What is the best way to set up a nested for loop?

Hi,

I need help creating an efficient, nested for-loop. I have to run through a couple of equations on a few of variables but need to know the best way to increment these variables and to store their new values.

Thanks,

Steve

[Stephen23]

The best way:

1. Preallocate the required output arrays.
2. Use indexing in the nested loops.

[Steve]

How so?

[Steve]

How would this be applied to my specific files? (see attached).

[darova]

Your code looks ok. What do you need help?

[Steve]

i need to create another loop to find and store the 3 angles that each Triplet makes with their corresponding F_point. Then, I would like to store all of this in one file.

Answer 1

[Sulaymon Eshkabilov]

Hi Steven,

You loop is not doing much. SInce you are saving the last value from your nested loops. If this is what you want then, just remove the loops as follows:

load('Triplets');
load('F_points');
i=953; j=953;
xC = Triplets{i}(2,1);
xB = Triplets{i}(2,2);
xA = Triplets{i}(2,3);
yC = Triplets{i}(1,1);
yB = Triplets{i}(1,2);
yA = Triplets{i}(1,3);

yb = F_points{j}(2,1);
xb = F_points{j}(1,1);
xyb = [yb,xb];
xye = [Triplets{j}(1,3), Triplets{j}(2,3); Triplets{j}(1,2) Triplets{j}(2,2); Triplets{j}(1,1) Triplets{j}(2,1)];
    
CosTheta1 = dot(xye(1,:)-xyb(1,:),xye(2,:)-xyb(1,:))/(norm(xye(1,:)-xyb(1,:))*norm(xye(2,:)-xyb(1,:)));
ThetaInDegrees1 = acosd(CosTheta1);
CosTheta2 = dot(xye(2,:)-xyb(1,:),xye(3,:)-xyb(1,:))/(norm(xye(2,:)-xyb(1,:))*norm(xye(3,:)-xyb(1,:)));
ThetaInDegrees2 = acosd(CosTheta2);
CosTheta3 = dot(xye(3,:)-xyb(1,:),xye(1,:)-xyb(1,:))/(norm(xye(3,:)-xyb(1,:))*norm(xye(1,:)-xyb(1,:)));
ThetaInDegrees3 = acosd(CosTheta3);
degsum=ThetaInDegrees1+ThetaInDegrees2+ThetaInDegrees3;

If you do wnat to save all values of variables (xC, xB, ... xye, ... degsum) from your computations, then you'd need to specify the indexes and then you'd need to do the memory allocation:

xA = zeros(1,953);  % Memory allocation
xB = zeros(1,953);
xC = zeros(1,953);
yA = zeros(1,953);
yB = zeros(1,953);
yC = zeros(1,953);
...
    
for i = 1 : length (Triplets)
   
    for j = 1 : i
           
xC(i) = Triplets{i}(2,1);
xB(i) = Triplets{i}(2,2);
xA(i) = Triplets{i}(2,3);
yC(i) = Triplets{i}(1,1);
yB(i) = Triplets{i}(1,2);
yA(i) = Triplets{i}(1,3);

yb(j) = F_points{j}(2,1);
xb(j) = F_points{j}(1,1);
...

    end
end

[Stephen23]

"memory allocation" -> array preallocation:

https://www.mathworks.com/help/matlab/matlab_prog/preallocating-arrays.html

[Steve]

Thanks for the info. What about the angle finding portion of the code? See below:

CosTheta1 = dot(xye(1,:)-xyb(1,:),xye(2,:)-xyb(1,:))/(norm(xye(1,:)-xyb(1,:))*norm(xye(2,:)-xyb(1,:)));

ThetaInDegrees1 = acosd(CosTheta1);

CosTheta2 = dot(xye(2,:)-xyb(1,:),xye(3,:)-xyb(1,:))/(norm(xye(2,:)-xyb(1,:))*norm(xye(3,:)-xyb(1,:)));

ThetaInDegrees2 = acosd(CosTheta2);

CosTheta3 = dot(xye(3,:)-xyb(1,:),xye(1,:)-xyb(1,:))/(norm(xye(3,:)-xyb(1,:))*norm(xye(1,:)-xyb(1,:)));

ThetaInDegrees3 = acosd(CosTheta3);

degsum=ThetaInDegrees1+ThetaInDegrees2+ThetaInDegrees3;

