matrix calculation have wrong answer

I have the following code,

clear;

time = datestr(now,'dd-mm-yyyy HH-MM-SS');
input=xlsread('Book1.xlsx');
load('beta0_1.mat');
load('kbeta_bao_hoa.mat');
load('x_delta.mat');
load('kI');
load('Ka');
Z1 = 48;
Z2 = 40;
x = [221.1321  336.6439  0.6575  5.3102  345.8133  7.6177  9.5410  19.3115  0.7990  8.8932  17.1360  8.8941  8.8954  30.9960  2.0003 ];
y_test = linspace(0.5,4,100);
z_test = linspace(1, 10, 100);
[X,Y] = meshgrid(y_test,z_test);
m_k = - m_k_ranhTQ_1(x,input,beta0,kbeta_bao_hoa,x_delta,kI,Ka,time,Z1,Z2,X,Y);
surf(X,Y,m_k);
hold on;
m_k_x = - m_k_ranhTQ_1(x,input,beta0,kbeta_bao_hoa,x_delta,kI,Ka,time,Z1,Z2,1.419,5.545);
plot3(0.7990,8.8932,m_k_x,'.r','markersize',10);

m_k_ranhTQ_1 is a function with 1 ouput. When i run this script, m_k is a 100x100 matrix. The problem is the result in the m_k matrix have wrong answers. Like you can see in the script, i try to highlight the point (1.419; 5.545) on the surface of the 3d graph, but it is not even on the surface, it is way off. I try to go to the Workspace to check manually, and typing in the command window like:

m_k_x = - m_k_ranhTQ_1(x,input,beta0,kbeta_bao_hoa,x_delta,kI,Ka,time,Z1,Z2,X(100,100),Y(100,100))

and compare it with m_k (100,100), and it is not the same answer. I think m_k (x,y) = f[X(x,y); Y(x,y)]. I check and if i use low index, like

m_k_x = - m_k_ranhTQ_1(x,input,beta0,kbeta_bao_hoa,x_delta,kI,Ka,time,Z1,Z2,X(30,16),Y(30,16))

In this case, m_k_x = m_k(30, 16)

Can someone tell me where i did wrong and how can i fix this?

[Bjorn Gustavsson]

How could we possibly? You will have to show the details of your m_k_ranhTQ_1 function for us to be able to help.

[quan ng]

I have attach all files needed to run the above script

[Catalytic]

I have attach all files needed to run the above script

Why so many? Why not put everything in one .mat file so it's easier for us to download?

[quan ng]

sorry, i use grab it to create it from image file, and i just left it to use

[quan ng]

i now use loop to calculate the m_k matrix:

count_i = 0;

count_j = 0;

for i = y_test

count_i = count_i + 1;

if count_j == 100

count_j = 0;

end

for j = z_test

count_j = count_j + 1;

m_k(count_i,count_j) = - m_k_ranhTQ_1(x,input,beta0,kbeta_bao_hoa,x_delta,kI,Ka,time,Z1,Z2,i,j);

end

end

surf(X,Y,m_k);

[quan ng]

now the m_k matrix is right (i checked it), but the surf is wrong

i use:

surf(X,Y,m_k);

hold on;

m_k_x = - m_k_ranhTQ_1(x,input,beta0,kbeta_bao_hoa,x_delta,kI,Ka,time,Z1,Z2,0.799,8.8932);

plot3(0.7990,8.8932,m_k_x,'.r','markersize',10);

and it is still not on the surface

I attach the graph for you to see

[Bjorn Gustavsson]

Another point you might consider is to write your code in "english".

It might seem a bit rude of me to suggest that your native tounge is not suitabel - but that's not what I insinuate. Once uppon a time I programmed in my native language, function and variable-names and comments. Then a colleague that did not speak my language tried to use/port the programme, which made me feel very rude. Since then I've stuck to working in english - even if neither of me nor my colleagues have it as a first language we can get by...

[quan ng]

thanks for your feedbak, next time i will try. Quick question though, is the "surf" function somehow try to alter the result to smooth the surface or something? I try surf and have values that is not in the matrix. I try to plot each points using loops and it is correct, while the graph of the surf function is not. I will attach the image of the point by point graph, you can compare with the surf.png i attach above. The red dot is the point i use to check if it is on the surface. If you can take a quick look, i will be gratefull.

[quan ng]

sorry i have to delete the .mat files so i can upload other question with attached files. Mathworks limit 10 files daily upload. If you can answer my question about the surf function, please go to:

https://www.mathworks.com/matlabcentral/answers/1654295-surf-function-give-wrong-graph

Answer 1

[Voss]

I think the difference you see between the case when you run m_k_ranhTQ_1() with matrix X,Y vs the case when you run it with scalar X,Y in for loops, is due to the logic near the bottom of m_k_ranhTQ_1.m here:

if h_px > (h1_r + h2_r)
% ...
end

if h_px<=(h1_r + h2_r)
% ...
end

When you send scalar X,Y in, then those lines are ok (if redundant), but when you send matrix X,Y, then h_px and h1_r are both matrices. Using a matrix conditional in an if statement evaluates to true only if the condition is true for all elements of the matrix. Therefore you get different results when you send in the whole matrix vs when you send the elements one at a time.

The solution is to adapt your function m_k_ranhTQ_1() to handle matrix inputs correctly (if you want to use it like that).

[quan ng]

yes, i will try that. For now i think i will use the loops to calculate the m_k matrix. And you answer my surf question too! You have been a great help!

[Voss]

I'm glad to help! And thanks for accepting this answer too!

[quan ng]

No problem! Thank you too!