How to have a sum of 2d fields in a 3d matrix so the time dimension remains?

Hello,

I have the following problem:

I have a 3D matrix with dimensions seen below and I am trying to get mean values of fields on first two dimensions,

while keeping the last dimension resolution, so the final matrix would be a 31 length vector. I have tried as seen below,

but that failed for the moment.

I would greatly appreciate a suggestion on how to do that correctly

% ice_sum_1st_aug is a matrix 480x30x31 - trying to get sum of values on 480x30 bits against time
% the needed result would be a vector of 31 points in lenght
for k=1:31
    icesum_1st_aug(k)=sum(ice_1st_august(:,:,k));
end

Answer 1

[Walter Roberson]

icesum_1st_aug = sum(sum(ice_1st_august,1),2);

If you have R2018b or later then

icesum_1st_aug = sum(ice_1st_august, [1 2]);

If you really want to loop,

nk = size(ice_1st_august,3);
icesum_1st_aug = zeros(1,1,nk);
for k = 1 : nk
    icesum_1st_aug(1,1,k) = sum(reshape(ice_1st_august(:,:,k),[],1));
end

[Mikolaj Jankowski]

Thank you, it worked, I found out how to apply it and get the result I need :)

Answer 2

[Steven Lord]

If you're using release R2018b or later, both sum and mean (your code used sum but your description of your goal indicated you're looking for the mean) accept a vector as the dimension input argument to operate over multiple dimensions simultaneously.

A = randi(10, [4, 5, 3]);
meanOfEachPage = mean(A, [1 2]);

You can check this, for example for the first page of A:

firstPage = A(:, :, 1);
meanOfFirstPage = mean(firstPage(:));
meanOfFirstPage - meanOfEachPage(1)

See the Release Notes for a list of the functions that support this functionality.