faq.htm

optimization toolbox

    more efficient and general solver. Future versions of YALMIP may resolve 
	this issue. Constraints defined using the <a href="kyp.htm">KYP</a> operator 
	is efficiently handled in some cases already.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="cddhangs"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b><a href="solvers.htm#penbmi">
        CDD</a> hangs</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0">Try 
    <code>sdpsettings('cdd.method','dual-simplex')</code></p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%" id="table2">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="fminconcrashes"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b><a href="solvers.htm#fmincon">
        fmincon</a> crashes</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0">Do you have <a href="solvers.htm#mosek">MOSEK</a> 
	installed? This can cause problems due to an inconsistency between MATLABs 
	and <a href="solvers.htm#mosek">MOSEKs</a> implementation of the file 
	optimget.m. Remove <a href="solvers.htm#mosek">MOSEK</a> 
	from your path.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%" id="table4">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="maxdetcrashes"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b><a href="solvers.htm#maxdet">
        MAXDET</a> fails</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0"><a href="solvers.htm#maxdet">
        MAXDET</a> seems to be sensitive to unbounded feasible regions and 
	unconstrained variables. Try to add redundant bound constraints on all your 
	variables.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="xpress"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b><a href="solvers.htm#xpress">
        XPRESS</a> performs badly, claims infeasibility etc.</a></b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0">Adding bounds on the involved 
    variables solves this issue in many cases.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="selectsolver"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>There are so many solvers, which 
    one should I use?</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"> <a href="solvers.htm#sedumi">SeDuMi</a> 
        and
        <a href="solvers.htm#sdpt3">SDPT3</a> are good general purpose SDP 
    solvers (not necessarily the 
    best solvers though), efficient also on LP and SOCP problems and reasonably 
        efficient on small QP problems. Hans D. Mittelmanns
        <a target="_blank" href="http://www.optimization-online.org/DB_HTML/2001/07/358.html">benchmark</a> might be helpful. 
        If you mainly solve LPs, make sure to try the free solvers
        <a href="solvers.htm#glpk">GLPK</a>,
        <a href="solvers.htm#qsopt">QSOPT</a>,
        <a href="solvers.htm#clp">CLP</a> and
        <a href="solvers.htm#cdd">CDD</a></td>
      </tr>
    </table>
    <p style="margin-top: 0; margin-bottom: 0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="infeasible"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>The solution I get in an SDP is not 
    feasible but has eigenvalues around, say, -1e-6.</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0">Most solvers actually use 
    infeasible/exterior algorithms, so slightly infeasible 
    solutions are common.</p>
        </td>
      </tr>
    </table>
    <p style="margin-top: 0; margin-bottom: 0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="bmi"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>Can I solve BMIs without
    <a href="solvers.htm#penbmi">PENBMI</a>?</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0">The BMI-examples in
    <a href="reference.htm#yalmipdemo">yalmipdemo</a> show some alternative 
    ways to code your own solver rather easily, but for performance and 
    robustness,
    <a href="solvers.htm#penbmi">PENBMI</a> is highly recommended.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <hr noShade SIZE="1" color="#FF0000">        
    <p style="margin-top: 0; margin-bottom: 0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="typical"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>My solution is not what I 
		expected.</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0">Is your problem what you expected? 
	Use the command checkset to see that you actually have the constraints that 
	you meant to declare (does it say matrix inequality or element-wise 
	inequality etc). </p>
	<p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
	<p style="margin-bottom: 0; margin-top:0">Are your variables really what you 
	meant to declare (display them to see if they are symmetric, Hermitian, 
	full, etc.). </p>
	<p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
	<p style="margin-bottom: 0; margin-top:0">Most common error is that you have 
	declared a square matrix but accidentally forgotten to declare it as full 
	and obtained a symmetric matrix instead (square matrices are full by 
	default!)</p></td>
      </tr>
    </table>
    <p style="margin-top: 0; margin-bottom: 0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%" id="table15">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="typeset"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>Typing <code>help set</code> gives 
    me no information on the YALMIP function <code>set</code>.</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
    <p style="margin-bottom: 0; margin-top:0">Yep, a bit tricky since <code>set</code> also is a built-in 
    function. Type <code>help sdpvar/set</code> and you will find what you are 
    looking for.</p></td>
      </tr>
    </table>
    <p style="margin-top: 0; margin-bottom: 0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="setclass"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>Is there really a <code>set</code> class?</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0">No...<code>set</code> is only a wrapper to 
        call the old class <code>lmi</code>. To much work to re-write all code 
        just for a name change.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="normalset"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>Is <code>set</code> related to 
        the &quot;normal&quot; set command in MATLAB</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0">No... The name <code>set</code> 
        was selected since it is short. An alternative would be to define 
        constraints, or feasible sets, using a command named, e.g., <code>
        constraint</code>. However, my keyboard typing speed is too slow to 
        allow for such a long command name. In my opinion, the name <code>set</code> 
        is at-least better than the old name <code>lmi</code>.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="lmielement"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>I define a 
      semidefinite constraint, but YALMIP declares it &quot;element-wise&quot;.</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0">YALMIP detects semidefinite 
        constraints by checking symmetry. In some cases (working with very 
        ill-conditioned data), numerical problems may lead to a small violation 
        of symmetry in MATLAB, and YALMIP will declare the constraint as 
        element-wise. To solve this problem, just symmetrize your variable 
        first.</p>
        </td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="reallystrict"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>Are  inequalities really 
    strict?</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0">By default, strict (&lt;,&gt;) and 
        non-strict (&lt;=,&gt;=) inequalities are treated in the same way in YALMIP, 
        and the result depends on the solver. However, by using the field <code>
        shift</code> in <a href="reference.htm#sdpsettings">sdpsetttings</a>, 
        inequalities defined using &lt; and &gt; will be treated slightly different. 
        YALMIP will add a small perturbation to these inequalities to 
        increase the likelihood of a strictly feasible solution. Warning : If you 
		have an integer variable, and add a constraint <code>set(x<2)</code>, this will not be interpreted as <code>set(x<=1)</code> 
		. To avoid confusion, use <code>set(x<=1)</code>. The same holds for rank constraints <code>set(rank(x)<2)<n)</code>.</code></td>
      </tr>
    </table>
    <p style="margin-bottom: 0; margin-top:0">&nbsp;</p>
    <table border="0" cellpadding="0" style="border-collapse: collapse" width="100%" id="table5">
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><b><font color="#0000FF">
        <a name="gevp"></a>Q:</font></b></td>
        <td>
        <p style="margin-bottom: 0; margin-top:0"><b>How do I solve generalized 
        eigenvalue problems (like <code>gevp</code> in <a href="solvers.htm#lmilab">LMILAB</a>)?</b></td>
      </tr>
      <tr>
        <td width="26" valign="top">
        <p style="margin-bottom: 0; margin-top:0"><font color="#0000FF"><b>A</b></font><b><font color="#0000FF">:</font></b></td>
        <td>