<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
<div class="moz-cite-prefix">On 1/19/20 11:38 AM, Mark Adams wrote:<br>
</div>
<blockquote type="cite" cite="mid:CADOhEh7+d9wLMfb5SdM3VS_gduM_Q1FngXGSwNZ6N5GmQ=fQFQ@mail.gmail.com">
<div dir="ltr">Can you recommend a higher order method that I might try?</div>
<br>
</blockquote>
<p>Mark, all of 2e, 3, 4, 5 are high order with really good properties. They have error estimators that are cheaper but less reliable (most of the time work well enough).</p>
<p>Emil<br>
</p>
<blockquote type="cite" cite="mid:CADOhEh7+d9wLMfb5SdM3VS_gduM_Q1FngXGSwNZ6N5GmQ=fQFQ@mail.gmail.com">
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Sun, Jan 19, 2020 at 10:37 AM Jed Brown <<a href="mailto:jed@jedbrown.org" moz-do-not-send="true">jed@jedbrown.org</a>> wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px
0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
Use -ts_adapt_monitor to see the rationale.<br>
<br>
Note that 1bee is backward Euler with an extrapolation error estimator<br>
(for adaptive control). It's still only first order accurate, and the<br>
longer step may be part of your SNES issues.<br>
<br>
You can set a maximum time step (-ts_adapt_dt_max) or be more aggressive<br>
about reducing time step in response to SNES failure<br>
(-ts_adapt_scale_solve_failed) or remember that failure for longer<br>
before increasing the step again (-ts_adapt_time_step_increase_delay) or<br>
more gradually increase time step when permitted (-ts_adapt_clip).<br>
<br>
Mark Adams <<a href="mailto:mfadams@lbl.gov" target="_blank" moz-do-not-send="true">mfadams@lbl.gov</a>> writes:<br>
<br>
> I am using -ts_type arkimex -ts_arkimex_type 1bee -ts_max_snes_failures -1<br>
> -ts_rtol 1e-6 -ts_dt 1.e-7<br>
><br>
> First ,Jed gave me these parameters. This is not a DAE, just a fully<br>
> implicit solve. Advice on parameters welcome.<br>
><br>
> Second, TS is reporting a large time step (0.0505357) that is wrong.<br>
><br>
> Third, it repeatedly takes this extra one or two (its a 3 step method) step<br>
> due to SNES failure. I wonder if that can be optimized.<br>
><br>
> Thanks,<br>
> Mark<br>
><br>
><br>
> ....<br>
> 9 SNES Function norm 1.438395286712e-06<br>
> 10 SNES Function norm 8.050454869525e-07<br>
> Nonlinear solve converged due to CONVERGED_SNORM_RELATIVE iterations 10<br>
> [0] TSAdaptChoose_Basic(): Estimated scaled local truncation error<br>
> 0.00461254, *accepting step of size 0.00304954*<br>
> 600 TS *dt 0.0304954* time 0.697817<br>
> 0 SNES Function norm 1.018387577463e-02<br>
> ...<br>
> 23 SNES Function norm 6.583420045281e-05<br>
> 24 SNES Function norm 5.959294539241e-05<br>
> 25 SNES Function norm 5.394347124131e-05<br>
> Nonlinear solve did not converge due to *DIVERGED_MAX_IT* iterations 25<br>
> 0 SNES Function norm 1.018387577468e-02<br>
> ...<br>
> 24 SNES Function norm 1.000717662032e-06<br>
> 25 SNES Function norm 7.741622573808e-07<br>
> Nonlinear solve converged due to *CONVERGED_SNORM_RELATIVE* iterations 25<br>
> 0 SNES Function norm 1.014795904701e-02<br>
> ...<br>
> 15 SNES Function norm 1.334407891279e-06<br>
> 16 SNES Function norm 9.148934277015e-07<br>
> Nonlinear solve converged due to *CONVERGED_SNORM_RELATIVE* iterations 16<br>
> 0 SNES Function norm 1.016588008759e-02<br>
> ...<br>
> 16 SNES Function norm 9.144418053264e-07<br>
> Nonlinear solve converged due to *CONVERGED_SNORM_RELATIVE* iterations 16<br>
> [0] TSAdaptChoose_Basic(): Estimated scaled local truncation error<br>
> 0.0184347, *accepting step of size 0.00762384*<br>
> 601 TS *dt 0.0505357 *time 0.705441<br>
> 0 SNES Function norm 1.014792968017e-02<br>
> 1 SNES Function norm 1.026477259201e-03<br>
> 2 SNES Function norm 6.170336507030e-04<br>
> 3 SNES Function norm 5.433176612554e-04<br>
> 4 SNES Function norm 5.196626557375e-04<br>
> 5 SNES Function norm 4.977855046309e-04<br>
</blockquote>
</div>
</blockquote>
</body>
</html>