optimizer.m

optimization toolbox

function sys = optimizer(F,h,ops,x,u)
%OPTIMIZER  Container for optimization problem
%
%   [OPT,PROBLEM] = OPTIMIZER(F,h,options,x,u) exports an object that
%   contains precompiled numerical data to be solved for varying arguments
%   x, returning the optimal value of the variable u.
%
%   OPTIMIZER typically only makes sense if the varying data x enters the
%   optmization problem affinely.
%
%   Example
%    The following problem creates an LP with varying upper and lower
%    bounds on the decision variable.
%
%    The optimizing argument is obtained by indexing the optimizer object
%    with the point of interest. The argument should be a column vector (if
%    the argument has a width larger than 1, YALMIP assumes that the
%    optimal solution should be computed in several points)
%   
%     A = randn(10,3);
%     b = rand(10,1)*19;
%     c = randn(3,1);
%
%     z = sdpvar(3,1);
%     sdpvar UB LB
%
%     F = set(A*z < b) + set(LB < z < UB);
%     h = c'*z
%     optZ = optimizer(F,h,[],[LB; UB],z);
%     
%     % Solve for LB=1, UB = 3;
%     zopt = optZ([1; 3])
%
%     % Solve for (LB,UB) [1;3] and (LB,UB) [2;6]
%     zopt = optZ([[1; 3], [2;6]])

if nargin < 5
    error('OPTIMIZER requires 5 inputs');
end

x = x(:);
n = length(x);
[aux1,aux2,aux3,model] = export(set(x == repmat(pi,n,1))+F,h,ops,[],[],0);

if norm(model.F_struc(1:n,1)-repmat(pi,length(x),1),inf) > 1e-10
    error('Failed exporting the model (try to specify another solver)')    
end

map = [];
for i = 1:length(u)
    var = getvariables(u(i));
    map = [map;find(var == model.used_variables)];
end

sys.recover = aux2;
sys.model = model;
sys.n = n;
sys.map = map;
sys = class(sys,'optimizer');