<div dir="ltr"><div dir="ltr"><div>Dear colleagues,</div><div>I found your discussion on <a href="http://listserv.zib.de/pipermail/scip/2021-February/">http://listserv.zib.de/pipermail/scip/2021-February/</a> because I began to receive the same messages:<br><b>(node xxxxxx) numerical troubles in LP yyyyyyy -- unresolved</b></div><div></div><div>It was almost the first time I saw them in the SCIP log.<br></div>It was a global optimization problem with continuous variables, linear, bilinear and SOS2 constraints.<div>E.g. the following dimension (but I saw that "numerical troubles" for smaller problems of the same kind as well)<br><b>presolved problem has 337 variables (0 bin, 0 int, 0 impl, 337 cont) and 81 constraints<br> 20 constraints of type <SOS2><br> 40 constraints of type <linear><br> 21 constraints of type <nonlinear></b></div><div><br></div><div>Can you, please, answer a couple of questions:</div><div>1. As far as I understood, Marc Pfetsch recommended SCIP build with IPOPT, but my SCIP was built with it already.<div><b>I use SCIP 8.0.1/Soplex 6.0.1/Ipopt 3.14.9 now.</b> <br></div><div>Is it possible to "stimulate" SCIP to use "more Ipopt" than LP by some SCIP parameters ?</div><div></div><div><br></div><div>2. After the 10 warnings I see <br><b>(node 30317) numerical troubles in LP 175340 -- unresolved -- further messages will be suppressed (use display/verblevel=5 to see all)</b></div><div>Although SCIP continued solving to the normal completion it seems that these troubles slow down the solving process.</div><div>Is it true? Are there any known evaluations of loss in performance for the case?</div></div></div><div><br></div><div>Sincerely yours,<br></div><div>Vladimir Voloshiov.<br><br></div><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Fri, Feb 26, 2021 at 2:00 PM James Cussens <<a href="mailto:james.cussens@bristol.ac.uk">james.cussens@bristol.ac.uk</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 class="msg-3353679587632330002">
<div dir="ltr">
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
Hi Marc,</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
<br>
</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
Yes, using IPOPT is a good idea. I've used IPOPT before on similar instances (not ones with numerical problems) and, much to my surprise, solving was slower. But avoiding</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
these numerical problems is more important and perhaps I can get to the bottom of (and hopefully avoid) the performance problem.</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
<br>
</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
Thanks,</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
<br>
</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
James<br>
</div>
<div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
<br>
</div>
<div id="m_-3353679587632330002Signature">
<div>
<div></div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
James Cussens</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
Dept of Computer Science, University of Bristol</div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
<a href="https://jcussens.github.io/" target="_blank">https://jcussens.github.io/</a></div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif;font-size:12pt;color:rgb(0,0,0)">
<span style="font-size:11.5pt">Funded PhDs available in Bristol in the following areas: <a href="http://www.bristol.ac.uk/cdt/compass/" rel="noopener noreferrer" target="_blank"><span>Data Science</span></a>, <a href="http://www.bristol.ac.uk/cdt/interactive-ai/" rel="noopener noreferrer" target="_blank"><span>Interactive
AI</span></a>, <a href="http://www.bristol.ac.uk/cdt/cyber-security/" rel="noopener noreferrer" target="_blank"><span>Cyber Security</span></a> or <a href="http://www.bristol.ac.uk/cdt/digital-health/" rel="noopener noreferrer" target="_blank"><span>Digital
Health</span></a>. </span><br>
</div>
</div>
</div>
</div>
<div id="m_-3353679587632330002appendonsend"></div>
<hr style="display:inline-block;width:98%">
<div id="m_-3353679587632330002divRplyFwdMsg" dir="ltr"><font style="font-size:11pt" face="Calibri, sans-serif" color="#000000"><b>From:</b> Scip <<a href="mailto:scip-bounces@zib.de" target="_blank">scip-bounces@zib.de</a>> on behalf of Marc Pfetsch <<a href="mailto:pfetsch@mathematik.tu-darmstadt.de" target="_blank">pfetsch@mathematik.tu-darmstadt.de</a>><br>
<b>Sent:</b> 26 February 2021 10:51<br>
<b>To:</b> <a href="mailto:scip@zib.de" target="_blank">scip@zib.de</a> <<a href="mailto:scip@zib.de" target="_blank">scip@zib.de</a>><br>
<b>Subject:</b> Re: [SCIP] unresolved numerical troubles</font>
<div> </div>
</div>
<div><font size="2"><span style="font-size:11pt">
<div><br>
<br>
Hi James,<br>
<br>
unresolved numerical troubles, means that all efforts of SCIP to get a<br>
feasible and optimal LP solution were fruitless. SCIP tries to tighten<br>
tolerances, changes the simplex method, restarts from scratch etc. in<br>
order to do so. If the LP solver fails to produce a solution, SCIP has<br>
to stop.<br>
<br>
In your particular case, there seems to be one simple thing to improve<br>
your situation: You should build SCIP with IPOPT - this solves your<br>
instance for me. IPOPT helps with finding feasible solutions, which<br>
seems to stablize the process.<br>
<br>
Best<br>
<br>
Marc<br>
<br>
<br>
<br>
On 26/02/2021 10.05, James Cussens wrote:<br>
> Hi all,<br>
> <br>
> I use a subMIQP for a pricer I have. Sometimes when I run it I get<br>
> <br>
> [solve.c:3912] ERROR: (node 1) unresolved numerical troubles in LP 32<br>
> cannot be dealt with<br>
> <br>
> I am keen to understand what might have led to such (fatal!) problems<br>
> and if I can avoid them. I attach an example of this problem. In the<br>
> attached file you have the original problem and the output when solving<br>
> the (presolved, transformed problem). The odd thing is that, in the<br>
> example I have attached, after presolving, the problem should not be too<br>
> hard to solve: all variables are continuous and we only have convex<br>
> constraints (22 second-order cone and 21 linear).<br>
> <br>
> One of my variables is fixed to a small positive value, perhaps that has<br>
> something to do with it.<br>
> <br>
> I am using SoPlex as the LP solver.<br>
> <br>
> James<br>
> <br>
> James Cussens<br>
> Dept of Computer Science, University of Bristol<br>
> <a href="https://jcussens.github.io/" target="_blank">https://jcussens.github.io/</a><br>
> Funded PhDs available in Bristol in the following areas: Data Science<br>
> <<a href="http://www.bristol.ac.uk/cdt/compass/" target="_blank">http://www.bristol.ac.uk/cdt/compass/</a>>, Interactive AI<br>
> <<a href="http://www.bristol.ac.uk/cdt/interactive-ai/" target="_blank">http://www.bristol.ac.uk/cdt/interactive-ai/</a>>, Cyber Security<br>
> <<a href="http://www.bristol.ac.uk/cdt/cyber-security/" target="_blank">http://www.bristol.ac.uk/cdt/cyber-security/</a>> or Digital Health<br>
> <<a href="http://www.bristol.ac.uk/cdt/digital-health/" target="_blank">http://www.bristol.ac.uk/cdt/digital-health/</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" target="_blank">https://listserv.zib.de/mailman/listinfo/scip</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" target="_blank">https://listserv.zib.de/mailman/listinfo/scip</a><br>
</div>
</span></font></div>
</div>
_______________________________________________<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>
</div></blockquote></div><br clear="all"><br></div>