<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Hi Hong,<br>
    <br>
    Thanks for checking this. A mechanical model was added at the time
    when the solver failed, causing some problem. We need to improve
    this part in the code.<br>
    <br>
    Thanks again and best wishes,<br>
    <br>
    Danyang<br>
    <br>
    <div class="moz-cite-prefix">On 15-12-08 08:10 PM, Hong wrote:<br>
    </div>
    <blockquote
cite="mid:CAGCphBvHgFR=p2vpouee2Bv4tYsz3to16Ka-KgsJYiQvGBDzEQ@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div class="gmail_extra">
          <div class="gmail_quote">Danyang :</div>
          <div class="gmail_quote">Your matrices are ill-conditioned,
            numerically singular with </div>
          <div class="gmail_quote"><span style="font-size:12.8px">Recip.
              condition number = 6.000846e-16</span></div>
          <div class="gmail_quote"><span style="font-size:12.8px">Recip.
              condition number = 2.256434e-27</span><span
              style="font-size:12.8px"><br>
            </span></div>
          <div class="gmail_quote"><span style="font-size:12.8px">Recip.
              condition number = 1.256452e-18</span><span
              style="font-size:12.8px"><br>
            </span></div>
          <div class="gmail_quote"><span style="font-size:12.8px">i.e.,
              condition numbers = O(1.e16 - 1.e27), there is no accuracy
              in computed solution.</span></div>
          <div class="gmail_quote"><span style="font-size:12.8px"><br>
            </span></div>
          <div class="gmail_quote"><span style="font-size:12.8px">I
              checked your </span><span style="font-size:12.8px">matrix
               168 - 172, got </span><span style="font-size:12.8px">Recip.
              condition number = 1.548816e-12.</span></div>
          <div class="gmail_quote"><br>
          </div>
          <div class="gmail_quote"><span style="font-size:12.8px">You
              need check your model to understand why the matrices are
              so ill-conditioned.</span></div>
          <div class="gmail_quote"><br>
          </div>
          <div class="gmail_quote"><span style="font-size:12.8px">Hong</span></div>
          <div class="gmail_quote"><span style="font-size:12.8px"><br>
            </span>
            <blockquote class="gmail_quote" style="margin:0px 0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
              <div text="#000000" bgcolor="#FFFFFF"> Hi Hong,<br>
                <br>
                Sorry to bother you again. The modified code works much
                better than before using both superlu or mumps. However,
                it still encounters failure. The case is similar with
                the previous one, ill-conditioned matrices.<br>
                <br>
                The code crashed after a long time simulation if I use
                superlu_dist, but will not fail if use superlu. I
                restart the simulation before the time it crashes and
                can reproduce the following error<br>
                <br>
                 timestep:      22 time: 1.750E+04 years   delt:
                2.500E+00 years   iter:  1 max.sia: 5.053E-03 tol.sia:
                1.000E-01<br>
                 Newton Iteration Convergence Summary:<br>
                 Newton       maximum      maximum     solver<br>
                 iteration    updatePa  updateTemp    residual   
                iterations  maxvolpa   maxvoltemp   nexvolpa  
                nexvoltemp<br>
                     1       0.1531E+08 0.1755E+04   0.6920E-05       
                1     5585          4402        5814        5814<br>
                <br>
                *** Error in `../program_test': malloc(): memory
                corruption: 0x0000000003a70d50 ***<br>
                Program received signal SIGABRT: Process abort signal.<br>
                Backtrace for this error:<br>
                <br>
                The solver failed at timestep 22, Newton iteration 2. I
                exported the matrices at timestep 1 (matrix 1) and
                timestep 22 (matrix 140 and 141). Matrix 141 is where it
                failed.  The three matrices here are not ill-conditioned
                form the estimated value.<br>
                <br>
                I did the same using the new modified ex52f code and
                found pretty different results for matrix 141. The norm
                by superlu is much acceptable than superlu_dist. In this
                test, memory corruption was not detected. The codes and
                example data can be download from the link below.<br>
                <br>
                <a moz-do-not-send="true"
href="https://www.dropbox.com/s/i1ls0bg0vt7gu0v/petsc-superlu-test2.tar.gz?dl=0"
                  target="_blank">https://www.dropbox.com/s/i1ls0bg0vt7gu0v/petsc-superlu-test2.tar.gz?dl=0</a><br>
                <br>
                <br>
                ****************More test on
                matrix_and_rhs_bin2*******************<br>
                mpiexec.hydra -n 1 ./ex52f -f0
                ./matrix_and_rhs_bin2/a_flow_check_1.bin -rhs
                ./matrix_and_rhs_bin2/b_flow_check_1.bin -loop_matrices
                flow_check -loop_folder ./matrix_and_rhs_bin2
                -matrix_index_start 140 -matrix_index_end 141  -pc_type
                lu -pc_factor_mat_solver_package superlu
                -ksp_monitor_true_residual -mat_superlu_conditionnumber<span><br>
                   -->loac matrix a<br>
                   -->load rhs b<br>
