<html><head></head><body><div style="font-family:garamond, new york, times, serif;font-size:13px;"><div>Thank you Barry for the suggestion, I will definitely look in to TS implementation.<br></div><div>I have implemented my own version of reduced time step size using the <span><span style="font-family:Consolas, Courier, monospace;">SNESGetConvergedReason</span></span>, but encountered the same problem after another 30 time-steps. So, I am experimenting with different set of initial conditions.<br></div><div>Ling, thank you for the code, I will see if my understanding is different from yours.</div><div><br></div><div>--</div><div>Rahul.<br></div><div><br></div><div id="yahoo_quoted_6336599421" class="yahoo_quoted">
<div style="font-family:'Helvetica Neue', Helvetica, Arial, sans-serif;font-size:13px;color:#26282a;">
<div>
On Thursday, August 30, 2018, 11:25:35 PM GMT+5:30, Smith, Barry F. <bsmith@mcs.anl.gov> wrote:
</div>
<div><br></div>
<div><br></div>
<div><div dir="ltr"><br clear="none"><br clear="none">> On Aug 30, 2018, at 12:49 PM, Fande Kong <<a shape="rect" ymailto="mailto:fdkong.jd@gmail.com" href="mailto:fdkong.jd@gmail.com">fdkong.jd@gmail.com</a>> wrote:<br clear="none">> <br clear="none">> Hi Barry,<br clear="none">> <br clear="none">> I haven't had time to look into TS so far. But it is definitely interesting. One simple question would like this: If I have a simple loop for time steppers, and each step SNES is called. How hard to convert my code to use TS? <br clear="none"><br clear="none"> You would need to look at the examples in ts/examples/tutorials but essentially you need to provide a slightly different version of FormFunction() and FormJacobian() and call TSSolve() etc instead of having your own loop.<div class="yqt5895941878" id="yqtfd76634"><br clear="none"><br clear="none"> Barry<br clear="none"><br clear="none">> <br clear="none">> Any suggestion? Where should I start from?<br clear="none">> <br clear="none">> <br clear="none">> Fande,<br clear="none">> <br clear="none">> On Thu, Aug 30, 2018 at 11:29 AM Smith, Barry F. <<a shape="rect" ymailto="mailto:bsmith@mcs.anl.gov" href="mailto:bsmith@mcs.anl.gov">bsmith@mcs.anl.gov</a>> wrote:<br clear="none">> <br clear="none">> Note also that the PETSc TS component has a large variety of timesteppers with automatic adaptivity which adjust the time-step for accuracy and convergence. Depending on the exact needs of your time-stepper it might be easier in the long run to use PETSc's time-steppers rather than write your own.<br clear="none">> <br clear="none">> Barry<br clear="none">> <br clear="none">> <br clear="none">> > On Aug 30, 2018, at 10:46 AM, Ling Zou <<a shape="rect" ymailto="mailto:Ling.Zou@inl.gov" href="mailto:Ling.Zou@inl.gov">Ling.Zou@inl.gov</a>> wrote:<br clear="none">> > <br clear="none">> > Rahul, please see the logic I used to reduce time step size.<br clear="none">> > Hope this helps you.<br clear="none">> > <br clear="none">> > Ling<br clear="none">> > <br clear="none">> > for (timestep = 1; timestep <= N; timestep++) // loop over time steps<br clear="none">> > {<br clear="none">> > // before trying to solve a time step, 1) it is still not successful; 2) we are not giving up; 3) haven't failed yet<br clear="none">> > bool give_up = false; bool success = false; bool experienced_fail_this_time_step = false;<br clear="none">> > // save solutions and old solutions into u_copy and u_old_copy, in case a time step fails and we restart from this saved solutions<br clear="none">> > VecCopy(u, u_copy); VecCopy(u_old, u_old_copy);<br clear="none">> > <br