4D data /Index exceed array bounds

Hey all,

I'm wrting a code which needs to store 2D position and velocity data in 4D cell arrays. I have to for loops first one is for each trial (k=1:5 which is the 4th dim), second one is for each words (l=1:134) which is the 3rd dim. I need to get the mean of data for each word over 5 trials and write a .dat file. In the tecplot section I got :

M_Vel(1,2,nsyll{k},k) Index in position 2 exceeds array bounds (must not exceed 1).

but my M_Vel is :

val(:,:,1,1) = {21600×4×135×2 double}

I couldn't fix it. If you're able to hepl me I'll be appreciate. Thanks in advance

[DGM]

You might want to include some example data that can allow the problem to be reproduced.

Answer 1

[Are Mjaavatten]

I think you may have inadvertently created a cell array containing your whole data matrix in element 1. Something like this:

>> M_vel = {rand([2,3,4,5])};
>> M_vel(:,:,2,1)
Index in position 3 exceeds array bounds (must not exceed 1).

To check this, type:

size(M_vel)

My guess is that you want a 4D array of doubles instead of a cell array. You may fix this by

M_vel = M_vel{1};

but a better idea is to look at your code and make sure that you make an array of doubles to start with. In my example:

>> M_vel = rand([2,3,4,5]);
>> M_vel(:,:,2,1)
ans =
    0.4294    0.2976    0.1192
    0.2573    0.4249    0.4951

[mehtap agirsoy]

Thanks for the help. You're right my code store the whole data in a single array which I don't want. I've made some changes.Please correct me if I my idea is wrong, size of 3rd dimension changes with the 4rd dim so I need cell array.

M_Vel=cell(1,1,nsyll{k},k) creates what I want val(:,:,1) ={21600×4 double}

but when it starts to 4th dim doesn't store the data val(:,:,1,1) = {0×0 double}

don't know how to fix

Answer 2

[Are Mjaavatten]

I think it may be better to let M_Val be a cell array of 5 ordinary arrays, where M_Val{k} is of dimension 134 by 2 by nsyll(k). This creates a demo version with random numbers:

n_trials = 5;
nsyll = [6,4,5,8,7,9];
n_words = 134;
M_Vel = cell(1,n_trials);
for k = 1:n_trials
    M_Vel{k} = randn(n_words,2,nsyll(k));
end

You say want the mean of data for each word, averaged over the 5 trials. The data you average must have fixed dimensions, independent of k. One example might be to calculate the the min,max and variance for position and velocity for each given trial and word. To prepare for the averaging over trials, I suggest to collect these vaues in a 5 by 134 by 2 by 3 array of doubles:

results = zeros(n_trials,n_words,2,3);
for k = 1:n_trials
  for j = 1:n_words
    results(k,j,:,:) = [min(M_Vel{k}(j,:,:),[],3)',max(M_Vel{k}(j,:,:),[],3)'...
      ,var(M_Vel{k}(j,:,:),[],3)'];
  end
end

Average over trials by taking the mean (over dimension 1) and remove the trials dimension by the squeeze command:

averages  = squeeze(mean(results));
size(averages)
ans =
   134     2     3

It is easy to get a little lost when handling multidimensional arrays, but I hope this is close enough to what you want, or that it may serve a starting point.

If you need more help, I shall need (a subset of) your data files.

Answer 3

[mehtap agirsoy]

Many thanks for the help. I've tried your suggestion but couldn't. I really got lost. I'm not good at matlab so have a real uphill struggle ahead of me now. I've attached some data if you're able to help, appreciate it. Many thanks again.

[Are Mjaavatten]

I had a look at your Archive.zip file and I am unable to figure out what you are trying to do. My main impression is that you are trying to do way too much at once. For large, complex projects like this, the strategy is to divide and conquer! Try to split your project into (much) smaller parts and test that each part works and does what you intend. Then combine these building blocks a few at a time, always testing that you obtain the expected results.