最优化之BFGS/DFP拟牛顿算法MATLAB代码（注释完整，附求解函数以及Armjio算法求步长）

function [x, k] = BFGS(xk, H0, kmax, fun, grd, Armjio) exp = 1e-6; k = 0; Hk = H0; gk = feval(grd, xk); while k <= kmax if norm(gk) < exp break; end dk = -Hk * gk; alpha = feval(Armjio, fun, grd, xk, dk); x_ = xk; %上一步xk xk = x_ + alpha * dk; g_ = gk; %上一步gk gk = feval(grd, xk); sk = xk - x_; yk = gk - g_; if sk' * yk > 0 v = yk' * sk; Hk = Hk + (1 + (yk' * Hk * yk) / v) * (sk * sk') / v - (sk * yk' * Hk + Hk * yk * sk') / v; end k = k + 1; end x = xk; end