Kunena Site Syndication http://sedumi.ie.lehigh.edu/ Tue, 21 May 2013 13:39:03 -0400 Kunena 1.5.12 http://sedumi.ie.lehigh.edu/components/com_kunena/template/default_ex/images/english/emoticons/rss.gif http://sedumi.ie.lehigh.edu/ Kunena Site Syndication http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=16&id=218177#218178 I am not going to give the SeDuMi numerical format, because that is a mess. Most people use an interface such as YALMIP or CVX to create models for SeDuMi. If you use YALMIP, you would do A = sdpvar(n,n); B = sdpvar(n,n,'full'); Model = [A >= 0, [A B/2;B'/2 A] >= 0]; If you want to look at the model that would have been sent to sedumi, use export (here assuming objective is trace(A)) sedumimodel = export(Model,trace(A),sdpsettings('solver','sedumi')) Fri, 17 May 2013 12:59:44 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=16&id=218176#218176 Hi Everyone, I'm new using sedumi and I would like to know how to use "K.ycomplex" or/and "K.xcomplex", since the HELP of sedumi is not clear. I have the next model. min b'x st: D'x+f>=0 || A*x+c || Thu, 16 May 2013 12:56:58 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=16&id=218172#218173 There is no way (or reason) to define predefine F or something like that. What takes time is not the creation of the elements of F, but the matrix multiplications etc. Your code look correct. Next time, please post YALMIP questions on the YALMIP forum https://groups.google.com/forum/?fromgroups#!forum/yalmip Sun, 12 May 2013 02:04:01 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=19&id=7274&limit=6&limitstart=12#218171 just bumping this to the top Wed, 08 May 2013 10:12:03 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=19&id=218169#218170 Please post YALMIP specific questions on the YALMIP forum https://groups.google.com/forum/?fromgroups#!forum/yalmip Wed, 08 May 2013 10:10:36 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=2&id=7211&limit=6&limitstart=12#218168 Hello, I was trying to install SEDUMI on my computer, hence I downloaded it from the website, added and set the path of its folder and run the command install_sedumi. This needs the program to run some C compilation, but Matlab cannot find any C compiler. It returns the following error: No supported SDK or compiler was found on this computer. For a list of supported compilers, see http://www.mathworks.com/support/compilers/R2013a/win64.html Error using mex (line 206) Unable to complete successfully. Error in install_sedumi (line 113) eval( temp ); I installed some of the compilers suggested in the website, like Microsoft SDK 7.1, Microsoft Visual C++ 2008 Express and Java Development Kit (JDK) 1.6 and added them to the path, but the error is always the same. Can you help me? Thank you and kind regards, Giampaolo Torrisi Tue, 07 May 2013 12:50:23 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=16&id=218166#218167 a complex-valued problem is converted internally to a real-valued anyway (note that sedumi 1.3 has a bug which causes problems on complex-valued problems) Tue, 07 May 2013 11:48:18 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=19&id=218165#218165 Is it more effecient to solve a real-valued problem than complex-valued problem using SeDuMi? Tue, 07 May 2013 09:37:19 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=16&id=7670#218133 First, SET is obsolete, so please don't use that http://users.isy.liu.se/johanl/yalmip/pmwiki.php?n=Commands.Set Your problem lies in the horrible numerics of your model. Most importantly the data matrix Ckk, which you probably think is symmetric. Look at Ckk-Ckk' for the two cases. In the first case, it has residuals on the order 10^-13, which will be cleaned away by YALMIP when YALMIP decides symmetry etc (YALMIP uses an error tolerance of roughly 10^-12), while in the second case it is 10^-9, which YALMIP assumes is intended data and thus, your matrix Amat is deemed unsymmetric by YALMIP, and the constraint Amat>=0 is interpreted as elementwise constraints First case [code] >> F +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | ID| Constraint| Type| +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | #1| Numeric value| Element-wise inequality 1x1| | #2| Numeric value| Matrix inequality (complex) 4x4| | #3| Numeric value| Matrix inequality (complex) 8x8| +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ [/code] Second case [code] >> F ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | ID| Constraint| Type| ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | #1| Numeric value| Element-wise inequality 1x1| | #2| Numeric value| Matrix inequality (complex) 4x4| | #3| Numeric value|Element-wise inequality (complex) 52x1| ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++[/code] Wed, 24 Apr 2013 01:54:50 -0400 http://sedumi.ie.lehigh.edu/index.php?option=com_kunena&Itemid=78&func=view&catid=16&id=7664&limit=6&limitstart=6#218132 Please post reproducible code to illustrate the issue. Wed, 24 Apr 2013 01:45:09 -0400