Optimization Toolbox of
MATLAB
Optimization of Engineering Systems
Topics
• MATLAB Toolbox Solvers
• Linear Programming
• Nonlinear unconstrained
• Nonlinear constrained
MATLAB Toolbox Solvers
MATLAB Optimization Toolbox Solvers
MATLAB Optimization Toolbox Solvers
MATLAB Optimization Toolbox Solvers
• fmincon (constrained nonlinear minimization)
• Trust‐region‐reflective (default)
• Allows only bounds or linear equality constraints, but not both.
• Active‐set (solve Karush‐Kuhn‐Tucker (KKT) equations and used quasi‐Newton
method to approximate the hessian matrix)
• Interior‐point
• Sequential Quadratic Programming (SQP)
• fminunc (unconstrained nonlinear minimization)
• Large‐Scale Problem: Trust‐region method based on the interior‐reflective
Newton method
• Medium–Scale: BFGS Quasi‐Newton method with a cubic line search
procedure.
Linear Programming
Linear Programming
• Command syntax
[x, fval, exitflag, output] = linprog (f, A, b, Aeq, beq, lb, [], [], optimset
('Display','iter'));
• Vector f defines objective function coefficients. Matrix A and vector b define
linear inequality constraints
• as Ax<=b. Matrix Aeq and vector beq define linear equality constraints as Aeq
x <= beq. Lower bounds on
• variables are defined as vector lb.
• f, b, and beq must be column vectors. Lower bounds, even when they are
zero, must also be specified.
Example
Solution
9
Example (Contd…)
MATLAB Output
Nonlinear Unconstrained
Nonlinear Constrained
• Command syntax
Nonlinear Constrained (Contd…)
• Without gradient
Nonlinear Constrained (Contd…)
• With gradient
Nonlinear Constrained (Contd…)
• With gradient
Nonlinear Constrained
Nonlinear Constrained
• Command syntax
• Objective function is defined as a function file ‘fun’
• x0 is vector of starting solution
• Matrix A and vector b define linear inequality constraints as Ax<=b. Matrix
Aeq and vector beq define linear equality constraints as Aeq x <= beq.
• Lower and upper bounds on variables are defined as vectors lb and ub
• Nonlinear constraints are defined in a function file ‘nonlcon’.
• Input arguments to the function is vector x.
• Output must be two vectors, one defining inequality constraints and the
other equality constraints.
Nonlinear Constrained(Contd…)
• Without gradient
Nonlinear Constrained(Contd…)
• Without gradient
Nonlinear Constrained(Contd…)
• Without gradient
Nonlinear Constrained(Contd…)
• With gradient
Nonlinear Constrained(Contd…)
• With gradient
Nonlinear Constrained(Contd…)
• With gradient
THANKS
