how to store a double loop result into a matrix column by column?

I have a matrix cycle; What I am doing is to find out certain points satisfying the condition and store them into Array up and down;

When cycle1 is one column, it could work; but when cycle1 increases to 2 columns, then the double loop doesn't work.

up and down's size is not certain; I just estimate their size 100*12;

now the code has no error, but give wrong results.

please give me some suggestion about this code.

cycle1 = [butterworth_filter1,butterworth_filter2];
[m,n] = size(cycle1)
up = zeros(100,12);
down = zeros(100,12);
k=1;
for j = 1:n
    for i = 1:m
        if i == m
            break
        elseif cycle1(i,j) < 1.5e-5 && cycle1(i+1,j) > 1.5e-5
            up(k,:) = i + 1;
            k = k+1;
        elseif cycle1(i,j) > 1.5e-5 && cycle1(i+1,j) < 1.5e-5
            down(k,:) = i;
            k = k+1;
        end
    end
end

[Geoff Hayes]

Lei - you say that the code has no error, but gives wrong results. What is wrong about the results? What are you expecting? Should you be using k for both the up and down arrays? Why are these arrays 100x12 - do you really need 12 columns or would one suffice?

[LEI Li]

I use the code to test it; no error; but it writes the result to the array; but I didn't know how to eliminate the redundent zero. I need 12 columns bc I have 12 data array;

I try to find certain value in cycle1 and record their locations; then I will use these locations to find the data set between up and down.

close all;
clear;
clc;
cycle1 = randn(100,12);
[m,n] = size(cycle1)
up = zeros(100,12);
down = zeros(100,12);
k=1;
for j = 1:n
    for i = 1:m
        if i == m
            break
        elseif cycle1(i,j) < 0.2 && cycle1(i+1,j) > 0.2
            up(k,:) = i + 1;
            k = k+1;
        elseif cycle1(i,j) > 0.2 && cycle1(i+1,j) < 0.2
            down(k,:) = i;
            k = k+1;
        end
    end
end

[Geoff Hayes]

Lei - where are the 12 arrays in your code? Or do you have variables called cycle2, cycle3, etc.? From your comment I try to find certain value in cycle1 and record their locations;... since the location is a pair (i,j) do you need to store these two values? Why k?

[LEI Li]

this is the original code.

the reason I use k is to push the selected location into cycle1. the index (i,j) will be too big for selected location.

close all;

clear all;

clc;

emg = fopen('C:\MGasR.csv');

%% Read Data Columns;

res = textscan(emg, '%f %f %f %f %f %f %f %f %f %f %f %f','headerlines',2);

fclose(emg);

%% Data Matrix

T = cell2table(res);

A = cell2mat(res);

%% Raw Signal

figure(1)

plot(A(:,1),A(:,8), A(:,1), A(:,9), A(:,1),A(:,10));

grid on

xlabel('Time(s)');ylabel('EMG (V)');title('Raw EMG Signal')

legend ('Subject 2')

%% Rectified Signal

rec_emg=abs(A);

figure(2)

plot(rec_emg(:,1),rec_emg(:,8), rec_emg(:,1),rec_emg(:,7));

grid on

xlabel('Time(s)');ylabel('EMG (V)');title('Rectified EMG Signal [Full-Wave Rectification]')

legend ('Subject 2')

%% Filtered Signal

[a,b]= butter(4, 0.002);

butterworth_filter1 = filter(a, b, rec_emg(:,8));

butterworth_filter2 = filter(a, b, rec_emg(:,7));

butterworth_filter3 = filter(a, b, rec_emg(:,4));

butterworth_filter4 = filter(a, b, rec_emg(:,5));

butterworth_filter5 = filter(a, b, rec_emg(:,6));

figure (3)

%Plotting style(res{1,1}, res{1,2}, res{1,1}, res{1,3})

plot(rec_emg(:,1),butterworth_filter1, rec_emg(:,1),butterworth_filter2);

xlabel('Time(s)');ylabel('EMG(V)');title('Filtered EMG Signal [Butterworth]')

legend ('Subject 2')

cycle1 = butterworth_filter1;

up = [];

down = [];

for i = 1:length(cycle1)

if i == length(cycle1)

break

elseif cycle1(i) < 1.5e-5 && cycle1(i+1) > 1.5e-5

up(end+1) = i + 1;

elseif cycle1(i) > 1.5e-5 && cycle1(i+1) < 1.5e-5

down(end+1) = i;

end

end

figure (4)

plot(cycle1)

hold on

plot(up, cycle1(up), 'go', 'MarkerSize', 8, 'MarkerFaceColor', 'g')

plot(down, cycle1(down), 'ro', 'MarkerSize', 8, 'MarkerFaceColor', 'r')

xlabel('Time(s)');ylabel('EMG(V)');title('Muscle Activation & Cycles')

legend ('Subject 1')

[Geoff Hayes]

So if (i,j) is "too big" then is k good enough? If not, where is the error?

[Stephen23]

LEI Li's incorrectly posted "Answer" moved here:

from the test code, up has the result; now each column is the same; I need to chech the random matrix;

next thing is there should not exist 0 in the column; this is the reason I use k;

previously when I deal with one dimension array, I use

up(end+1) = i + 1; to store the location; but now I need to deal with a matrix; I stuck here.

Answer 1

[Jan]

You are using one counter k for the up and the down lists.

cycle1 = [butterworth_filter1,butterworth_filter2];
[m, n] = size(cycle1)
up     = zeros(100, 12);
down   = zeros(100, 12);
kup    = 1;
kdown  = 1;
for j = 1:n
    for i = 1:m - 1  % Simpler than: if i==m, break
        if cycle1(i,j) < 1.5e-5 && cycle1(i+1,j) > 1.5e-5
            up(kup,:) = i + 1;  % Are you sure you want to set the complete row to i+1?
            kup = kup + 1;
        elseif cycle1(i,j) > 1.5e-5 && cycle1(i+1,j) < 1.5e-5
            down(kdown, :) = i;
            kdown = kdown + 1;
        end
    end
end

[LEI Li]

close all;

clear;

clc;

cycle1 = abs(randn(100,12));

for i = 1:12

plot(cycle1(:,i))

hold on

end

[m,n] = size(cycle1)

up = zeros(100,12);

down = zeros(100,12);

kup=1;

kdown=1;

for j = 1:n

for i = 1:m-1

% if i == m-1

% break

if cycle1(i,j) < 0.2 && cycle1(i+1,j) > 0.2

up(kup,:) = i + 1; %up only has one value now??

kup = kup+1;

elseif cycle1(i,j) > 0.2 && cycle1(i+1,j) < 0.2

down(kdown,:) = i;

kdown = kdown+1; %down only has one value now??

end

end

end

figure (4)

plot(cycle1)

hold on

plot(kup, cycle1(kup), 'go', 'MarkerSize', 8, 'MarkerFaceColor', 'g')

plot(kdown, cycle1(kdown), 'ro', 'MarkerSize', 8, 'MarkerFaceColor', 'r')

xlabel('Time(s)');ylabel('EMG(V)');title('Muscle Activation & Cycles')

legend ('Subject 1')