1 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> |
---|
2 | <html> |
---|
3 | |
---|
4 | <head> |
---|
5 | <meta http-equiv="Content-Language" content="en-us"> |
---|
6 | <title>YALMIP Example : Basics</title> |
---|
7 | <meta http-equiv="Content-Type" content="text/html; charset=windows-1251"> |
---|
8 | <link href="yalmip.css" type="text/css" rel="stylesheet"> |
---|
9 | <base target="_self"> |
---|
10 | </head> |
---|
11 | |
---|
12 | <body leftmargin="0" topmargin="0"> |
---|
13 | |
---|
14 | <div align="left"> |
---|
15 | <table border="0" cellpadding="4" cellspacing="3" style="border-collapse: collapse" bordercolor="#000000" width="100%" align="left" height="100%"> |
---|
16 | <tr> |
---|
17 | <td width="100%" align="left" height="100%" valign="top"> |
---|
18 | <h2>Basics: The sdpvar and set object</h2> |
---|
19 | <hr size="1" color="#000000"> |
---|
20 | <p>The most important command in YALMIP is <a href="reference.htm#sdpvar"> |
---|
21 | sdpvar</a>. This command is used to the define decision variables. To define |
---|
22 | a matrix (or scalar) <b>P</b> with height <b>n</b> and width <b>m</b>, we |
---|
23 | write</p> |
---|
24 | <table cellpadding="10" width="100%"> |
---|
25 | <tr> |
---|
26 | <td class="xmpcode"> |
---|
27 | <pre>P = sdpvar(n,m)</pre> |
---|
28 | </td> |
---|
29 | </tr> |
---|
30 | </table> |
---|
31 | <p><font color="#FF0000">A square matrix is symmetric by default!</font>. To obtain a fully parameterized |
---|
32 | square matrix, a third argument is needed.</p> |
---|
33 | <table cellpadding="10" width="100%"> |
---|
34 | <tr> |
---|
35 | <td class="xmpcode"> |
---|
36 | <pre>P = sdpvar(3,3,'full')</pre> |
---|
37 | </td> |
---|
38 | </tr> |
---|
39 | </table> |
---|
40 | <p>The third argument can be used to obtain a number of pre-defined types |
---|
41 | of variables, such as Toeplitz, Hankel, symmetric and skew-symmetric matrices. |
---|
42 | See the help text on <a href="reference.htm#sdpvar">sdpvar</a> for details. |
---|
43 | Alternatively, the associated standard commands can be applied to a |
---|
44 | suitable vector</p> |
---|
45 | <table cellpadding="10" width="100%" id="table2"> |
---|
46 | <tr> |
---|
47 | <td class="xmpcode"> |
---|
48 | <pre>x = sdpvar(n,1); |
---|
49 | D = diag(x) ; % Diagonal matrix |
---|
50 | H = hankel(x); % Hankel matrix |
---|
51 | T = toeplitz(x); % Hankel matrix</pre> |
---|
52 | </td> |
---|
53 | </tr> |
---|
54 | </table> |
---|
55 | <p>Scalars can be defined in three different ways.</p> |
---|
56 | <table cellpadding="10" width="100%" id="table1"> |
---|
57 | <tr> |
---|
58 | <td class="xmpcode"> |
---|
59 | <pre>x = sdpvar(1,1); y = sdpvar(1,1); |
---|
60 | x = sdpvar(1); y = sdpvar(1); |
---|
61 | sdpvar x y</pre> |
---|
62 | </td> |
---|
63 | </tr> |
---|
64 | </table> |
---|
65 | <p>The <a href="reference.htm#sdpvar">sdpvar</a> objects are manipulated in |
---|
66 | MATLAB as any other variable and (almost)<font color="#FF0000"> all standard |
---|
67 | functions are overloaded</font>. Hence, the following commands are valid</p> |
---|
68 | <table cellpadding="10" width="100%"> |
---|
69 | <tr> |
---|
70 | <td class="xmpcode"> |
---|
71 | <pre>P = sdpvar(3,3) + diag(sdpvar(3,1)); |
---|
72 | X = [P P;P eye(length(P))] + 2*trace(P); |
---|
73 | Y = X + sum(sum(P*rand(length(P)))) + P(end,end)+hankel(X(:,1));</pre> |
---|
74 | </td> |
---|
75 | </tr> |
---|
76 | </table> |
---|
77 | <p>To define constraints, the command <a href="reference.htm#set">set</a> |
---|