Answer 2

[Sulaymon Eshkabilov]

Hi Steven,

Here is the complete solution of your problem:

clearvars
%% Part 0. Loading data
load('Triplets');
load('F_points');
N = length(Triplets);

%% Part 1. Saving all values of arrays from the calculations

% Memory allocation
xA = zeros(1,N);  
xB = zeros(1,N);
xC = zeros(1,N);
yA = zeros(1,N);
yB = zeros(1,N);
yC = zeros(1,N);
xb = zeros(1,N);
yb = zeros(1,N);
for ii = 1 : N
    for jj = 1 : N
        xC(ii) = Triplets{ii}(2,1);
        xB(ii) = Triplets{ii}(2,2);
        xA(ii) = Triplets{ii}(2,3);
        yC(ii) = Triplets{ii}(1,1);
        yB(ii) = Triplets{ii}(1,2);
        yA(ii) = Triplets{ii}(1,3);
        yb(jj) = F_points{jj}(2,1);
        xb(jj) = F_points{jj}(1,1);
    end
end
xyb = [yb; xb]';  % Augmented matrix array = xyb

% Memory allocation:
xye1 = zeros(2,N); 
xye2 = zeros(2,N);
xye3 = zeros(2,N);
for k = 1:N
xye1(:,k) = [Triplets{k}(1,3), Triplets{k}(2,3)]; 
xye2(:,k) = [Triplets{k}(1,2) Triplets{k}(2,2)];   
xye3(:,k) = [Triplets{k}(1,1) Triplets{k}(2,1)];  
end
XYE0 = [xye1', xye2', xye3'];  % Augmented matrix array = 'xye'

%% Part 2. Compute Angles

% Memory allocation: 
 CosTheta1 = zeros(1,N);    
 ThetaInDegrees1=zeros(1,N);
 CosTheta2 = zeros(1,N);
 ThetaInDegrees2 = zeros(1,N);
 CosTheta3 =  zeros(1,N);
 ThetaInDegrees3 = zeros(1,N);
for jj = 1:N
CosTheta1(jj) = dot(XYE0(jj,1:2)-xyb(jj, :),XYE0(jj,3:4)-xyb(jj,:))/(norm(XYE0(jj,1:2)-xyb(jj,:))*norm(XYE0(jj,3:4)-xyb(jj,:)));
ThetaInDegrees1(jj) = acosd(CosTheta1(jj));
CosTheta2(jj) = dot(XYE0(jj,3:4)-xyb(jj, :),XYE0(jj,5:6)-xyb(jj,:))/(norm(XYE0(jj,3:4)-xyb(jj,:))*norm(XYE0(jj,5:6)-xyb(jj,:)));
ThetaInDegrees2(jj) = acosd(CosTheta2(jj));
CosTheta3(jj) = dot(XYE0(jj,5:6)-xyb(jj, :),XYE0(jj,1:2)-xyb(jj,:))/(norm(XYE0(jj,5:6)-xyb(jj,:))*norm(XYE0(jj,1:2)-xyb(jj,:)));
ThetaInDegrees3(jj) = acosd(CosTheta3(jj));
degsum(jj)=ThetaInDegrees1(jj)+ThetaInDegrees2(jj)+ThetaInDegrees3(jj);
end

good luck

[Steve]

Thanks so much for the info! It is truly appreciated. I'm getting an error when I run it though:

Index exceeds the number of array elements (3).

It seems it doesn't like the size of F_points vs Triplets. Can I just store them both by pairs of their x, y coordinates?

Thanks again for all your help!

[Steve]

I just wanted to add that I loaded different F_point and Triplets files and the code ran, but the angles are not correct. They are all coming up as decimals (instead of values like 120 for example). Any idea why this might be the case?

Thanks!

[Sulaymon Eshkabilov]

Ok. Glad that can of some help. Hit accept of this answer, pl.

Make sure that your formualtions are giving the values in degrees not radians. If other way around, then you should use conversion the values rad2deg(). With rad2deg() you will get the values in degrees. Always watch out the size match.

[Steve]

I don't think the angles are in radians or degrees. My original code portion for finding the angles gave true angles in degrees. Attached are the new input files; they may not be in the correct row/column arrangements, and that may be causing the problem with the angles. How does it look to you?

Thanks!