<html>
  <head>

    <meta http-equiv="content-type" content="text/html; charset=iso-8859-2">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Dear All,<br>
    <br>
    I'm using scip 3.1.1 with soplex 2.0.1. (on 1 thread) through C++.
    I'm deailing with a branch-and-price procedure for a MILP. The
    problem contains continuous and binary variables where the binary
    variables have lazy upper bound (the minimization problem has a set
    partition structure so an LP solution should respect these bounds).<br>
    <br>
    I have made a pricer and a branching rule, of course, and I'm
    expecting SCIP to call my function scip_execlp. And SCIP does it
    usually, but sometimes calls the function scip_execps, and I can't
    find its reason.<br>
    <br>
    Okay, it's clear from the documentation that SCIP calls scip_execps
    if 'an integral solution violates one or more constraints and this
    infeasibility could not be resolved...', but what does it mean?
    Hasn't SCIP solved the LP in the focused node? I have made an event
    handler to check the solving steps and I've got the following
    output:<br>
    <i>...</i><i><br>
    </i><i>[nodefocused] node 11 has been focused and is now the focus
      node</i><i><br>
    </i><i>[firstlpsolved] initial LP of node 11 was solved</i><i><br>
    </i><i>[lpsolved] LP of node 11 was completely solved with cut &
      price</i><i><br>
    </i><i>[nodebranched] the focus node 11 has been solved by branching</i><i><br>
    </i><i>...</i><br>
    so it means for me that the LP has been solved. But the branching
    procedure used the function scip_execps instead of scip_execlp, and
    I still don't see the reason. Can you explain it me?<br>
    <br>
    On the other hand, I wanted to check if a solution violates some
    constraints (something like SCIPchecksol). I can get the variables
    of a constraint, but I haven't found any procedure to get the
    coefficient of a variable in a given constraint. Is there an easy
    way to get it?<br>
    <br>
    Best Regards,<br>
    Markó Horváth
  </body>
</html>