<div dir="ltr">On Tue, Oct 22, 2013 at 2:18 AM, Gong Ding <span dir="ltr"><<a href="mailto:gdiso@ustc.edu" target="_blank">gdiso@ustc.edu</a>></span> wrote:<br><div class="gmail_extra"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi,<br>
I updated petsc from 3.2 to 3.4. However, I found that the nonlinear solver previously worked well now failed with massage DIVERGED_LINE_SEARCH.<br>
My problem is extremely nonlinear, cubic line search may fail but Newton method works.<br>
<br>
With petsc 3.2, I only use basic Newton by calling<br>
SNESLineSearchSet(snes, SNESLineSearchNo, PETSC_NULL)<br>
With 3.4 I call basic Newton by<br>
SNESLineSearchSetType(snes, SNESLINESEARCHBASIC).<br>
<br>
However, petsc 3.4 still reports<br>
[0] SNESSolve_NEWTONLS(): fnorm=3.7500000000000006e-02, gnorm=3.7500000000000006e-02, ynorm=1.0000000000000000e+00, lssucceed=0<br>
[0] SNESNEWTONLSCheckLocalMin_Private(): || J^T F|| 1.414213562373e+00 near zero implies found a local minimum<br>
<br>
<br>
It seems petsc 3.4 can not disable line search. Does anyone have the same problem?</blockquote><div><br></div><div>Send -snes_view</div><div><br></div><div>   Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<span class="HOEnZb"><font color="#888888"><br>
Gong Ding<br>
<br>
<br>
<br>
</font></span></blockquote></div><br><br clear="all"><div><br></div>-- <br>What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>
-- Norbert Wiener
</div></div>