                   size l,m,n,mm       90000       90000      
                  90000       90000<br>
                </span>   Recip. condition number = 6.000846e-16<br>
                  0 KSP preconditioned resid norm 1.146871454377e+08
                true resid norm 4.711091037809e+03 ||r(i)||/||b||
                1.000000000000e+00<br>
                  1 KSP preconditioned resid norm 2.071118508260e-06
                true resid norm 3.363767171515e-08 ||r(i)||/||b||
                7.140102249181e-12<br>
                Norm of error  3.3638E-08 iterations     1<br>
                 -->Test for matrix          140<br>
                  Recip. condition number = 2.256434e-27<br>
                  0 KSP preconditioned resid norm 2.084372893355e+14
                true resid norm 4.711091037809e+03 ||r(i)||/||b||
                1.000000000000e+00<br>
                  1 KSP preconditioned resid norm 4.689629276419e+00
                true resid norm 1.037236635337e-01 ||r(i)||/||b||
                2.201690918330e-05<br>
                Norm of error  1.0372E-01 iterations     1<br>
                 -->Test for matrix          141<br>
                  Recip. condition number = 1.256452e-18<br>
                  0 KSP preconditioned resid norm 1.055488964519e+08
                true resid norm 4.711091037809e+03 ||r(i)||/||b||
                1.000000000000e+00<br>
                  1 KSP preconditioned resid norm 2.998827511681e-04
                true resid norm 4.805214542776e-04 ||r(i)||/||b||
                1.019979130994e-07<br>
                Norm of error  4.8052E-04 iterations     1<span><br>
                   --> End of test, bye<br>
                  <br>
                  <br>
                </span> mpiexec.hydra -n 1 ./ex52f -f0
                ./matrix_and_rhs_bin2/a_flow_check_1.bin -rhs
                ./matrix_and_rhs_bin2/b_flow_check_1.bin -loop_matrices
                flow_check -loop_folder ./matrix_and_rhs_bin2
                -matrix_index_start 140 -matrix_index_end 141  -pc_type
                lu -pc_factor_mat_solver_package superlu_dist<span><br>
                   -->loac matrix a<br>
                   -->load rhs b<br>
                   size l,m,n,mm       90000       90000      
                  90000       90000<br>
                </span> Norm of error  3.6752E-08 iterations     1<br>
                 -->Test for matrix          140<br>
                Norm of error  1.6335E-01 iterations     1<br>
                 -->Test for matrix          141<br>
                Norm of error  3.4345E+01 iterations     1<span><br>
                   --> End of test, bye<br>
                  <br>
                </span> Thanks,<span><br>
                  <br>
                  Danyang<br>
                  <br>
                  <div>On 15-12-07 12:01 PM, Hong wrote:<br>
                  </div>
                </span>
                <div>
                  <div>
                    <blockquote type="cite">
                      <div dir="ltr">
                        <div class="gmail_extra">
                          <div class="gmail_quote">Danyang:</div>
                          <div class="gmail_quote">Add 'call
                            MatSetFromOptions(A,ierr)' to your code.</div>
                          <div class="gmail_quote">Attached below is
                            ex52f.F modified from your ex52f.F to be
                            compatible with petsc-dev.</div>
                          <div class="gmail_quote"><br>
                          </div>
                          <div class="gmail_quote">Hong</div>
                          <div class="gmail_quote"><br>
                            <blockquote class="gmail_quote"
                              style="margin:0px 0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
                              <div text="#000000" bgcolor="#FFFFFF">
                                Hello Hong,<br>
                                <br>
                                Thanks for the quick reply and the
                                option "-mat_superlu_dist_fact
                                SamePattern" works like a charm, if I
                                use this option from the command line. <br>
                                <br>
                                How can I add this option as the
                                default. I tried using
                                PetscOptionsInsertString("-mat_superlu_dist_fact
                                SamePattern",ierr) in my code but this
                                does not work. <br>
                                <br>
                                Thanks,<br>
                                <br>
                                Danyang
                                <div>
                                  <div><br>
                                    <br>
                                    <div>On 15-12-07 10:42 AM, Hong
                                      wrote:<br>
                                    </div>
                                    <blockquote type="cite">
                                      <div dir="ltr">
                                        <div class="gmail_extra">
                                          <div class="gmail_quote">Danyang
                                            :</div>
                                          <div class="gmail_quote"><br>
                                          </div>
                                          <div class="gmail_quote">Adding

