Split a vector into multiple vectors based on a value in the original vector

Please, I would like to split a vector into multiple vectors. For example a vector [2 0 5 7 0 14 0 17 19 20] should be split using the 0 in it as a delimiter to obtain

a1 = [2].... a2 = [5 7]...... a3 = [14]....... a4 = [17 19 20].

I would be very grateful. Thanks

[Stephen23]

Note that it is a really bad code to create lots of separate variables dynamically. Although beginners love doing this, in fact it is a slow, buggy, and obfuscated way to write code. Read this to know why:

http://www.mathworks.com/matlabcentral/answers/196952-how-to-create-new-variables-in-batches-with-strcat

Instead you should simply put all of the output vectors into one cell array.

Answer 1

[Star Strider]

One approach:

V = [2 0 5 7 0 14 0 17 19 20];
idx = [find(V == 0) length(V)+1];                   % Find Zeros & End Indices
didx = diff([0 idx])-1;                             % Lengths Of Nonzero Segments
V0 = V;                                             % Create ‘V0’ (Duplicate Of ‘V’)
V0(V0 == 0) = [];                                   % Delete Zeros
C = mat2cell(V0, 1, didx);                          % Create Cell Array
Result = C(cellfun(@(x) size(x,2)~=0, C));          % Delete Empty Cells

It’s a bit more complicated here than it needs to be for your vector, because I also designed it to be robust to leading and trailing zeros as well, even though your vector doesn’t have them.

I did not assign them as ‘a1’ ... ‘a4’ because creating dynamic variables is poor programming practise. They are instead elements of the ‘Result’ cell array.

If you have a specific reason to refer to an element of ‘Result’ as a particular variable, ‘a4’ would be:

a4 = Result{4}

a4 =
    17    19    20

Answer 2

[Jos (10584)]

V = [0 2 0 0 5 7 0 14 0 0 0 17 19 20]
tf = V==0
ix = cumsum(tf & [true diff(tf) > 0])
Result = arrayfun(@(k) V(ix==k & ~tf),1:ix(end),'un',0)

[Amir Pasha Zamani]

Great job Genius :)

THANKS

Answer 3

[Andrei Bobrov]

V = [0 2 0 0 5 7 0 14 0 0 0 17 19 20];

V0 = V(:)';
i0 = diff(strfind([0,V0,0],0));
out = mat2cell(V0(V0 ~= 0),1,i0(i0 > 1) - 1);

or with bwlabel from Image Processing Toolbox

x = bwlabel(V(:));
t = x ~= 0;
out = accumarray(x(t),V(t),[],@(x){x'});