Question

我需要在matlab中使用quadprog来最小化alpha

1/2*alpha.'*H*alpha+(-1.')*alpha

subject to: y.'alpha=0 and 0<=alpha<=inf

我已经制作了矩阵H

    for a=1:8

    for b=1:8

    H(a,b)=y(a)*y(b)*dot((x(a)).',x(b));

    end

end

但我不确定是否必须制定约束

Answer 1

这有用吗？

n = size(H,1);
f = -ones(n,1);       // linear term
aeq = randn(1,n);     // equality constraint
lb = zeros(n,1);      // lower bound
ub = inf * ones(n,1); // upper bound
alpha = quadprog(H,f,[],[],aeq,0,lb,ub)

线性约束的二次规划（Matlab）

1 个答案: