matlab中Funval(f,var,x)如何定义

Question

如题所述

Answer 1

%初始搜索点: x;%线性无关的初始向量组: P;%自变量向量: var%精度: eps;%目标函数取最小值时的自变量值: x;%目标函数的最小值: minf;format long;if nargin==3eps=1.0e-6;endn=length(var)+1;syms l;while 1y=zeros(size(P));y(:,1)=x0;for i=1:n-1 %在每个搜索方向上进行一维搜索yv=y(:,i)+l*P(:,i);fy=Funval(f,var,yv);[a,b]=minJT(fy,0,0.1);tl=minHJ(fy,a,b);y(:,i+1)=y(:,i)+tl*P(:,i);endP(:,n)=y(:,n)-y(:,1);if norm((P(:,n))<=eps) %精度判断x=y(:,n);break;elsefor j=1:nFY(j)=Funval(f,var,y(:,j));endmaxDF=-inf;m=0;for j=1:n-1 %求出算法中对应的mdf=FY(j)-FY(j+1);if df>maxDFmaxDF=df;m=j+1;endendtmpF=Funval(f,var,2*y(:,n)-y(:,1));fl=FY(1)-2*FY(n)+tmpF;if fl<2*maxDFyv=y(:,n)+l*P(:,n);fy=Funval(f,var,yv);[a,b]=minJT(fy,0,0.1);tl=minHJ(fy,a,b);x0=y(:,n)+tl*P(:,n);P(:,m:(n-1))=P(:,(m+1):n); %重新设置搜索方向elsex0=y(:,n);endendendminf=Funval(f,var,x);format short;powell方法解决无约束最优化问题