Dud*_*Wah 5 optimization matlab mathematical-optimization numerical-methods gradient-descent
我正在使用最速下降法来找出具有5x5希尔伯特矩阵的线性系统的解.我相信代码很好,因为它给了我正确的答案.
我的问题是:
我认为这需要太多的迭代才能得到正确的答案.我相信我可能错过了算法中的一些东西,但我不知道此时是什么.
我不确定这是否是实现算法的最有效方法,另外,选择"tol"有点令人困惑.
任何有关这些的见解将不胜感激(尤其是1.).谢谢!
% Method of Steepest Descent with tol 10^-6
h = hilb(5); %Hilbert 5x5 matrix
b = [1;1;1;1;1]; %solution matrix
solution = zeros(d,1); %Initialization
residual = h*solution - b;
tol = 10^(-6)
count = 0;
while residual'*residual > tol;
roe = (residual'*residual)/(residual'*h*residual);
solution = solution - roe*residual;
residual = h*solution - b;
count = count + 1;
end
count
solution
%Method of Steepest Descent with tol 10^-12
solution = zeros(d,1);
residual = h*solution - b;
tol = 10^(-12)
count = 0;
while residual'*residual > tol;
roe = (residual'*residual)/(residual'*h*residual);
solution = solution - roe*residual;
residual = residual - roe*h*residual;
count = count + 1;
end
count
solution
%another_solution = invhilb(5)*b %Check for solution
Run Code Online (Sandbox Code Playgroud)
我没有从数学方面处理你的问题的知识,但从编程的角度来看,我可以注意到一点。
确实你是对的。它需要太多迭代才能得到结果:
Elapsed time is 6.487824 seconds.
count =
292945
Run Code Online (Sandbox Code Playgroud)
当您定义步长以接近最终结果时,必须优化步长。否则,你要么接近答案的速度太慢,要么因为步长太大而多次经过最佳答案并围绕它进行锯齿形调整。
为了优化步长,我首先根据您的脚本形成一个函数(加上一些小的修改):
function [solution, count, T] = SDhilb(d, step)
h = hilb(d);
tic
b = ones(d, 1);
solution = zeros(d, 1);
residual = h * solution - b;
res2 = residual' * residual;
tol = 10^(-6);
count = 0;
while res2 > tol;
roe = res2/(residual' * h * residual);
solution = solution - step * roe * residual; % here the step size appears
residual = h * solution - b;
count = count + 1;
res2 = residual' * residual; % let's calculate this once per iteration
end
T = toc;
Run Code Online (Sandbox Code Playgroud)
现在将此函数用于一系列步长 (0.1:0.1:1.3) 和几个希尔伯特矩阵 (d = 2, 3, 4, 5),很明显这1不是一个有效的步长:
相反,step = 0.9似乎效率要高得多。让我们看看在以下情况下它的效率如何hilb(5):
[result, count, T] = SDhilb(5, 0.9)
result =
3.1894
-85.7689
481.4906
-894.8742
519.5830
count =
1633
T =
0.0204
Run Code Online (Sandbox Code Playgroud)
速度快了两个数量级以上。
以类似的方式,您可以尝试不同的值,tol看看它对速度的影响有多大。在这种情况下,没有最佳值:容差越小,需要等待的时间就越长。