<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <div class="moz-cite-prefix">Dear Vladimir V. Voloshinov,<br>
      <br>
      On 14.03.25 16:36, Vladimir V. Voloshinov wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAH87Kps3Ub++HiW2xrTrGPnJ2pbG9GZW4LfH3f9erh+R3naSiw@mail.gmail.com">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <div dir="ltr">
        <div dir="ltr">
          <div dir="ltr">
            <div>Dear SCIP team,</div>
            <div><span style="white-space:pre-wrap">we</span><span
              style="white-space:pre-wrap"> have been </span><span
              style="white-space:pre-wrap">successfully</span><span
              style="white-space:pre-wrap"> </span><span
              style="white-space:pre-wrap">using</span><span
              style="white-space:pre-wrap"> </span><span
              style="white-space:pre-wrap">SCIP & FiberSCIP & ParaSCIP</span><span
              style="white-space:pre-wrap"> for a </span><span
              style="white-space:pre-wrap">long</span><span
              style="white-space:pre-wrap"> </span><span
              style="white-space:pre-wrap">time</span><span
              style="white-space:pre-wrap">.</span></div>
            <div><span style="white-space:pre-wrap">But sometimes some problems arise...
</span></div>
            <div>Recently we found that FiberSCIP works very very slowly
              in deterministic mode.</div>
            <div><span style="white-space:pre-wrap">Moreover, it looks like <b><i>Deterministic=TRUE</i></b> (in ug.set file) switches off multithreading and regardless of <b><i>-sth</i></b> setting total CPU load becomes the same as for the case <b><i>-sth=1</i></b>.</span></div>
            <div>But performance becomes even worse than for one
              threaded SCIP... See example below.</div>
            <div><br>
            </div>
            <div>The question is: maybe there are some other ways to get
              a deterministic behaviour of FiberSCIP keeping
              performance? The question arose when we tried to compare
              FiberSCIP performance and results on different computing
              environments...</div>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    No, unfortunately. The deterministic mode was developed only to
    debug FiberSCIP more efficiently. <br>
    Please read the FiberSCIP paper: <br>
    <a class="moz-txt-link-freetext" href="https://pubsonline.informs.org/doi/10.1287/ijoc.2017.0762">https://pubsonline.informs.org/doi/10.1287/ijoc.2017.0762</a><br>
    <style>@font-face
        {font-family:"Cambria Math";
        panose-1:2 4 5 3 5 4 6 3 2 4;
        mso-font-charset:0;
        mso-generic-font-family:roman;
        mso-font-pitch:variable;
        mso-font-signature:-536870145 1107305727 0 0 415 0;}@font-face
        {font-family:"Yu Gothic";
        panose-1:2 11 4 0 0 0 0 0 0 0;
        mso-font-alt:游ゴシック;
        mso-font-charset:128;
        mso-generic-font-family:swiss;
        mso-font-pitch:variable;
        mso-font-signature:-536870145 717749759 22 0 131231 0;}@font-face
        {font-family:Aptos;
        panose-1:2 11 0 4 2 2 2 2 2 4;
        mso-font-charset:0;
        mso-generic-font-family:swiss;
        mso-font-pitch:variable;
        mso-font-signature:536871559 3 0 0 415 0;}@font-face
        {font-family:"\@游ゴシック";
        panose-1:2 11 4 0 0 0 0 0 0 0;
        mso-font-alt:"\@Yu Gothic";
        mso-font-charset:128;
        mso-generic-font-family:swiss;
        mso-font-pitch:variable;
        mso-font-signature:-536870145 717749759 22 0 131231 0;}p.MsoNormal, li.MsoNormal, div.MsoNormal
        {mso-style-unhide:no;
        mso-style-qformat:yes;
        mso-style-parent:"";
        margin:0mm;
        text-align:justify;
        text-justify:inter-ideograph;
        mso-pagination:none;
        font-size:10.5pt;
        mso-bidi-font-size:12.0pt;
        font-family:"Aptos",sans-serif;
        mso-ascii-font-family:Aptos;
        mso-ascii-theme-font:minor-latin;
        mso-fareast-font-family:"Yu Gothic";
        mso-fareast-theme-font:minor-fareast;
        mso-hansi-font-family:Aptos;
        mso-hansi-theme-font:minor-latin;
        mso-bidi-font-family:"Times New Roman";
        mso-bidi-theme-font:minor-bidi;
        mso-font-kerning:1.0pt;}.MsoChpDefault
        {mso-style-type:export-only;
        mso-default-props:yes;
        font-size:10.5pt;
        mso-ansi-font-size:10.5pt;
        font-family:"Aptos",sans-serif;
        mso-bidi-font-family:"Times New Roman";
