<div dir="ltr"><div>Dear Benjamin, <br></div>your recommendation to try "separating/minefficacy* is very helpful, but the next questions appeared.<div><br></div><div>Step by step.</div><div><br></div><div>1. When I used settings</div><div>separating/minefficacyroot = 5000</div><div>separating/minefficacy = 0.01</div><div>My SCIPAMPL solved the problem in ~400 sec in the root node also.</div><div>See the last line of log:</div><div><div>==============================================================<br></div></div><div>397s| 1 | 0 | 4204k| - | 43M| 0 | - | 427 | 438 | 427 |2664 |8377 | 0 | 0 | 1.615240e-01 | 1.615240e-01 | 0.00%</div><div><div>==============================================================<br></div></div><div>I used desktop with i7-6700 CPU @ 3.40GHz, SCIP 6.0.2, Ubuntu 16.04.06, GCC v. 5.4.0 (20160609).</div><div><br></div><div>But I have the following questions.</div><div>1.1. You wrote "I also disabled heuristics ...". <br></div><div>But how did you do that? I can not find proper settings... :o(</div><div>May be it is the reason of that "my" final number of cuts (8377) is twice bigger than "yours" (3902)?</div><div><br></div><div>1.2. Is it possible to reduce the number of annoying "every second" log lines? For the current running it is about 700 Kb!<br></div><div>Before, we used [display/freq] to reduce log file size, but in this case number of nodes does not change!<br></div><div><br></div><div>1.3. And still, is possible to force branching for this problems somehow ?</div><div><br></div><div>And now BAD news and question.<br></div><div>2. Is it possible that the result of deterministic SCIP-code is hardware dependent ?</div><div><br></div><div>The thing is that we could not reproduce successful solving on another computers with the same: OS (Ubuntu 16.04.06), GCC and SCIP version (including Ipopt version)!</div><div>All SCIP settings were same as for desktop running! <br></div><div>There were two notebooks: one(CPU i5-8250U @ 1.6Ghz (max 3.4Ghz) and Virtualbox 6.0.10 on another iMac notebook.</div><div>On both machines SCIP fell in stuck like it was before your advice to try "minefficacy" options. <br></div><div>Number of cuts did not grows dramatically and according to "perf top" tool SCIP "fall in SoPlex" (on Virtualbox):</div><div>==============================================================<br></div><div>Samples: 105K of event 'cpu-clock', Event count (approx.): 7318241336<br>Overhead Shared Object Symbol<br> 17.60% scip-6.0.2.linux.x86_64.gnu.opt.spx2.none [.] soplex::SPxDevexPR::selectEnterDenseCoDim<br> 14.94% scip-6.0.2.linux.x86_64.gnu.opt.spx2.none [.] soplex::CLUFactor::vSolveUright<br> 10.00% scip-6.0.2.linux.x86_64.gnu.opt.spx2.none [.] soplex::SPxSolver::setupPupdate</div>...........................<br><div>==============================================================</div><div><br></div><div>Moreover, I copied /usr/local/bin/scipampl executable (ONLY, and I did not copy all related libraries) from notebook to desktop.</div><div>Linux diff told that this SCIPAMPL file differs from the "native" one that has been compiled on desktop computer. <br></div><div>But when I ran this "guest" SCIPAMPL on desktop computer the problem has been solved successfully (as the "native" SCIPAMPL did before)!</div><div><br></div><div>Best regards and thanks,</div><div>Vladimir.<br></div><div><br></div><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Oct 8, 2019 at 11:26 PM Benjamin Müller <<a href="mailto:benjamin.mueller@zib.de">benjamin.mueller@zib.de</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">
<div bgcolor="#FFFFFF">
<tt>Dear Vladimir,<br>
<br>
good news, I configured SCIP in a way that I could solve your
problem to global optimality:<br>
<br>
732s| 1 | 0 | 2043k| - |7712k| 0 | - | 427 |
438 | 427 |1522 |3902 | 0 | 0 | 1.615240e-01 | 1.615240e-01
| 0.00%<br>
<br>
SCIP Status : problem is solved [optimal solution found]<br>
Solving Time (sec) : 732.03<br>
Solving Nodes : 1<br>
Primal Bound : +1.61523993625945e-01 (1 solutions)<br>
Dual Bound : +1.61523993625945e-01<br>
Gap : 0.00 %<br>
<br>
The way how SCIP enforces nonconvex nonlinear constraints is not
working well for this instance. SCIP will always prioritize
separation over branching when enforcing nonconvex nonlinear
constraints. The only condition here is that the computed cuts
have a efficacy above some threshold. Surprisingly, SCIP finds an
insane number of cuts during the enforcement rounds and thus SCIP
will spend a pretty long time in the root node. To overcome this
problem, I increased the minimum efficacy</tt><tt> drastically,
i.e., </tt><br>
<tt><tt><tt>set separating minefficacyroot 5000. As a result, the
problem is solved after 732 seconds in the root node. I also
disabled heuristics and used the solution that you sent me.<br>
<br>
Yes, you can just write "display statistics" in the
interactive shell. You can also terminate the current solving
process by CTRL+C and then display the statistics. I don't
really know what you mean with "... not any appropriate
settings might be inserted in *.set file", but displaying the
statistics doesn't depend on the settings.<br>
<br>