                                            '-mat_superlu_dist_fact
                                            SamePattern' fixed the
                                            problem. Below is how I
                                            figured it out.</div>
                                          <div class="gmail_quote"><br>
                                          </div>
                                          <div class="gmail_quote">
                                            <div class="gmail_quote">1.
                                              Reading ex52f.F, I see
                                              '-superlu_default' =
                                              '-pc_factor_mat_solver_package
                                              superlu_dist', the later
                                              enables runtime options
                                              for other packages. I use
                                              superlu_dist-4.2 and
                                              superlu-4.1 for the tests
                                              below.</div>
                                            <div class="gmail_quote"><br>
                                            </div>
                                            <div class="gmail_quote">2.
                                              Use the Matrix 168 to
                                              setup KSP solver and
                                              factorization, all
                                              packages, petsc,
                                              superlu_dist and mumps
                                              give same correct results:</div>
                                            <div class="gmail_quote"><br>
                                            </div>
                                            <div class="gmail_quote">./ex52f
                                              -f0
                                              matrix_and_rhs_bin/a_flow_check_168.bin
                                              -rhs
                                              matrix_and_rhs_bin/b_flow_check_168.bin
                                              -loop_matrices flow_check
                                              -loop_folder
                                              matrix_and_rhs_bin
                                              -pc_type lu
                                              -pc_factor_mat_solver_package
                                              petsc</div>
                                            <div class="gmail_quote"> -->loac
                                              matrix a</div>
                                            <div class="gmail_quote"> -->load
                                              rhs b</div>
                                            <div class="gmail_quote"> size
                                              l,m,n,mm       90000      
                                              90000       90000      
                                              90000</div>
                                            <div class="gmail_quote">Norm
                                              of error  7.7308E-11
                                              iterations     1</div>
                                            <div class="gmail_quote"> -->Test
                                              for matrix          168</div>
                                            <div class="gmail_quote">..</div>
                                            <div class="gmail_quote"> -->Test
                                              for matrix          172</div>
                                            <div class="gmail_quote">Norm
                                              of error  3.8461E-11
                                              iterations     1</div>
                                            <div class="gmail_quote"><br>
                                            </div>
                                            <div class="gmail_quote">./ex52f
                                              -f0
                                              matrix_and_rhs_bin/a_flow_check_168.bin
                                              -rhs
                                              matrix_and_rhs_bin/b_flow_check_168.bin
                                              -loop_matrices flow_check
                                              -loop_folder
                                              matrix_and_rhs_bin
                                              -pc_type lu
                                              -pc_factor_mat_solver_package
                                              superlu_dist</div>
                                            <div class="gmail_quote">Norm
                                              of error  9.4073E-11
                                              iterations     1</div>
                                            <div class="gmail_quote"> -->Test
                                              for matrix          168</div>
                                            <div class="gmail_quote">...</div>
                                            <div class="gmail_quote"> -->Test
                                              for matrix          172</div>
                                            <div class="gmail_quote">Norm
                                              of error  3.8187E-11
                                              iterations     1</div>
                                            <div class="gmail_quote"><br>
                                            </div>
                                            <div class="gmail_quote">
                                              <div class="gmail_quote">3.
                                                Use superlu, I get</div>
                                              <div class="gmail_quote">./ex52f
                                                -f0
                                                matrix_and_rhs_bin/a_flow_check_168.bin
                                                -rhs
                                                matrix_and_rhs_bin/b_flow_check_168.bin
                                                -loop_matrices
                                                flow_check -loop_folder
                                                matrix_and_rhs_bin
                                                -pc_type lu
                                                -pc_factor_mat_solver_package
                                                superlu</div>
                                              <div class="gmail_quote">Norm
                                                of error  1.0191E-06
                                                iterations     1</div>
                                              <div class="gmail_quote"> -->Test
                                                for matrix          168</div>
                                              <div class="gmail_quote">...</div>
                                              <div class="gmail_quote"> -->Test
                                                for matrix          172</div>
                                              <div class="gmail_quote">Norm
                                                of error  9.7858E-07
                                                iterations     1</div>
                                              <div class="gmail_quote"><br>
                                              </div>
                                              <div class="gmail_quote">Replacing
                                                default DiagPivotThresh:
                                                1. to 0.0, I get same
                                                solutions as other
                                                packages:</div>
                                              <div class="gmail_quote"><br>
                                              </div>
                                              <div class="gmail_quote">./ex52f
                                                -f0
                                                matrix_and_rhs_bin/a_flow_check_168.bin
                                                -rhs
                                                matrix_and_rhs_bin/b_flow_check_168.bin
                                                -loop_matrices
                                                flow_check -loop_folder
                                                matrix_and_rhs_bin
                                                -pc_type lu
                                                -pc_factor_mat_solver_package
                                                superlu
                                                -mat_superlu_diagpivotthresh
                                                0.0</div>
                                              <div class="gmail_quote"><br>
                                              </div>
                                              <div class="gmail_quote">Norm
                                                of error  8.3614E-11
                                                iterations     1</div>
                                              <div class="gmail_quote"> -->Test
                                                for matrix          168</div>
                                              <div class="gmail_quote">...</div>
                                              <div class="gmail_quote"> -->Test
                                                for matrix          172</div>
                                              <div class="gmail_quote">Norm
                                                of error  3.7098E-11
                                                iterations     1</div>
                                              <div class="gmail_quote"><br>
                                              </div>
                                              <div class="gmail_quote">4.</div>
                                              <div class="gmail_quote">using
                                                '-mat_view
                                                ascii::ascii_info', I
                                                found that
                                                a_flow_check_1.bin and
                                                a_flow_check_168.bin
                                                seem have same
                                                structure:</div>
                                              <div class="gmail_quote"><br>
                                              </div>
                                              <div class="gmail_quote"> -->loac

