<div>Barry,</div>
<div> </div>
<div>We have tried tiny tol and lu but the problem is still there. It doesn't make sense to us that this problem is expected and common in iterative solvers instead we think this might be caused by some mistakes we made when we setup the ksp, pc or even the linear system. </div>
<div> </div>
<div> </div>
<div> </div>
<div> </div>
<div> </div>
<div>On Sat, Oct 27, 2012 at 6:07 PM, Barry Smith <span dir="ltr"><<a href="mailto:bsmith@mcs.anl.gov" target="_blank">bsmith@mcs.anl.gov</a>></span> wrote:<br></div>
<div class="gmail_quote">
<blockquote style="BORDER-LEFT:#ccc 1px solid;MARGIN:0px 0px 0px 0.8ex;PADDING-LEFT:1ex" class="gmail_quote"><br><br> PETSc, by default, using iterative solvers which only solve the problem within a certain accuracy. You can control this accuracy by using -ksp_rtol tol where for example you could pick tol = 1.0e-12 You can also use a direct solver<br>
-pc_type lu Run with -ksp_monitor_true_residual -ksp_converged_reason to see if the linear solver is solving to the requested accuracy.<br><span class="HOEnZb"><font color="#888888"><br> Barry<br></font></span>
<div class="HOEnZb">
<div class="h5"><br>On Oct 27, 2012, at 3:46 PM, Chung-Kan Huang <<a href="mailto:ckhuangf@gmail.com">ckhuangf@gmail.com</a>> wrote:<br><br>> Hi,<br>><br>> I am solving a linear system A * X = B that sometime contains A_{I,I} != 0 A_{I,J} = 0. where I!=J and B_I = 0. so I would expect X_i to be perfect zero not some tiny small value. The same linear system is solved by other solvers but they gave zero for aforementioned situation even the system is fair large to our applications. I have tried several configurations in PETSc hoping that problem can be solved but all attempts failed. I wonder if someone can give me some pointers that help to solve this issue. Thanks,<br>
><br>> Chung-Kan<br>><br>><br>><br>> --<br>> Cheers<br>><br>><br><br></div></div></blockquote></div><br><br clear="all"><br>-- <br>
<p><strong>Cheers</strong></p><br>