what's the difference

Question

zilai si 2019-5-6

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/460431-what-s-the-difference

评论： Walter Roberson 2019-5-7

I want to piecewise plot a interlolation function. When I input the nodes like this

X = -1:0.1:1;
N = length(X);

the graph is continuous like this

However When I input the nodes like this :

X = zeros(1,21);
N = 21;
for i = 1: 21
X(i) = -1 + 2*(i-1)/(N-1);

there is an obvious gap in the graph like this :

I suppose these two codes are exactly the same, why the graphs are different??

the codes are as follows:

tic
clc
clear
f = @(x)1./(1+25.*x.^2);
syms x 
f_dri = diff(f(x));
%X = -1:0.1:1;
%N = length(X);
 X = zeros(1,21);
 N = 21;
    F = zeros(1, N);
    F_dri = zeros(1, N);
    for i = 1 : N
       X(i) = -1 + 2*(i-1)/(N-1);
        F(i) = f(X(i));
        F_dri(i) = subs(f_dri, x, X(i));
    end
    for j = 1 : N-1
        M = zeros(4,4);
        M(1,1) = F(j);
        M(2,1) = F(j);
        M(3,1) = F(j+1);
        M(4,1) = F(j+1);
        M(2,2) = F_dri(j);
        M(3,2) = (F(j+1)-F(j))/(X(j+1)-X(j));
        M(4,2) = F_dri(j+1);
        M(3,3) = (M(3,2)-M(2,2))/(X(j+1)-X(j));
        M(4,3) = (M(4,2)-M(3,2))/(X(j+1)-X(j));
        M(4,4) = (M(4,3)-M(3,3))/(X(j+1)-X(j));
    f_inp = M(1,1)+(x-X(j))*M(2,2)+(x-X(j))^2*M(3,3)+(x-X(j+1))*(x-X(j))^2*M(4,4);
    a = X(j):0.01:X(j+1);
    b = subs(f_inp, x, a);
    G = plot(a, b, 'r', 'LineWidth', 2);
    hold on 
    end
     saveas(gcf, 'pic_2', 'jpg'); 
% r = -1:0.01:1;
% f_ori = f(r);
% G(5) = plot(r, f_ori, 'k', 'LineWidth', 2);
legend(G,'N=21');
toc

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2019-5-6

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/460431-what-s-the-difference#answer_373719

X(13) is eps/4 less than 0.2 . When you use X(j):0.01:X(j+1) that is just enough that the previous plot ends at 0.19 instead of 0.20

Always remember that when you calculate your endpoints using floating point numbers that they will usually not be **exactly* nice multiples of 1/10 or 1/100 .

2 个评论
显示无隐藏无

zilai si 2019-5-7

thanks, i undersatand. But how to solve such problems?

Walter Roberson 2019-5-7

You have a fundamental mismatch of expecting that breaking up the range 0 to 1 into (N-1) subdivisions (N total points including both endpoints) will give you ranges that are nicely divisible into increments of 0.01

Instead of using X(j):0.01:X(j+1) consider using a = linspace(X(j), X(j+1), 11)

请先登录，再进行评论。