                                                matrix a</div>
                                              <div class="gmail_quote">Mat
                                                Object: 1 MPI processes</div>
                                              <div class="gmail_quote"> 
                                                type: seqaij</div>
                                              <div class="gmail_quote"> 
                                                rows=90000, cols=90000</div>
                                              <div class="gmail_quote"> 
                                                total: nonzeros=895600,
                                                allocated
                                                nonzeros=895600</div>
                                              <div class="gmail_quote"> 
                                                total number of mallocs
                                                used during MatSetValues
                                                calls =0</div>
                                              <div class="gmail_quote"> 
                                                  using I-node routines:
                                                found 45000 nodes, limit
                                                used is 5</div>
                                              <div class="gmail_quote"><br>
                                              </div>
                                              <div class="gmail_quote">
                                                <div class="gmail_quote">5.</div>
                                                <div class="gmail_quote">Using

                                                  a_flow_check_1.bin, I
                                                  am able to reproduce
                                                  the error you
                                                  reported: all packages
                                                  give correct results
                                                  except superlu_dist:</div>
                                                <div class="gmail_quote">./ex52f
                                                  -f0
                                                  matrix_and_rhs_bin/a_flow_check_1.bin
                                                  -rhs
                                                  matrix_and_rhs_bin/b_flow_check_168.bin
                                                  -loop_matrices
                                                  flow_check
                                                  -loop_folder
                                                  matrix_and_rhs_bin
                                                  -pc_type lu
                                                  -pc_factor_mat_solver_package
                                                  superlu_dist</div>
                                                <div class="gmail_quote">Norm
                                                  of error  2.5970E-12
                                                  iterations     1</div>
                                                <div class="gmail_quote"> -->Test
                                                  for matrix        
                                                   168</div>
                                                <div class="gmail_quote">Norm
                                                  of error  1.3936E-01
                                                  iterations    34</div>
                                                <div class="gmail_quote"> -->Test
                                                  for matrix        
                                                   169</div>
                                                <div class="gmail_quote"><br>
                                                </div>
                                                <div class="gmail_quote">I
                                                  guess the error might
                                                  come from reuse of
                                                  matrix factor.
                                                  Replacing default</div>
                                                <div class="gmail_quote">-mat_superlu_dist_fact


                                                  <SamePattern_SameRowPerm>
                                                  with</div>
                                                <div class="gmail_quote">-mat_superlu_dist_fact


