How can I reset my variables after each iteration and then save them ?

Hello everyone,

I am having trouble with reseting my variables for each iteration. So the main idea is this:

I have 6 measurements (Data series called 'OUTPUT',each one has 6 Columns: 1. time, 2. and 3. sound pressure, 4.,5. and 6. wind velocities xyz).

For each measurement I have to evaluate the mean, std and so on, so thats why I am dividing my signal into blocks of 5s.

For example, 1. Measurement lasts 40s, each block will be 5s. (Sample frequency 100Hz).

But each measurements doesnt have the same amount of blocks , because they dont last all 40s.

So outside the main evaluations I added a loop : for j=1:6, to choose everytime which measurement I want to read in.

And after choosing it, at the block parameter I added:

for j=1:6
    %reading in the Input File Array

    %Block parameters
    blocks=[8 6 4 6 6 6]; %Block number for each measurement
    i_blocks=blocks(j); %Block number for each measurement
    n_block_res=floor(length(OUTPUT_resample)/i_blocks); %Number of samples for each block
    t_block=n_block_res/fs_resample; %Duration of each block [s]
            %Measurements
            for ii=1:i_blocks
            Mean(ii,:,j)=....    %here I added the j at the end so I will get  a 3D matrix for each measurement j iteration
            Std(ii,:,j)=....
            ........
            end
end

So the problem is that, after the first iteration, the variables wont be overwriten and therefore I get an error because the size on the left side doesnt agree with the one on the right (as I said each measuement has different number of blocks).

So maybe the Mean has a size at measuement 1 , 8(blocks)x6, but the Mean at measurement 2, wont be 6(blocks)x6, but still 8x6.

I hope my problem is making sense. Thanks for your time and help!

[KALYAN ACHARJYA]

You may use clear varibale name after the iteration, but most experts generally not recomend to clear varaibale name within the code??

Wait, experts will give you the best possible way out.

[Ifigenia Aslanidou]

Hey,

yeah I tried writing before each variable 'clear .....' , but if you have like 50 of them its impossible.

Thanks for your help!

[Stephen23]

You do not need to clear the variables, just preallocate them before that loop:

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

[Ifigenia Aslanidou]

Wow!It really worked!

Thank you very much :) Saved my day!

[Adam Danz]

"Wow! t really worked!"

And it's much faster and cleaner.

[Ifigenia Aslanidou]

:)

But now with Preallocating, I think, I cannot have all 6 values of measurements written in the Workspace Results.

Ex.1: Mean(ii,:,j) %j is for each measurement (3D Matrix)

1.1 Workspace Results: Mean(1,:,1)=[....] until Mean(8,:,1)=[....]

Mean(1,:,2)=[.....] until Mean(6,:,2)=[....]

................................................................

Mean(1,:,6)=[.....] until Mean(6,:,6)=[....]

-How would you extract/save all values for all measuerements?

[Ifigenia Aslanidou]

This could work:

save(['Turb_Intensität_Länge' num2str(j) '.mat']);

PROBLEM:

In this case, if I want to combine the results of all measurements (j loop) in one Plot, then it won't work, because they are in different workspaces.

Is there a way to combine plots from different workspaces and then edit them?

There is only 1 script.

[Adam Danz]

Could you share your updated code that includes pre-allocation?

[Ifigenia Aslanidou]

My script looks like that:

(sorry I believe I am not allowd to upload the whole one,because I didnt write it on my own)

for j=1:6
 %reading in the Input File Array
    %Block parameters
    blocks=[8 6 4 6 6 6]; %Block number for each measurement
    i_blocks=blocks(j); %Block number for each measurement
    n_block_res=floor(length(OUTPUT_resample)/i_blocks); %Number of samples for each block
    t_block=n_block_res/fs_resample; %Duration of each block [s]
 %% Calling 'Preallocating_variables_ii'
    Preallocating_variables_ii; 
        for ii = 1:i_blocks
         %% Measurements
            Mean(ii,:) = mean(OUTPUT_resample(((ii-1)*n_block_res+1):(ii*n_block_res),:));
            Std(ii,:) = std(detrend(OUTPUT_resample(((ii-1)*n_block_res+1):(ii*n_block_res),:),1));
            Detr((ii-1)*n_block_res+1:(ii*n_block_res),:) = detrend(OUTPUT_resample(((ii-  ...                     ... 1)*n_block_res+1):(ii*n_block_res),:));
            T_u(ii) = Std(ii,4)/Mean(ii,4)*100;
            T_v(ii) = Std(ii,5)/Mean(ii,4)*100;
            ......... % (some other variables as well)
            end

