<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">On 09/08/2013 03:38 PM, Dave May wrote:<br>
    </div>
    <blockquote
cite="mid:CAJ98EDqLTC5LFeSz+QBa=Z-VjZGWv4SSXwZov-gMNB4i_6oYWQ@mail.gmail.com"
      type="cite">
      <meta http-equiv="Content-Type" content="text/html;
        charset=ISO-8859-1">
      <div dir="ltr">
        <div>
          <div>Remove the option<br>
               -fieldsplit_1_pc_type icc<br>
          </div>
          It is clobbering the LU option Matt suggested you try first (
          -fieldsplit_1_pc_type lu)<br>
        </div>
      </div>
    </blockquote>
    Dear Dave,<br>
    <br>
    I also realized this but and changed that with <br>
    <br>
    -fieldsplit_2_pc_type icc<br>
    <br>
    but thanks for looking at that, need a break I guess ;-)<br>
    <blockquote
cite="mid:CAJ98EDqLTC5LFeSz+QBa=Z-VjZGWv4SSXwZov-gMNB4i_6oYWQ@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div><br>
        </div>
        When debigging such solvers, you should always run with
        -ksp_view to check your solver is configured correctly.<br>
      </div>
    </blockquote>
    Thanks for this<br>
    <blockquote
cite="mid:CAJ98EDqLTC5LFeSz+QBa=Z-VjZGWv4SSXwZov-gMNB4i_6oYWQ@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div><br>
        </div>
      </div>
      <div class="gmail_extra"><br>
        <br>
        <div class="gmail_quote">
          On 8 September 2013 15:32, Umut Tabak <span dir="ltr"><<a
              moz-do-not-send="true" href="mailto:u.tabak@tudelft.nl"
              target="_blank">u.tabak@tudelft.nl</a>></span> wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div bgcolor="#FFFFFF" text="#000000">
              <div class="im">
                <div>On 09/08/2013 03:00 PM, Matthew Knepley wrote:<br>
                </div>
                <blockquote type="cite">
                  <div dir="ltr">
                    <div class="gmail_extra">
                      <div class="gmail_quote"><br>
                        <div><br>
                        </div>
                        <div>If you want to solve hard systems, you must
                          learn to debug a solver yourself. What does
                          "just hangs" mean? Start with a full</div>
                        <div>Schur complement factorization with LU on
                          block 1 and a tiny tolerance on block 2. This
                          will converge in 1 iterate. Then</div>
                        <div>start slowly making the preconditioner
                          weaker.</div>
                      </div>
                    </div>
                  </div>
                </blockquote>
              </div>
              Hi Matt,<br>
              <br>
              Thanks for the suggestion but I did not completely
              understand the options to use for this, I am still
              learning the fieldsplit interface. It was my mistake to
              not to give more in formation on what 'Just hangs' means,
              sorry.<br>
              <br>
              For the previous message: that meant extremely slow
              iterations, the iteration process was not advancing at
              all.<br>
              <br>
              After your reply and looking at the fieldsplit options a
              bit more, I came up with these set of options, now
              iterations are going fast but the decrease in relative
              residual is so slow. But apparently this does not converge
              in 1 iteration, I am still mistaken at some point.<br>
              <br>
              ./ex3 -pc_type fieldsplit -pc_fieldsplit_type schur
              -fieldsplit_1_pc_type lu -fieldsplit_1_ksp_type preonly
              -fieldsplit_1_pc_type icc -fieldsplit_2_ksp_type minres
              -fieldsplit_2_ksp_rtol 1e-8
              -pc_fieldsplit_schur_factorization_type full
              -ksp_monitor_true_residual<br>
              <br>
              Maybe you can help me at this point again.<br>
              <br>
              Thanks.<span class="HOEnZb"><font color="#888888"><br>
                  Umut</font></span>
              <div>
                <div class="h5"><br>
                   <br>
                  <blockquote type="cite">
                    <div dir="ltr">
                      <div class="gmail_extra">
                        <div class="gmail_quote">
                          <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"> The system is
                            a symmetric system in this case given as<br>
                            <br>
                            A =<br>
                            A11 C<br>
                            C^T A22<br>
                            <br>
                            C is of size 1490X2432<br>
                            <br>
                            Could you please comment on this point for
                            me, do I still have an error on the problem
                            setup or on the usage of the options?<br>
                            <br>
                            BR,<br>
                            Umut
                            <div>
                              <div><br>
                                <blockquote class="gmail_quote"
                                  style="margin:0 0 0
                                  .8ex;border-left:1px #ccc
                                  solid;padding-left:1ex"> <br>
                                   
 ISCreateBlock(PETSC_COMM_SELF,bs,n1,inputindices,PETSC_COPY_VALUES,&is_row1);<br>
                                   
 ISCreateBlock(PETSC_COMM_SELF,bs,n2,inputindices,PETSC_COPY_VALUES,&is_row2);<br>
                                  <br>
                                  We could check for this case, but it
                                  would require a little code and<br>
                                  isn't currently done.  This is
                                  probably what you wanted:<br>
                                  <br>
                                  diff --git i/ex2.c w/ex2.c<br>
                                  index 94cea2c..0cce3e6 100644<br>
                                  --- i/ex2.c<br>
                                  +++ w/ex2.c<br>
                                  @@ -45,7 +45,7 @@ int main(int
                                  argc,char **args)<br>
                                      PetscInt bs = 2, n1=1, n2=1,
                                  inputindices[]={0,1}; /* block size is
                                  2 */<br>
                                           
ISCreateBlock(PETSC_COMM_SELF,bs,n1,inputindices,PETSC_COPY_VALUES,&is_row1);<br>
                                  -
 ISCreateBlock(PETSC_COMM_SELF,bs,n2,inputindices,PETSC_COPY_VALUES,&is_row2);<br>
                                  +
 ISCreateBlock(PETSC_COMM_SELF,bs,n2,inputindices+n1,PETSC_COPY_VALUES,&is_row2);<br>
                                       
                                  ISView(is_row1,PETSC_VIEWER_STDOUT_SELF);<br>
                                     
                                  ISView(is_row2,PETSC_VIEWER_STDOUT_SELF);<br>
                                </blockquote>
                                <br>
                              </div>
                            </div>
                          </blockquote>
                        </div>
                        <br>
                        <br clear="all">
                        <div><br>
                        </div>
                        -- <br>
                        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>
                  </blockquote>
                  <br>
                </div>
              </div>
            </div>
          </blockquote>
        </div>
        <br>
      </div>
    </blockquote>
    <br>
  </body>
</html>