clear="none">> > while ((!give_up) && (!success)) // as long as not successful and not giving up, we solve again with smaller time step<br clear="none">> > {<br clear="none">> > if (time_step_size < dt_min) { give_up = true; break; } // ok, bad luck, give up due to time step size smaller than a preset value<br clear="none">> > if (experienced_fail_this_time_step) { // get the vectors from backups if this is a re-try, i.e., already failed with a larger time step<br clear="none">> > VecCopy(u_old_copy, u); VecCopy(u_old_copy, u_old);<br clear="none">> > }<br clear="none">> > <br clear="none">> > try {<br clear="none">> > SNESSolve(snes, NULL, u);<br clear="none">> > SNESGetConvergedReason(snes, &snes_converged_reason);<br clear="none">> > <br clear="none">> > if (snes_converged_reason > 0) success = true; // yes, snes converged<br clear="none">> > else { // no, snes did not converge<br clear="none">> > cutTimeStepSize(); // e.g., dt / 2<br clear="none">> > experienced_fail_this_time_step = true;<br clear="none">> > }<br clear="none">> > }<br clear="none">> > catch (int err) { // in case your own pieces of code throws an exception<br clear="none">> > std::cout << "An exception occurred." << std::endl;<br clear="none">> > success = false;<br clear="none">> > cutTimeStepSize(); // e.g., dt / 2<br clear="none">> > experienced_fail_this_time_step = true;<br clear="none">> > }<br clear="none">> > }<br clear="none">> > <br clear="none">> > if (success) {<br clear="none">> > // output, print, whatever<br clear="none">> > // duplicate current solution to old solution in preparing next time step<br clear="none">> > VecCopy(u, u_old);<br clear="none">> > // you can increase time step size here, e.g. * 2<br clear="none">> > increaseTimeStepSize();<br clear="none">> > }<br clear="none">> > <br clear="none">> > if (give_up) {<br clear="none">> > simulationFailed = true;<br clear="none">> > std::cerr << "Simulation failed.\n";<br clear="none">> > //exit(1);// dont exit(1) now, just break the for-loop, let PETSc clean its workspace.<br clear="none">> > break;<br clear="none">> > }<br clear="none">> > }<br clear="none">> > <br clear="none">> > From: Rahul Samala <<a shape="rect" ymailto="mailto:srahul_05@yahoo.co.in" href="mailto:srahul_05@yahoo.co.in">srahul_05@yahoo.co.in</a>><br clear="none">> > Sent: Wednesday, August 29, 2018 10:37:30 PM<br clear="none">> > To: Ling Zou; Smith, Barry F.<br clear="none">> > Cc: PETSc Users List<br clear="none">> > Subject: Re: [petsc-users] Problem with SNES convergence<br clear="none">> > <br clear="none">> > Thank you Ling, I would definitely like to look at your code for reducing timestep size.<br clear="none">> > Thank you Barry for your inputs.<br clear="none">> > <br clear="none">> > --<br clear="none">> > Rahul.<br clear="none">> > <br clear="none">> > On Wednesday, August 29, 2018, 9:02:00 PM GMT+5:30, Smith, Barry F. <<a shape="rect" ymailto="mailto:bsmith@mcs.anl.gov" href="mailto:bsmith@mcs.anl.gov">bsmith@mcs.anl.gov</a>> wrote:<br clear="none">> > <br clear="none">> > <br clear="none">> > <br clear="none">> > Current time (before start of timestep) 52.5048, iter=5380 Timestep=864.000000 <br clear="none">> > 0 SNES Function norm 1.650467412595e+05 <br clear="none">> > 0 KSP preconditioned resid norm 3.979123221160e+03 true resid norm 1.650467412595e+05 ||r(i)||/||b|| 1.000000000000e+00<br clear="none">> > 1 KSP preconditioned resid norm 9.178246525982e-11 true resid norm 7.006473307032e-09 ||r(i)||/||b|| 4.245144892632e-14<br clear="none">> > Linear solve converged due to CONVERGED_RTOL iterations 