Best,<br>
Benjamin<br>
</tt></tt><br>
</tt><br>
<div>Am 08.10.19 um 19:39 schrieb Vladimir
V. Voloshinov:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div dir="ltr">
<div>Dear Benjamin,</div>
<div>thank you for your reply!</div>
<div><br>
</div>
<div>1.<br>
</div>
<div>I attached problem's data in CIP format
[tmpabc0007_0000start-bound.cip]. <br>
</div>
<div>Just in case I attached initial solution also [
tmpabc0007_0000start.cip.init.sol]. Objective value had been
changed form 0 to "correct" value manually...</div>
<div><br>
</div>
<div>2. You wrote "print statistics ..." <br></div>
<div>Am I right that it is possible by SCIP console command
"display statistics" only, and there is not any appropriate
settings might be inserted in *.set file ?<br>
</div>
<div><br>
</div>
<div>Best regards,</div>
<div>Vladimir<br>
</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Tue, Oct 8, 2019 at 2:40
PM Benjamin Müller <<a href="mailto:benjamin.mueller@zib.de" target="_blank">benjamin.mueller@zib.de</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">Dear Vladimir,<br>
<br>
yes, SCIP did not start to branch yet. Due to the large
number of <br>
generated cutting planes, I would suspect that most of the
time is spent <br>
in solving the LP relaxations, but it could also be that
some plugins <br>
are too expensive. You could print the statistics and check
which <br>
plugins could be disabled/configured to improve SCIP's
performance.<br>
<br>
Could you send me your instance in the CIP format? Then I
could have a <br>
look at your instance.<br>
<br>
Best regards,<br>
Benjamin<br>
<br>
On 10/8/19 10:11 AM, Vladimir V. Voloshinov wrote:<br>
> Dear SCIP developers,<br>
> we tried to solve some "middle-size" nonlinear problem
with polynomial <br>
> functions in constraints with out any discrete
variables (extr. from <br>
> NL-file header below)<br>
> 430 442 1 0 401 # vars, constraints, objectives,
ranges, eqns<br>
> 199 1 0 0 0 0 # nonlinear constrs, objs; ccons:
lin, nonlin, nd, nzlb<br>
> 129 221 26 # nonlinear vars in constraints,
objectives, both<br>
> 0 0 0 0 0 # discrete variables: binary,
integer, <br>
> nonlinear (b,c,o)<br>
> Some of nonlinear constraints are non-convex quadratic
and others (100 <br>
> totally) are 3rd order polynomial, i.e. sums of the
items like " <br>
> <var1>*sqr(<var2>)" (in CIP format).<br>
> Local solver IPOPT founds local optimum (we suspect
that it is global) <br>
> in 10 seconds.<br>
> But SCIP with default settings (and FiberSCIP) can not
prove global <br>
> optimality in an hours and I DO NOT SEE ANY BRANCHING
(domain <br>
> decomposition), see fragment of SCIP-log:<br>
> =============================<br>
> time | node | left |LP iter |LP it/n| mem |mdpt
|frac |vars |cons <br>
> |cols |rows |cuts | ,,, | dualbound |
primalbound | gap<br>
> 5004s| 1 | 0 | 12932k| - | 77M|
0 | 0 | <br>
> 427 | 438 | 427 | 89k | 190k | ... |-2.167723e-01 |
1.615240e-01 | Inf<br>
> ...........<br>
> 358m| 1 | 0 | 31028k| - | 103M|
0 | 0 | <br>
> 427 | 438 | 427 | 126k| 259k| ... |-1.800902e-01 |
1.615240e-01 | Inf<br>
> SCIP Status : solving was interrupted [user
interrupt]<br>
> Solving Time (sec) : 21469.69<br>
> Solving Nodes : 1<br>
> Primal Bound : +1.61523993625950e-01 (2
solutions)<br>
> Dual Bound : -1.80090245701914e-01<br>
> =============================<br>
> <br>
> I understand that it is a hard problem, but it is
strange that solver <br>
> did not try to split bounded domain ! Only number of
linear cuts has <br>
> been increasing...<br>
> It looks like SCIP only tries to approximate objective
function epigraph <br>
> by polyhedron (as a result lower bound increases
slowly).<br>
> <br>
> Is it normal behaviour of SCIP or some settings for
branching rules may <br>
> be applied ?<br>
> <br>
> Sincerely yours,<br>
> -- <br>
> Vladimir V. Voloshinov,<br>
> Ph.D, head of lab. C-3 "Distributed computing
algorithms", <br>
> <a href="http://www.iitp.ru/ru/researchlabs/1040.htm" rel="noreferrer" target="_blank">http://www.iitp.ru/ru/researchlabs/1040.htm</a>,<br>
> Center for Distributed Computing, Institute for
Information Transmission <br>
> Problems RAS, <a href="http://www.iitp.ru" rel="noreferrer" target="_blank">http://www.iitp.ru</a><br>
> web: our site is temporarily offline <br>
> <<a href="https://scholar.google.ru/citations?hl=en&user=-m4QhNEAAAAJ&view_op=list_works&sortby=pubdate" rel="noreferrer" target="_blank">https://scholar.google.ru/citations?hl=en&user=-m4QhNEAAAAJ&view_op=list_works&sortby=pubdate</a>><br>
> <br>
> _______________________________________________<br>
> Scip mailing list<br>
> <a href="mailto:Scip@zib.de" target="_blank">Scip@zib.de</a><br>
> <a href="https://listserv.zib.de/mailman/listinfo/scip" rel="noreferrer" target="_blank">https://listserv.zib.de/mailman/listinfo/scip</a><br>
> <br>
<br>
-- <br>
______________________________<br>
Benjamin Müller<br>
Zuse Institute Berlin<br>
Takustr. 7, 14195 Berlin<br>
<a href="mailto:benjamin.mueller@zib.de" target="_blank">benjamin.mueller@zib.de</a><br>
+49 30 841 85-195<br>
</blockquote>
</div>
<br clear="all">
<br>
</div>
</blockquote>
<br>
</div>
</blockquote></div><br></div>