SeDuMi Interface

A tool for solving LMI problems with SeDuMi

Yann labit,
Dimitri Peaucelle
Didier Henrion

LAAS-CNRS, Toulouse, France

 Motivation

 Importance of Linear matrix Inequalities (LMI)

in Control theory and applications
 Limitations of existing tools:
– Problem size
– Computation time
– Convivial, relevant display
 Build together an adapted software

2
 Solver & Interface objectives
 Selected solver : SeDuMi.
 Necessity to write an Interface.
 Not a GUI tool.

3
 SeDuMi : not ideal nor definitive but
 Low computational complexity
 Sparse data format
 Works with Matlab
 Free software
 LMI and equality constraints
Jos F. Sturm  Complex valued constraints
 Work in progress

4
 Interface to build canonical
expressions for optimisation tools

LMIs and LMEs are Optimisation tools use vectors of

matrix dependent decision variables
m
expressions.
p = ( p1  p m ) F0 + Σ Fi p i  0
i =0

E.g.
 − 2a21 − a11 − a22 
 a11 a12   p1 p2  F0 = 0 F2 = 
A= P= − a11 − a22 − 2a12 
a21 a22  p 2 p 3  − 2a11 − a12   0 − a21 
F1 =  F3 = 
A′P + PA 0  − a12 0  − a21 − 2a22 


5
 Existing Interfaces/Parsers

 LMIlab,
LMITOOL, sdpsol, YALMIP
 Some critiques
– Lack of functionalities (LMIlab, sdpsol)
– Slow conversion (LMITOOL,sdpsol)
– Difficulties to create Matlab functions to generate
LMI problems (LMITOOL, sdpsol)
– Difficulties to analyse the obtained solution…
not a GUI and no symbolic declarations

6
 Interface dedicated to LMIs
 Simplified declaration
– Structured variables : symmetric, diagonal, Hermitian…
– Structured constraints : block decomposition, Kronecker…
– Predefined objectives : trace, log(det)…
– Analysis of the solution : margin on constraints, matrix
format...
 Software constraints
– Speed : simple algebaric manipulations
– Memory space : sparse format
– Open to modifications : Matlab free source code

7
 Using SeDuMi Interface :
An example

H ∞ - state feedback
max γ
Q = Q'  0
 AQ + QA'+ BY + Y ' B'+ B B ' QC∞' + Y' D∞' 
 ∞ ∞
0
 C∞ Q + D∞ Y γI



−1
Optimal controller K = YQ Optimal norm −γ

8
 Step 1 : Name the LM problem

 The LM problem = a Matlab object

>> quiz = sdmpb(‘Hinfty state feedback’)
LM problem: Hinfty state feedback
no matrix variable
no equality constraint
no inequality constraint
no linear objective
Unsolved
>> whos quiz
Name Size Bytes Class
quiz 1x1 3358 sdmpb object

9
 Step 2 : declare matrix variables
>> [quiz, Yindex] = sdmvar(quiz, m, n, 'Y');
>> [quiz, gindex] = sdmvar(quiz, 1, 1, 'gamma');
>> [quiz, Qindex] = sdmvar(quiz, n, 's', 'Q=Q’’')
LM problem: Hinfty state feedback
matrix variables: index name
1 Y
2 gamma
3 Q=Q’
no equality constraint
no inequality constraint
no linear objective
unsolved

Symmetric, diagonal, hermitian, structured…

10
 Step 3 : declare inequalities
>> [quiz, lmi1] = sdmlmi(quiz, n, 'Q>0');
>> quiz = sdmineq(quiz, -lmi1, Qindex);

>> [quiz, lmi2] = sdmlmi(quiz, [n p], 'Hinfty');

>> quiz = sdmineq(quiz, [lmi2 1 1], Qindex, A, 1);
>> quiz = sdmineq(quiz, [lmi2 1 1], Yindex, B, 1);
>> quiz = sdmineq(quiz, [lmi2 1 1], 0, Bi);
>> quiz = sdmineq(quiz, [lmi2 2 1], Qindex, Ci, 1);
>> quiz = sdmineq(quiz, [lmi2 2 1], Yindex, Di, 1);
>> quiz = sdmineq(quiz, [lmi2 2 2], gindex);

Symmetric terms automatically added

Possible to define Kronecker products

11
 Step 4 : declare the objective
>> quiz = sdmobj(quiz, gindex, 1, 1, ‘gamma’)
LM problem: Hinfty state feedback
matrix variables: index name
1 Y
2 gamma
3 Q=Q’
no equality constraint
inequality constraints: index meig name
1 -Inf Q>0
2 -Inf Hinfty
maximise objective: gamma
unsolved

Possible to define : el Xer or trace( MX)

12
 Step 5 : solve the LM problem
>> quiz = sdmsol(quiz)
... computation ...
LM problem: Hinfty state feedback
matrix variables: index name
1 Y
2 gamma
3 Q=Q’
no equality constraint
inequality constraints: index meig name
1 eps Q>0
2 eps Hinfty
maximise objective: gamma = -27.3
feasible

13
 Step 6 : analyse the result

Feasibility margins on each constraints:

inequality constraints: index meig name
1 eps Q>0
2 eps Hinfty
maximise objective: gamma = -27.3
feasible

Matrix formatted result:

>> K = quiz(Yindex)/quiz(Qindex)
K = 1.0e+03 *
-2.5989 -8.5124 -0.1898 -4.0176

14
 Other features of the Interface

 Matrixequalities
 Complex valued constraints
 Complex valued variables
 Radius on the vector of decision variables
 Adapted tuning of SeDuMi options

Acknowledgements to K. Taitz

15
 Future evolutions
 Warm-start with feasible solution
 Pre-conditioning of the optimisation problem
 Other predefined options (user’s feedback)
 Platform incorporating other solvers
 Predefined LM problems for Automatic control
– Robust analysis
– Performance (pole location, H ∞ , H 2 …)
– State and Output feedback
– …

16
 Conclusion

http://www.laas.fr/~peaucell/sedumiint

mailto:[email protected]

17