1<br clear="none">> > 1 SNES Function norm 6.722712947273e+02 <br clear="none">> > Linear solve did not converge due to DIVERGED_NANORINF iterations 0<br clear="none">> > Nonlinear solve did not converge due to DIVERGED_LINEAR_SOLVE iterations 1<br clear="none">> > <br clear="none">> > This usually is an indicator that the LU (ILU) factorization has hit a zero pivot (hence the linear solve has a divide by zero so gives the DIVERGED_NANORINF flag). <br clear="none">> > <br clear="none">> > You can/should call SNESGetConvergedReason() immediately after each SNESSolve(), if the result is negative that means something has failed in the nonlinear solve and you can try cutting the time-step and trying again.<br clear="none">> > <br clear="none">> > Good luck,<br clear="none">> > <br clear="none">> > Barry<br clear="none">> > <br clear="none">> > <br clear="none">> > > On Aug 29, 2018, at 10:11 AM, Ling Zou <<a shape="rect" ymailto="mailto:Ling.Zou@inl.gov" href="mailto:Ling.Zou@inl.gov">Ling.Zou@inl.gov</a>> wrote:<br clear="none">> > > <br clear="none">> > > 1) My experience is that this kind of bug or sudden death (everything is fine till suddenly something is broken) is very difficult to debug/fix. I looked at your txt files and could not give any quick comments. Maybe PETSc developers have better idea on this.<br clear="none">> > > 2) I do have successful experience on reducing time step size when PETSc fails solving or my own piece of code throws an exception. If you are interested, I can share them.<br clear="none">> > > <br clear="none">> > > -Ling<br clear="none">> > > From: petsc-users <<a shape="rect" ymailto="mailto:petsc-users-bounces@mcs.anl.gov" href="mailto:petsc-users-bounces@mcs.anl.gov">petsc-users-bounces@mcs.anl.gov</a>> on behalf of Rahul Samala <<a shape="rect" ymailto="mailto:srahul_05@yahoo.co.in" href="mailto:srahul_05@yahoo.co.in">srahul_05@yahoo.co.in</a>><br clear="none">> > > Sent: Wednesday, August 29, 2018 8:36:58 AM<br clear="none">> > > To: PETSc Users List<br clear="none">> > > Subject: [petsc-users] Problem with SNES convergence<br clear="none">> > > <br clear="none">> > > Hello PetSc users,<br clear="none">> > > <br clear="none">> > > 1) I have problem with SNES convergence. I call SNESSolve in a time loop and use the inbuilt Jacobian feature. The code works fine for about 5380 time steps after which it breaks down. The solution till that point looks fine. I have used newtonls of type l2. (newtontr and others aren't working). Since I have used inbuilt Jacobian feature and the code worked for about 5000 time steps I don't understand the reason for failure, is it an incorrect function evaluation? Attached are the outputs with -pc_type lu and ilu along with -snes_linesearch_type l2 -snes_converged_reason -snes_monitor -snes_view -ksp_converged_reason -ksp_monitor_true_residual<br clear="none">> > > <br clear="none">> > > 2) How to get hold of failure signal, like Nonlinear solve DIVERGED_MAX_IT or DIVERGED_LINEAR_SOLVE so that whenever it occurs I can use a reduced time step and see if the code converges.<br clear="none">> > > <br clear="none">> > > Thank you,<br clear="none">> > > Rahul.<br clear="none">> > > <br clear="none">> > > output_ilu.txt<br clear="none">> > > <br clear="none">> > > <br clear="none">> > > output_ilu.txt<br clear="none">> > > <br clear="none">> > > <br clear="none">> > > <br clear="none">> > > output_lu.txt<br clear="none">> > > <br clear="none">> > > <br clear="none">> > > output_lu.txt<br clear="none">> <br clear="none"></div></div></div>
</div>
</div></div></body></html>