<html dir="ltr">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style>
<!--
@font-face
{font-family:"Cambria Math"}
@font-face
{font-family:Calibri}
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
margin-bottom:.0001pt;
font-size:11.0pt;
font-family:"Calibri",sans-serif}
a:link, span.MsoHyperlink
{color:#0563C1;
text-decoration:underline}
a:visited, span.MsoHyperlinkFollowed
{color:#954F72;
text-decoration:underline}
span.E-MailFormatvorlage17
{font-family:"Calibri",sans-serif;
color:windowtext}
.MsoChpDefault
{font-family:"Calibri",sans-serif}
@page WordSection1
{margin:70.85pt 70.85pt 2.0cm 70.85pt}
div.WordSection1
{}
-->
</style><style type="text/css" id="owaParaStyle">P {margin-top:0;margin-bottom:0;}</style>
</head>
<body style="" fpstyle="1" ocsi="0" vlink="#954F72" link="#0563C1" lang="DE">
<div style="direction: ltr;font-family: Tahoma;color: #000000;font-size: 10pt;">
<div class="WordSection1">
<p class="MsoNormal">Hello,</p>
<p class="MsoNormal"> </p>
<p class="MsoNormal">I'm trying to solve a problem using column generation and solving the master problem requires many pricing iterations leading to a large number of generated variables. </p>
<p class="MsoNormal">This slows the lp solver down significantly. According to the log file the number of variables equals the number of columns so I assume Scip did not drop any columns from the lp:</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal"> time | node | left |LP iter|LP it/n| mem |mdpt |frac |vars |cons |cols |rows |cuts |confs|strbr| dualbound | primalbound | gap
<br>
r1207s| 1 | 0 | 2238k| - | 156M| 0 | 0 | 10k|8380 | 10k|8380 | 0 | 0 | 0 | -- | 7.160129e+04 | Inf
<br>
r1208s| 1 | 0 | 2239k| - | 157M| 0 | 0 | 10k|8380 | 10k|8380 | 0 | 0 | 0 | -- | 7.160121e+04 | Inf
<br>
r1209s| 1 | 0 | 2240k| - | 157M| 0 | 0 | 10k|8380 | 10k|8380 | 0 | 0 | 0 | -- | 7.160044e+04 | Inf
<br>
r1211s| 1 | 0 | 2241k| - | 157M| 0 | 0 | 10k|8380 | 10k|8380 | 0 | 0 | 0 | -- | 7.159907e+04 | Inf
<br>
r1212s| 1 | 0 | 2243k| - | 157M| 0 | 0 | 10k|8380 | 10k|8380 | 0 | 0 | 0 | -- | 7.159789e+04 | Inf
<br>
r1213s| 1 | 0 | 2244k| - | 157M| 0 | 0 | 10k|8380 | 10k|8380 | 0 | 0 | 0 | -- | 7.159727e+04 | Inf
<br>
r1214s| 1 | 0 | 2245k| - | 157M| 0 | 0 | 10k|8380 | 10k|8380 | 0 | 0 | 0 | -- | 7.159654e+04 | Inf
<br>
</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal">(note that in this log snippet all integer variables are relaxed)<br>
</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal">As in an integer feasible solution only 20 of the priced variables can take non-zero values I assume that most of the priced variables are 0.<br>
</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal"><a name="_MailAutoSig"></a>I read <a href="https://scip.zib.de/doc-6.0.2/html/FAQ.php#howtodeletevariables" target="_blank" rel="noopener noreferrer">
https://scip.zib.de/doc-6.0.2/html/FAQ.php#howtodeletevariables</a></p>
<p class="MsoNormal">* SCIPvarMarkDeletable() is called for all priced variables</p>
<p class="MsoNormal">* SCIPsetBoolParam(scip, "lp/cleanupcols", TRUE);<br>
SCIPsetBoolParam(scip, "lp/cleanupcolsroot", TRUE);<br>
SCIPsetBoolParam(scip, "pricing/delvars", TRUE);<br>
SCIPsetBoolParam(scip, "pricing/delvarsroot", TRUE);</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal">After a quick look into lp.c I got the impression that these parameter only apply after a node was processed.</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal">Is there some mechanism to remove/delete columns within the pricing loop?</p>
<p class="MsoNormal">Obviously one could stop Scip and start a new instance seeded with the previously found variables (set Initial=FALSE). Or implement the initial pricing loop yourself...</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal">Best regards,</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal">Tobias<br>
</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal"><br>
</p>
</div>
</div>
</body>
</html>