                                                  SamePattern, I get</div>
                                                <div class="gmail_quote"><br>
                                                </div>
                                                <div class="gmail_quote">./ex52f
                                                  -f0
                                                  matrix_and_rhs_bin/a_flow_check_1.bin
                                                  -rhs
                                                  matrix_and_rhs_bin/b_flow_check_168.bin
                                                  -loop_matrices
                                                  flow_check
                                                  -loop_folder
                                                  matrix_and_rhs_bin
                                                  -pc_type lu
                                                  -pc_factor_mat_solver_package
                                                  superlu_dist
                                                  -mat_superlu_dist_fact
                                                  SamePattern</div>
                                                <div class="gmail_quote"><br>
                                                </div>
                                                <div class="gmail_quote">Norm
                                                  of error  2.5970E-12
                                                  iterations     1</div>
                                                <div class="gmail_quote"> -->Test
                                                  for matrix        
                                                   168</div>
                                                <div class="gmail_quote">Norm
                                                  of error  9.4073E-11
                                                  iterations     1</div>
                                                <div class="gmail_quote"> -->Test
                                                  for matrix        
                                                   169</div>
                                                <div class="gmail_quote">Norm
                                                  of error  6.4303E-11
                                                  iterations     1</div>
                                                <div class="gmail_quote"> -->Test
                                                  for matrix        
                                                   170</div>
                                                <div class="gmail_quote">Norm
                                                  of error  7.4327E-11
                                                  iterations     1</div>
                                                <div class="gmail_quote"> -->Test
                                                  for matrix        
                                                   171</div>
                                                <div class="gmail_quote">Norm
                                                  of error  5.4162E-11
                                                  iterations     1</div>
                                                <div class="gmail_quote"> -->Test
                                                  for matrix        
                                                   172</div>
                                                <div class="gmail_quote">Norm
                                                  of error  3.4440E-11
                                                  iterations     1</div>
                                                <div class="gmail_quote"> -->
                                                  End of test, bye</div>
                                                <div class="gmail_quote"><br>
                                                </div>
                                                <div class="gmail_quote">Sherry
                                                  may tell you why
                                                  SamePattern_SameRowPerm
                                                  cause the difference
                                                  here.</div>
                                                <div class="gmail_quote">Best
                                                  on the above
                                                  experiments, I would
                                                  set following as
                                                  default</div>
                                                <div class="gmail_quote">'-mat_superlu_diagpivotthresh


                                                  0.0' in petsc/superlu
                                                  interface.</div>
                                                <div class="gmail_quote">'-mat_superlu_dist_fact


                                                  SamePattern' in
                                                  petsc/superlu_dist
                                                  interface.</div>
                                                <div class="gmail_quote"><br>
                                                </div>
                                                <div class="gmail_quote">Hong</div>
                                              </div>
                                            </div>
                                          </div>
                                          <div class="gmail_quote"><br>
                                            <blockquote
                                              class="gmail_quote"
                                              style="margin:0px 0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
                                              <div bgcolor="#FFFFFF"
                                                text="#000000"> Hi Hong,<br>
                                                <br>
                                                I did more test today
                                                and finally found that
                                                the solution accuracy
                                                depends on the initial
                                                (first) matrix quality.
                                                I modified the ex52f.F
                                                to do the test. There
                                                are 6 matrices and
                                                right-hand-side vectors.
                                                All these matrices and
                                                rhs are from my reactive
                                                transport simulation.
                                                Results will be quite
                                                different depending on
                                                which one you use to do
                                                factorization. Results
                                                will also be different
                                                if you run with
                                                different options. My
                                                code is similar to the
                                                First or the Second test
                                                below. When the matrix
                                                is well conditioned, it
                                                works fine. But if the
                                                initial matrix is well
                                                conditioned, it likely
                                                to crash when the matrix
                                                become ill-conditioned.
                                                Since most of my case
                                                are well conditioned so
                                                I didn't detect the
                                                problem before. This
                                                case is a special one. <br>
                                                <br>
                                                <br>
                                                How can I avoid this
                                                problem? Shall I redo
                                                factorization? Can PETSc
                                                automatically detect
                                                this prolbem or is there
                                                any option available to
                                                do this?<br>
                                                <br>
                                                All the data and test
                                                code (modified ex52f)
                                                can be found via the
                                                dropbox link below. <br>
                                                <u><br>
                                                </u><u><a
                                                    moz-do-not-send="true"
href="https://www.dropbox.com/s/4al1a60creogd8m/petsc-superlu-test.tar.gz?dl=0"
                                                    target="_blank">https://www.dropbox.com/s/4al1a60creogd8m/petsc-superlu-test.tar.gz?dl=0</a></u><br>
                                                <br>
                                                <br>
                                                Summary of my test is
                                                shown below. <br>
                                                <br>
                                                First, use the Matrix 1
                                                to setup KSP solver and
                                                factorization, then
                                                solve 168 to 172<br>
                                                <br>
                                                mpiexec.hydra -n 1
                                                ./ex52f -f0
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/a_flow_check_1.bin
                                                -rhs
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/b_flow_check_1.bin
                                                -loop_matrices
                                                flow_check -loop_folder
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin

