<div dir="ltr"><div dir="ltr">Dear Yuji, <br></div><div dir="ltr">I tried the option "-sl gap.set" you wrote about. It works and I see a non-zero gap in fscip log!</div><div><br></div><div>BUT, there is the next very unpleasant problem: parallel, multithreaded, fscip <b>runs much slower than one-thread scip</b> !!?? <br>And, if I remember correctly, I saw the same effect for FiberSCIP of older versions too....<br></div><div></div><div>For example, on my Dell comp, CPU i5-8250U 3.40GHz, 32Gb, with disabled Hyperthreading (to have "real" 4 CPUs) I run SCIP 8.0.3 and FSCIP 8.0.3 on the same problem [<a href="http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/tsp/burma14.tsp.gz" target="_blank">burma14</a>] from TSPLIB95 with limits/gap=1.e-01. Here SCIP had found a solution in ~800 sec and FSCIP (run on 4 CPUs, with "-sth 4") - in ~1400 sec.<br></div><div><br></div><div>Please, can you tell me what is the possible reason for this strange situation? <br><br></div><div>SCIP and FiberSCIP have been run by commands (on Linux)<br></div><div><div><b>$ scip -s scip.set -f tsp_burma14_1.cip -l tsp_burma14_q.scip803.dell.gap_0.1.log.txt</b></div><div></div><b>$ fscip ug.set tsp_burma14_q.cip -s scip.set -sl gap.set -sth 4 -l tsp_burma14_q.fscip803.dell.gap_0.1.log.txt</b></div>Just in case, I attached tsp_burma14_q.cip, scip.set, gap.set and log files to the message. <br><div>I should say that TSP has been intentionally represented as 0-1 Quadratic Programming problem, similar to Quadratic Assignment problem, with bilinear objective function and linear constraints. <br></div><div><br></div><div>Best regards,<br></div><div>Vladimir.<br></div><div><br></div><div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Apr 24, 2023 at 9:46 AM Yuji Shinano <<a href="mailto:shinano@zib.de" target="_blank">shinano@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>
<div>Dear Vladimir,<br>
<br>
<code>Stefan Vigerske and I tried to implement it so that it works
more precisely.<br>
Probably, the code is already released in the version.<br>
Please try to use the command line option with "-sl gap.set",
and the file gap.set inclueds<br>
"limits/gap" or "limits/absgap". Then, it would work better than
the previous versions.<br>
Please let me know the results.<br>
</code><code>@Stefan: If you can say something else, please add
your comments. <br>
<br>
Best regards,<br>
Yuji<br>
</code><br>
On 24.04.23 1:06, Vladimir V. Voloshinov wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>Dear Yuji,<br>
</div>
excuse for bothering you, but I see that SCIP Opt. Suite 8.0.3
includes UG 1.0 already. <br>
<div>I see that the parameter CheckGapInLC, which we tried to
use, has been excluded from the list of UG options, so it
seems that some changes with "gap management" happened...</div>
<div>But I see that the gap set in the scip.set file does not
affect FiberSCIP which still solves problem till zero gap. <br>
Moreover, I see in ug/src/ug_scip/scipParaSolver.cpp, lines
#1259,#1260, that "limits/gap" and "limits/absgap" are
"hardcoded" to be 0...<br>
</div>
<div>Maybe we missed something, but is it possible to set a non
zero gap for FiberSCIP (and ParaSCIP) now, since version
8.0.3? It is very important in some circumstances...<br>
</div>
<div><br>
</div>
<div>Best regards,<br>
</div>
<div>Vladimir<br>
</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Mon, Sep 12, 2022 at
5:36 PM Yuji Shinano <<a href="mailto:shinano@zib.de" target="_blank">shinano@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">Hi
Sergey,<br>
<br>
On 12.09.22 16:17, Sergey Smirnov wrote:<br>
> Thank you for checking my instance.<br>
><br>
> We also tried setting CheckGapInLC=TRUE but it did not
help either.<br>
<br>
Thanks for testing it, too. Probably, I tried to implement it
once,<br>
but I had some issues realizing it. Before UG version 1.0,<br>
UG was a bit hard to specialize for a specific "base solver"
parallelization.<br>
In current UG version 1.0, it is easy to implement some
features<br>
specialized to a specific "base solver".<br>
I will let you know when I add the feature to FiberSCIP and
ParaSCIP.<br>
<br>
Best,<br>
Yuji<br>
<br>
><br>
>> On 11 Sep 2022, at 17:42, Yuji Shinano <<a href="mailto:shinano@zib.de" target="_blank">shinano@zib.de</a>>
wrote:<br>
>><br>
>> Hi Sergey,<br>
>><br>
>> Thanks for the instance and configuration. I checked
it.<br>
>> Sorry, in the current implementation of FiberSCIP and
ParaSCIP, controlling it only with the SCIP parameter looks
difficult when it takes time to solve<br>
>> (if it can be solved in the racing stage, the
"NoPreprocessingInLC = TRUE" might work).<br>
>> And it looks like the setting SCIP parameters for
that is not good for FiberSCIP and ParaSCIP in a sense how it
work is not clear.<br>
>> I will add the feature rather UG side in the future.
