<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body 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<br>
<br>
<div class="moz-cite-prefix">On 15-12-07 10:42 AM, Hong wrote:<br>
</div>
<blockquote
cite="mid:CAGCphBv6ahBz-cBfyXLkuwFsMTMmJpL8RBW7DRGNstr0vL=Hww@mail.gmail.com"
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>
</body>
</html>