78 | is used (with set meaning set as in convex set, non-convex set, set of integers |
---|
79 | etc, not as in set/get). The meaning of a constraint is context-dependent. |
---|
80 | If left-hand side and right-hand side are Hermitian, the constraint is interpreted |
---|
81 | in terms of positive definiteness, otherwise element-wise. Hence, declaring |
---|
82 | a symmetric matrix and a positive definiteness constraint is done with</p> |
---|
83 | <table cellpadding="10" width="100%"> |
---|
84 | <tr> |
---|
85 | <td class="xmpcode"> |
---|
86 | <pre>n = 3; |
---|
87 | P = sdpvar(n,n); |
---|
88 | F = set(P>0);</pre> |
---|
89 | </td> |
---|
90 | </tr> |
---|
91 | </table> |
---|
92 | <p>while a symmetric matrix with positive elements is defined with, e.g.,</p> |
---|
93 | <table cellpadding="10" width="100%"> |
---|
94 | <tr> |
---|
95 | <td class="xmpcode" > |
---|
96 | <pre>P = sdpvar(n,n); |
---|
97 | F = set(P(:)>0);</pre> |
---|
98 | </td> |
---|
99 | </tr> |
---|
100 | </table> |
---|
101 | <p>According to the rules above, a non-square matrix with positive elements |
---|
102 | can be defined using the > operator immediately</p> |
---|
103 | <table cellpadding="10" width="100%"> |
---|
104 | <tr> |
---|
105 | <td class="xmpcode"> |
---|
106 | <pre>P = sdpvar(n,2*n); |
---|
107 | F = set(P>0);</pre> |
---|
108 | </td> |
---|
109 | </tr> |
---|
110 | </table> |
---|
111 | <p>A list of several constraints is defined by just adding |
---|
112 | <a href="reference.htm#set">set</a> objects.</p> |
---|
113 | <table cellpadding="10" width="100%"> |
---|
114 | <tr> |
---|
115 | <td class="xmpcode"> |
---|
116 | <pre>P = sdpvar(n,n); |
---|
117 | F = set(P>0) + set(P(1,1)>2);</pre> |
---|
118 | </td> |
---|
119 | </tr> |
---|
120 | </table> |
---|
121 | <p>Of course, the involved expressions can be arbitrary |
---|
122 | <a href="reference.htm#sdpvar">sdpvar</a> objects, and equality constraints |
---|
123 | (==) can be defined, as well as constraints using <.</p> |
---|
124 | <table cellpadding="10" width="100%"> |
---|
125 | <tr> |
---|
126 | <td class="xmpcode"> |
---|
127 | <pre>F = set(P>0) + set(P(1,1)<2) + set(sum(sum(P))==10);</pre> |
---|
128 | </td> |
---|
129 | </tr> |
---|
130 | </table> |
---|
131 | <p>In fact, non-strict operators =< and >= may also be used (by default, there |
---|
132 | is no difference, but by using the option <code>shift</code> in |
---|
133 | <a href="reference.htm#sdpsettings">sdpsetttings</a>, it is possible to |
---|
134 | aim for |
---|
135 | strict feasibility). Note though that no solver can distinguish between |
---|
136 | strict and non-strict constraints. In fact, most solvers will not even |
---|
137 | respect a non-strict constraint but often return slightly infeasible |
---|
138 | solutions.</p> |
---|
139 | <table cellpadding="10" width="100%"> |
---|
140 | <tr> |
---|
141 | <td class="xmpcode"> |
---|
142 | <pre>F = set(P>=0) + set(P(1,1)<=2) + set(sum(sum(P))==10);</pre> |
---|
143 | </td> |
---|
144 | </tr> |
---|
145 | </table> |
---|
146 | <p>Finally, a convenient way to define several constraint is to use double-sided |
---|
147 | constraints.</p> |
---|
148 | <table cellpadding="10" width="100%"> |
---|
149 | <tr> |
---|
150 | <td class="xmpcode"> |
---|
151 | <pre>F = set(0 < P(1,1) < 2);</pre> |
---|
152 | </td> |
---|
153 | </tr> |
---|
154 | </table> |
---|
155 | </td> |
---|
156 | </tr> |
---|
157 | </table> |
---|
158 | </div> |
---|
159 | |
---|
160 | </body> |
---|
161 | |
---|
162 | </html> |
---|