                                                -pc_type lu
                                                -pc_factor_mat_solver_package
                                                superlu_dist<br>
                                                <br>
                                                Norm of error 
                                                3.8815E-11
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          168<br>
                                                Norm of error 
                                                4.2307E-01 iterations   
                                                32<br>
                                                 -->Test for
                                                matrix          169<br>
                                                Norm of error 
                                                3.0528E-01 iterations   
                                                32<br>
                                                 -->Test for
                                                matrix          170<br>
                                                Norm of error 
                                                3.1177E-01 iterations   
                                                32<br>
                                                 -->Test for
                                                matrix          171<br>
                                                Norm of error 
                                                3.2793E-01 iterations   
                                                32<br>
                                                 -->Test for
                                                matrix          172<br>
                                                Norm of error 
                                                3.1251E-01 iterations   
                                                31<br>
                                                <br>
                                                Second, use the Matrix 1
                                                to setup KSP solver and
                                                factorization using the
                                                implemented SuperLU
                                                relative codes. I
                                                thought this will
                                                generate the same
                                                results as the First
                                                test, but it actually
                                                not.<br>
                                                <br>
                                                mpiexec.hydra -n 1
                                                ./ex52f -f0
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/a_flow_check_1.bin
                                                -rhs
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/b_flow_check_1.bin
                                                -loop_matrices
                                                flow_check -loop_folder
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin

                                                -superlu_default<br>
                                                <br>
                                                Norm of error 
                                                2.2632E-12
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          168<br>
                                                Norm of error 
                                                1.0817E+04
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          169<br>
                                                Norm of error 
                                                1.0786E+04
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          170<br>
                                                Norm of error 
                                                1.0792E+04
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          171<br>
                                                Norm of error 
                                                1.0792E+04
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          172<br>
                                                Norm of error 
                                                1.0792E+04
                                                iterations     1<br>
                                                <br>
                                                <br>
                                                Third, use the Matrix
                                                168 to setup KSP solver
                                                and factorization, then
                                                solve 168 to 172<br>
                                                <br>
                                                mpiexec.hydra -n 1
                                                ./ex52f -f0
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/a_flow_check_168.bin



                                                -rhs
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/b_flow_check_168.bin



                                                -loop_matrices
                                                flow_check -loop_folder
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin

                                                -pc_type lu
                                                -pc_factor_mat_solver_package
                                                superlu_dist<br>
                                                <br>
                                                Norm of error 
                                                9.5528E-10
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          168<br>
                                                Norm of error 
                                                9.4945E-10
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          169<br>
                                                Norm of error 
                                                6.4279E-10
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          170<br>
                                                Norm of error 
                                                7.4633E-10
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          171<br>
                                                Norm of error 
                                                7.4863E-10
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          172<br>
                                                Norm of error 
                                                8.9701E-10
                                                iterations     1<br>
                                                <br>
                                                Fourth, use the Matrix
                                                168 to setup KSP solver
                                                and factorization using
                                                the implemented SuperLU
                                                relative codes. I
                                                thought this will
                                                generate the same
                                                results as the Third
                                                test, but it actually
                                                not.<br>
                                                <br>
                                                mpiexec.hydra -n 1
                                                ./ex52f -f0
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/a_flow_check_168.bin



                                                -rhs
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin/b_flow_check_168.bin



                                                -loop_matrices
                                                flow_check -loop_folder
                                                /home/dsu/work/petsc-superlu-test/matrix_and_rhs_bin

                                                -superlu_default<br>
                                                <br>
                                                Norm of error 
                                                3.7017E-11
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          168<br>
                                                Norm of error 
                                                3.6420E-11
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          169<br>
                                                Norm of error 
                                                3.7184E-11
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          170<br>
                                                Norm of error 
                                                3.6847E-11
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          171<br>
                                                Norm of error 
                                                3.7883E-11
                                                iterations     1<br>
                                                 -->Test for
                                                matrix          172<br>
                                                Norm of error 
                                                3.8805E-11
                                                iterations     1<br>
                                                <br>
                                                Thanks very much,<br>
                                                <br>
                                                Danyang<br>
                                                <br>
                                                <div>On 15-12-03 01:59
                                                  PM, Hong wrote:<br>
                                                </div>
                                                <blockquote type="cite">
                                                  <div dir="ltr">
                                                    <div
                                                      class="gmail_extra">
                                                      <div
                                                        class="gmail_quote">Danyang
                                                        :</div>
                                                      <div
                                                        class="gmail_quote">Further

testing a_flow_check_168.bin,</div>
                                                      <div
                                                        class="gmail_quote">
                                                        <div
                                                          class="gmail_quote">./ex10