        mso-bidi-theme-font:minor-bidi;
        mso-ligatures:none;}div.WordSection1
        {page:WordSection1;}</style>And, it would not always be deterministic
    with the current version of SCIP.<br>
    <br>
    Best regards,<br>
    Yuji<br>
    <br>
    <blockquote type="cite"
cite="mid:CAH87Kps3Ub++HiW2xrTrGPnJ2pbG9GZW4LfH3f9erh+R3naSiw@mail.gmail.com">
      <div dir="ltr">
        <div dir="ltr">
          <div dir="ltr">
            <div><br>
            </div>
            <div>Below is an example of results for some NLP global
              optimization problem without discrete variables but with
              SOS2 constraints:</div>
            <div><span style="font-family:arial,sans-serif">___________________________________________________________</span></div>
            <div><span style="font-family:arial,sans-serif">Presolved
                Problem  :<br>
                  Variables       : 145 (0 binary, 0 integer, 0 implicit
                integer, 145 continuous)<br>
                  Constraints    : 61<br>
                Constraints      : Number<br>
                  SOS2             : 15 <br>
                  linear              : 30 <br>
                  nonlinear        : 16 <br>
              </span></div>
            <div>--------------------------------------------------------------------------------------------------</div>
            <div>It was solved with total 10% gap, <b><i>-sth=16</i></b>
              and <b><i>Deterministic=FALSE</i></b> in 87 sec:</div>
            <div>
              <div><span style="font-family:arial,sans-serif">_____________________________________________________</span></div>
              SCIP Status        : solving was interrupted [given gap
              reached]<br>
              Total Time         : 86.80<br>
            </div>
            <div>B&B Tree           :   nodes (total)    : 202112<br>
              Solution           :<br>
                Primal Bound     : +3.10185241045302e-03<br>
                Dual Bound       : +2.98514588645749e-03<br>
              Gap                : 3.90958 %</div>
            <div>----------------------------------------------------------------------------------------</div>
            <div><br>
            </div>
            <div>But solving time with the same settings and <i><b>Deterministic=TRUE</b></i>
              became almost 50 times bigger:</div>
            <div><span style="font-family:arial,sans-serif">_____________________________________________________</span></div>
            <div>
              <div>SCIP Status        : solving was interrupted [given
                gap reached]<br>
                Total Time         : 4216.07<br>
                B&B Tree           :   nodes (total)    : 141474<br>
                Solution           :<br>
                  Primal Bound     : +3.09812857597093e-03<br>
                  Dual Bound       : +2.98180690970866e-03<br>
                Gap                : 3.90105 %<br>
----------------------------------------------------------------------------------------</div>
              <div><br>
              </div>
              Sincerely yours,</div>
            <div dir="ltr" class="gmail_signature"
              data-smartmail="gmail_signature">
              <div dir="ltr">
                <div>
                  <div dir="ltr">
                    <div>
                      <div dir="ltr">
                        <div>
                          <div dir="ltr">
                            <div>Vladimir V. Voloshinov,<br>
                              web: <a
href="https://scholar.google.ru/citations?hl=en&user=-m4QhNEAAAAJ&view_op=list_works&sortby=pubdate"
                                target="_blank" moz-do-not-send="true">GoogleScholar
                                profile</a></div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </div>
        </div>
      </div>
      <br>
      <fieldset class="moz-mime-attachment-header"></fieldset>
      <pre wrap="" class="moz-quote-pre">_______________________________________________
Scip mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Scip@zib.de">Scip@zib.de</a>
<a class="moz-txt-link-freetext" href="https://listserv.zib.de/mailman/listinfo/scip">https://listserv.zib.de/mailman/listinfo/scip</a>
</pre>
    </blockquote>
    <br>
    <br>
    <pre class="moz-signature" cols="0">-- 
--------------------------------------------------
Yuji Shinano
Zuse Institute Berlin
Takustrasse 7, 14195 Berlin-Dahlem, Germany
Phone:+49 30 84185-477, Fax:+49 30 84185-269</pre>
  </body>
</html>