I think it can not control vividly, but when LC recognized
that gap is reached,<br>
>> then LC interrupts all computations from the UG side.<br>
>><br>
>> Best regards,<br>
>> Yuji<br>
>><br>
>> --------------<br>
>> Hi Yuji,<br>
>><br>
>> Here are the instance and configuration files
attached.<br>
>><br>
>> Thank you in advance,<br>
>> Best regards,<br>
>> Sergey<br>
>><br>
>><br>
>> > On 10 Sep 2022, at 08:55, Yuji Shinano <<a href="mailto:shinano@zib.de" target="_blank">shinano@zib.de</a>>
wrote:<br>
>> ><br>
>> > Hi Sergey,<br>
>> ><br>
>> > Can you send me the instance and parameters (UG
and SCIP)? I will check it.<br>
>> ><br>
>> > Best regards,<br>
>> > Yuji<br>
>> ><br>
>> > On 10.09.22 6:50, Sergey Smirnov wrote:<br>
>> >> Hi Yuji,<br>
>> >><br>
>> >> Thank you for the suggestion. I tried to set
NoPreprocessingInLC = TRUE in ug.set but both fscip and
parascip still solved till zero gap.<br>
>> >><br>
>> >> Best regards,<br>
>> >> Sergey<br>
>> >><br>
>> >>> On 9 Sep 2022, at 21:20, Yuji Shinano
<<a href="mailto:shinano@zib.de" target="_blank">shinano@zib.de</a>>
wrote:<br>
>> >>><br>
>> >>> Hi,<br>
>> >>><br>
>> >>> This is one of the frequently asked
questions. FiberSCIP/ParaSCIP solves the presolved instance in
the Solvers.<br>
>> >>> Therefore, the value is applied to solve
the presolved instance. If you would like to control the gap
more precisely,<br>
>> >>> one way would be to set<br>
>> >>> NoPreprocessingInLC = TRUE<br>
>> >>> In the UG parameter file.<br>
>> >>> However, this means that layered
presolving does not work and may decrease its performance.<br>
>> >>> Please try to run with the above
parameter.<br>
>> >>><br>
>> >>> Best,<br>
>> >>> Yuji<br>
>> >>><br>
>><br>
>><br>
>> On 09.09.22 19:45, Sergey Smirnov wrote:<br>
>>> Hello, SCIP community,<br>
>>><br>
>>> I have problems setting relative gap in parallel
version of SCIP 8.0.1.<br>
>>><br>
>>> When I set limits/gap = .5e-01 in scip.set and
run scip like scip -f problem.cip -s scip.set scip stops
after reaching the gap specified. However, fscip and parascip
seem to ignore this parameter. For example, I run fscip the
following way and it solves until optimality:<br>
>>> fscip ug.set problem.cip -s scip.set<br>
>>><br>
>>> What is the correct way to set relative gap in
fscip/parascip?<br>
>>><br>
>>> Thank you in advance,<br>
>>> Best regards,<br>
>>> Sergey<br>
>>><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>
>> Yuji Shinano<br>
>> Zuse Institute Berlin<br>
>> Takustrasse 7, 14195 Berlin-Dahlem, Germany<br>
>> Phone:+49 30 84185-477, Fax:+49 30 84185-269<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>
--------------------------------------------------<br>
Yuji Shinano<br>
Zuse Institute Berlin<br>
Takustrasse 7, 14195 Berlin-Dahlem, Germany<br>
Phone:+49 30 84185-477, Fax:+49 30 84185-269<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>
</blockquote>
</div>
<br clear="all">
<br>
<span>-- </span><br>
<div dir="ltr">
<div dir="ltr">
<div>
<div dir="ltr">
<div>
<div dir="ltr">
<div>
<div dir="ltr">
<div>Vladimir V. Voloshinov,<br>
Ph.D, head of lab. C-3 "Distributed computing
algorithms", <a href="http://www.iitp.ru/ru/researchlabs/1040.htm" target="_blank">http://www.iitp.ru/ru/researchlabs/1040.htm</a>,<br>
Center for Distributed Computing, Institute for
Information Transmission Problems RAS, <a href="http://www.iitp.ru" target="_blank">http://www.iitp.ru</a><br>
web: <a href="https://scholar.google.ru/citations?hl=en&user=-m4QhNEAAAAJ&view_op=list_works&sortby=pubdate" target="_blank">GoogleScholar
profile</a></div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
<br>
<br>
<pre cols="0">--
--------------------------------------------------
Yuji Shinano
Zuse Institute Berlin
Takustrasse 7, 14195 Berlin-Dahlem, Germany
Phone:+49 30 84185-477, Fax:+49 30 84185-269</pre>
</div>
<u></u><u></u>
</blockquote></div><br clear="all"><br></div></div>