 %% Calling Tu und L Result Plots
    Tu_L_Plots_Script;
end

Preallocating Script looks like that:

   %% Preallocating variables
                columns=6;
                Mean = zeros(i_blocks,columns);
                Std= zeros(i_blocks,columns);
                Var = zeros(i_blocks,columns);
                Detr= zeros(length(OUTPUT_resample),columns);

................................and so on for other variables.

I am not sure how I can plot these variables in 1 Plot for each measurement.

In the attachment is my script for the plots that need to be made. In each figure should be 6 different results for each measurement.

I added some for loops again for testing but its not wroking so dont mind them.

[Adam Danz]

Let's distill this down to something more manageable. Let us know if this summarizes your (new) problem.

With the code below, every time j iterates to the next loop, you lose the values from var1 and var2. Is that correct?

for j = 1:6
    
    % [code....]
    
    % Pre-allocate
    var1 = zeros(4,5); 
    var2 = zeros(4,5); 
    for ii = 1:4
        var1(ii,:) = func1(); 
        var2(ii,:) = func2();         
    end
end

If that's the problem, you can easily add another dimension to the loop variables and move the pre-allocation outside of the j-loop.

% Pre-allocate
var1 = zeros(6,4,5); 
var2 = zeros(6,4,5); 

for j = 1:6
    
    % [code....]

    for ii = 1:4
        var1(j,ii,:) = func1(); 
        var2(j,ii,:) = func2();         
    end
end

[Ifigenia Aslanidou]

Thanks for your answer!

''With the code below, every time j iterates to the next loop, you lose the values from var1 and var2. Is that correct?''

Answer:Yes, I lose all the values at each iteration, so I wont be able to plot all of them in 1 plot.

Unfortunately I cant put the preallocating script outside the loop because its in the j loop where I read in my Input file (OUTPUT) and inside my preallocating script I have used the length of OUTPUT (see attachment:specific variable 'Detr'). And the same is with the i_blocks variable,it changes in the loop each time...

In the Preallocating script: Is there any other way that it recognizes everytime the length of OUTPUT ?Can I use another length?The thing is it has to be the same length before and after that...

[Adam Danz]

Doing the pre-allocation from within a script isn't a good choice. It makes the code much less readable. Anyway, here's a template that fits your description.

% Pre-allocate the variables that will save ALL of the data
var1 = zeros(6,4,5);
var2 = zeros(6,4,5);
for j = 1:6
    
    % Pre-allocate the ii-loop variables
    var1_temp = zeros(4,5);
    var2_temp = zeros(4,5);
    for ii = 1:4
        var1_temp(ii,:) = func1();
        var2_temp(ii,:) = func2();
    end
    % Store the values before going to next loop in J
    var1(j,:,:) = var1_temp; 
    var2(j,:,:) = var2_temp; 
end

[Ifigenia Aslanidou]

Thanks for your template!

But I am still puzzled,

the variables at the first preallocation(outside j, mentioning only the numbers 4 and 5) have to be the same size with the ones inside the j loop..

At each j, the size of var1_temp is different, so when it changes it wont take the var1 size into consideration.So,putting preallocation outside the loop which depends on the var inside the loop is bit of a problem... :/

Any thoughts there?

Sorry if I'm making you crazy with all those questions :P

[Adam Danz]

I see. One way around that is to store the var1_temp and var2_temp in a cell array.

% Pre-allocate the variables that will save ALL of the data
var1 = cell(1,6);
var2 = cell(1,6);
for j = 1:6
    
    % Pre-allocate the ii-loop variables
    var1_temp = zeros(4,5);
    var2_temp = zeros(4,5);
    for ii = 1:4
        var1_temp(ii,:) = func1();
        var2_temp(ii,:) = func2();
    end
    % Store the values before going to next loop in J
    var1{j} = var1_temp; 
    var2{j} = var2_temp; 
end

Another solution would be to pre-allocate the var1 variables to the largest possible var1_temp size and then pad the smaller var1_temp matrices with NaN or 0s.

[Ifigenia Aslanidou]

Thank you very much!It worked :)