                                                          -f0
                                                          /Users/Hong/Downloads/matrix_and_rhs_bin/a_flow_check_168.bin
                                                          -rhs
                                                          /Users/Hong/Downloads/matrix_and_rhs_bin/x_flow_check_168.bin
                                                          -pc_type lu
                                                          -pc_factor_mat_solver_package
                                                          superlu
                                                          -ksp_monitor_true_residual
-mat_superlu_conditionnumber</div>
                                                        <div
                                                          class="gmail_quote"> 
                                                          Recip.
                                                          condition
                                                          number =
                                                          1.610480e-12</div>
                                                        <div
                                                          class="gmail_quote"> 
                                                          0 KSP
                                                          preconditioned
                                                          resid norm
                                                          6.873340313547e+09
                                                          true resid
                                                          norm
                                                          7.295020990196e+03
                                                          ||r(i)||/||b||
1.000000000000e+00</div>
                                                        <div
                                                          class="gmail_quote"> 
                                                          1 KSP
                                                          preconditioned
                                                          resid norm
                                                          2.051833296449e-02
                                                          true resid
                                                          norm
                                                          2.976859070118e-02
                                                          ||r(i)||/||b||
4.080672384793e-06</div>
                                                        <div
                                                          class="gmail_quote">Number

                                                          of iterations
                                                          =   1</div>
                                                        <div
                                                          class="gmail_quote">Residual

                                                          norm 0.0297686</div>
                                                        <div
                                                          class="gmail_quote"><br>
                                                        </div>
                                                        <div
                                                          class="gmail_quote">condition

                                                          number of this
                                                          matrix =
                                                          1/1.610480e-12
                                                          = 1.e+12,</div>
                                                        <div
                                                          class="gmail_quote">i.e.,

