -mat_superlu_dist_replacetinypivot: <FALSE> Replace tiny pivots (None)</div><div>  -mat_superlu_dist_parsymbfact: <FALSE> Parallel symbolic factorization (None)</div><div>  -mat_superlu_dist_fact <SamePattern> Sparsity pattern for repeated matrix factorization (choose one of) SamePattern SamePattern_SameRowPerm (None)</div></div><div><br></div><div>The options inside <> are defaults. You may try others. This might help narrow down the bug.</div><div><br></div><div>Hong</div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div bgcolor="#FFFFFF"><span class="gmail-"><br>
    <blockquote type="cite">
      <div dir="ltr">
        <div class="gmail_extra">
          <div class="gmail_quote">Hong
            <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
              <div bgcolor="#FFFFFF"><span class="gmail-m_-8985211249997465386gmail-">
                  <div class="gmail-m_-8985211249997465386gmail-m_4450561754203495905moz-cite-prefix">On
                    10/24/2016 05:47 PM, Hong wrote:<br>
                  </div>
                  <blockquote type="cite">
                    <div dir="ltr">Barry,
                      <div>Your change indeed fixed the error of his
                        testing code.</div>
                      <div>As Satish tested, on your branch, ex16 runs
                        smooth.</div>
                      <div><br>
                      </div>
                      <div>I do not understand why on maint or master
                        branch, ex16 creases inside superlu_dist, but
                        not with mumps. </div>
                      <div><br>
                      </div>
                    </div>
                  </blockquote>
                  <br>
                </span> I also confirm that ex16 runs fine with latest
                fix, but unfortunately not my code.<br>
                <br>
                This is something to be expected, since my code
                preallocates once in the beginning. So there is no way
                it can be affected by multiple preallocations.
                Subsequently I only do matrix assembly, that makes sure
                structure doesn't change (set to get error otherwise).<br>
                <br>
                Summary: we don't have a simple test code to debug
                superlu issue anymore.<span class="gmail-m_-8985211249997465386gmail-HOEnZb"><font color="#888888"><br>
                    <br>
                    Anton</font></span><span class="gmail-m_-8985211249997465386gmail-"><br>
                  <br>
                  <blockquote type="cite">
                    <div dir="ltr">
                      <div>Hong<br>
                        <div class="gmail_extra"><br>
                          <div class="gmail_quote">On Mon, Oct 24, 2016
                            at 9:34 AM, Satish Balay <span dir="ltr"><<a href="mailto:balay@mcs.anl.gov" target="_blank">balay@mcs.anl.gov</a>></span>
                            wrote:<br>
                            <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><span>On
                                Mon, 24 Oct 2016, Barry Smith wrote:<br>
                                <br>
                                ><br>
                              </span><span>> > [Or perhaps Hong is
                                using a different test code and is
                                observing bugs<br>
                                > > with superlu_dist interface..]<br>
                                ><br>
                                >    She states that her test does a
                                NEW MatCreate() for each matrix load (I
                                cut and pasted it in the email I just
                                sent). The bug I fixed was only related
                                to using the SAME matrix from one
                                MatLoad() in another MatLoad().<br>
                                <br>
                              </span>Ah - ok.. Sorry - wasn't thinking
                              clearly :(<br>
                              <span class="gmail-m_-8985211249997465386gmail-m_4450561754203495905HOEnZb"><font color="#888888"><br>
                                  Satish<br>
                                </font></span></blockquote>
                          </div>
                          <br>
                        </div>
                      </div>
                    </div>
                  </blockquote>
                  <br>
                </span></div>
            </blockquote>
          </div>
          <br>
        </div>
      </div>
    </blockquote>
    <br>
  </span></div>

</blockquote></div><br></div></div>