5.19 fmincon

Function File: [x obj info iter nf lambda] = fmincon(objf, x0, A, b, Aeq, beq, lb, ub)

Wrap basic functionality of Matlab’s solver fmincon to Octave’s sqp.
Non-linear constraint functions provided by fmincon’s function handle nonlincon are NOT processed.
Return codes are also mapped according to fmincon expected return codes. Note: This function mimics the behaviour of fmincon only.
In order to speed up minimizing, a bounded minimization algorithm fminbnd is used in a first try. If this fails, sqp algorithm is called.

Matlab:

A*x <= b
Aeq*x = beq
lb <= x <= ub

Octave:

g(x) = -Aeq*x + beq = 0
h(x) = -A*x + b >= 0
lb <= x <= ub

See the following example:

[x obj info iter] = fmincon (@(x)100*(x(2)-x(1)^2)^2 + (1-x(1))^2,[0.5,0],[1,2],1,[2,1],1)
x = [0.41494,0.17011]
obj =  0.34272
info =  1
iter =  6

Explanation of Input Parameters:

• objf: pointer to objective function
• x0: initial values
• A: inequality constraint matrix
• b: inequality constraint vector
• Aeq: equality constraint matrix
• beq: equality constraint vector
• lb: lower bound (required for fminbnd, defaults to -10)
• ub: upper bound (required for fminbnd, defaults to 10)

See also: sqp.