                                                          this matrix is
ill-conditioned. </div>
                                                        <div
                                                          class="gmail_quote"><br>
                                                        </div>
                                                        <div
                                                          class="gmail_quote">Hong</div>
                                                        <div
                                                          class="gmail_quote"><br>
                                                        </div>
                                                        <div
                                                          class="gmail_quote"><br>
                                                        </div>
                                                        <blockquote
                                                          class="gmail_quote"
                                                          style="margin:0px
                                                          0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
                                                          <div
                                                          text="#000000"
bgcolor="#FFFFFF"> Hi Hong,<br>
                                                          <br>
                                                          The binary
                                                          format of
                                                          matrix, rhs
                                                          and solution
                                                          can be
                                                          downloaded via
                                                          the link
                                                          below.<br>
                                                          <br>
                                                          <a
                                                          moz-do-not-send="true"
href="https://www.dropbox.com/s/cl3gfi0s0kjlktf/matrix_and_rhs_bin.tar.gz?dl=0"
target="_blank">https://www.dropbox.com/s/cl3gfi0s0kjlktf/matrix_and_rhs_bin.tar.gz?dl=0</a><br>
                                                          <br>
                                                          Thanks,<br>
                                                          <br>
                                                          Danyang
                                                          <div>
                                                          <div><br>
                                                          <br>
                                                          On 15-12-03
                                                          10:50 AM, Hong
                                                          wrote:<br>
                                                          <blockquote
                                                          type="cite">
                                                          <div dir="ltr">
                                                          <div
                                                          class="gmail_extra">
                                                          <div
                                                          class="gmail_quote">Danyang:<br>
                                                          <blockquote
                                                          class="gmail_quote"
                                                          style="margin:0px
                                                          0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
                                                          <div
                                                          text="#000000"
bgcolor="#FFFFFF"><br>
                                                          <br>
                                                          To my
                                                          surprising,
                                                          solutions from
                                                          SuperLU at
                                                          timestep 29
                                                          seems not
                                                          correct for
                                                          the first 4
                                                          Newton
                                                          iterations,
                                                          but the
                                                          solutions from
                                                          iteration
                                                          solver and
                                                          MUMPS are
                                                          correct. <br>
                                                          <br>
                                                          Please find
                                                          all the
                                                          matrices, rhs
                                                          and solutions
                                                          at timestep 29
                                                          via the link
                                                          below. The
                                                          data is a bit
                                                          large so that
                                                          I just share
                                                          it through
                                                          Dropbox. A
                                                          piece of
                                                          matlab code to
                                                          read these
                                                          data and then
                                                          computer the
                                                          norm has also
                                                          been attached.
                                                          <br>
                                                          <u><a
                                                          moz-do-not-send="true"
href="https://www.dropbox.com/s/rr8ueysgflmxs7h/results-check.tar.gz?dl=0"
target="_blank">https://www.dropbox.com/s/rr8ueysgflmxs7h/results-check.tar.gz?dl=0</a></u></div>
                                                          </blockquote>
                                                          <div><br>
                                                          </div>
                                                          <div>Can you
                                                          send us matrix
                                                          in petsc
                                                          binary format?</div>
                                                          <div><br>
                                                          </div>
                                                          <div>e.g.,
                                                          call
                                                          MatView(M,
                                                          PETSC_VIEWER_BINARY_(PETSC_COMM_WORLD))</div>
                                                          <div>or
                                                          '-ksp_view_mat
                                                          binary'</div>
                                                          <div><br>
                                                          </div>
                                                          <div>Hong</div>
                                                          <blockquote
                                                          class="gmail_quote"
                                                          style="margin:0px
                                                          0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
                                                          <div
                                                          text="#000000"
bgcolor="#FFFFFF"><br>
                                                          <br>
                                                          Below is a
                                                          summary of the
                                                          norm from the
                                                          three solvers
                                                          at timestep
                                                          29, newton
                                                          iteration 1 to
                                                          5.<br>
                                                          <br>
                                                          Timestep 29<br>
                                                          Norm of
                                                          residual seq
                                                          1.661321e-09,
                                                          superlu
                                                          1.657103e+04,
                                                          mumps
                                                          3.731225e-11 <br>
                                                          Norm of
                                                          residual seq
                                                          1.753079e-09,
                                                          superlu
                                                          6.675467e+02,
                                                          mumps
                                                          1.509919e-13 <br>
                                                          Norm of
                                                          residual seq
                                                          4.914971e-10,
                                                          superlu
                                                          1.236362e-01,
                                                          mumps
                                                          2.139303e-17 <br>
                                                          Norm of
                                                          residual seq
                                                          3.532769e-10,
                                                          superlu
                                                          1.304670e-04,
                                                          mumps
                                                          5.387000e-20 <br>
                                                          Norm of
                                                          residual seq
                                                          3.885629e-10,
                                                          superlu
                                                          2.754876e-07,
                                                          mumps
                                                          4.108675e-21 <br>
                                                          <br>
                                                          Would anybody
                                                          please check
                                                          if SuperLU can
                                                          solve these
                                                          matrices?
                                                          Another
                                                          possibility is
                                                          that something
                                                          is wrong in my
                                                          own code. But
                                                          so far, I
                                                          cannot find
                                                          any problem in
                                                          my code since
                                                          the same code
                                                          works fine if
                                                          I using
                                                          iterative
                                                          solver or
                                                          direct solver
                                                          MUMPS. But for
                                                          other cases I
                                                          have tested, 
                                                          all these
                                                          solvers work
                                                          fine.<br>
                                                          <br>
                                                          Please let me
                                                          know if I did
                                                          not write down
                                                          the problem
                                                          clearly.<br>
                                                          <br>
                                                          Thanks,<br>
                                                          <br>
                                                          Danyang<br>
                                                          <br>
                                                          <br>
                                                          <br>
                                                          </div>
                                                          </blockquote>
                                                          </div>
                                                          <br>
                                                          </div>
                                                          </div>
                                                          </blockquote>
                                                          <br>
                                                          </div>
                                                          </div>
                                                          </div>
                                                        </blockquote>
                                                      </div>
                                                      <br>
                                                    </div>
                                                  </div>
                                                </blockquote>
                                                <br>
                                              </div>
                                            </blockquote>
                                          </div>
                                          <br>
                                        </div>
                                      </div>
                                    </blockquote>
                                    <br>
                                  </div>
                                </div>
                              </div>
                            </blockquote>
                          </div>
                          <br>
                        </div>
                      </div>
                    </blockquote>
                    <br>
                  </div>
                </div>
              </div>
            </blockquote>
          </div>
          <br>
        </div>
      </div>
    </blockquote>
    <br>
  </body>
</html>