merging two cell type matrix

start_id=[{'p1'};{'p2'};{'p1'};{'p1'};{'p3'}];  %cell
finish_id=[{'p2'};{'p3'};{'p3'};{'p4'};{'p4'}];  %cell
points_id=[{'p1'};{'p2'};{'p3'};{'p4'}];  %cell
m=numel(points_id);

for i=1:m
    IndexC_minus(:,i) = strfind(start_id, points_id{i});
    IndexC_positive(:,i) = strfind(finish_id, points_id{i}); 
end

I need to create IndexC_minus_double and IndexC_positive_double matrix as follows;

IndexC_minus_double=[-1 0 0 0;0 -1 0 0;-1 0 0 0;-1 0 0 0;0 0 -1 0]; %for each [1] cell, write -1 numeric and for each [] cell write 0 numeric.  

IndexC_positive_double=[0 1 0 0;0 0 1 0;0 0 1 0;0 0 0 1;0 0 0 1]; %for each [1] cell, write 1 numeric and for each [] cell write 0 numeric.

Finally, I need to merge these 2 matrix as below;

merged_matrix=[-1 1 0 0;0 -1 1 0;-1 0 1 0;-1 0 0 1;0 0 -1 1]; % overlapped 0 produces 0 in the merged_matrix. The location of 1 and -1 in the IndexC_minus_double and IndexC_positive_double matrix are the same in the merged_matrix.

[Guillaume]

Did you really mean to use strfind (which will find p1 in abcp1d) and not just compare the strings for equality with strcmp?

You have an odd way of constructing cell arrays. This is simpler:

start_id = {'p1'; 'p2'; 'p1'; 'p1'; 'p3'};

Answer 1

[Andrei Bobrov]

merged_matrix = bsxfun(@strcmp,finish_id,points_id')...
                      - bsxfun(@strcmp,start_id,points_id');

ADD

[~,~,c] = unique([start_id;finish_id;points_id]);
ii = zeros(numel(c),1);
n = cellfun(@numel,{start_id;finish_id;points_id});
ii(cumsum(n) - n + 1) = 1;
ii = cumsum(ii);
c3 = c(ii == 3)';
merged_matrix = bsxfun(@eq,c(ii==2),c3) - bsxfun(@eq,c(ii==1),c3)

in your case

v = {start_id;finish_id;points_id};
for ii = 1:3
    a = regexp(v{ii},'\d','match');
    v{ii} = str2double(cat(1,a{:}));
end
merged_matrix = bsxfun(@eq,v{2},v{3}') - bsxfun(@eq,v{1},v{3}');

other variant

n = numel(points_id);
s = repmat(start_id,1,n);
f = repmat(finish_id,1,n);
nn = repmat(points_id',numel(finish_id),1);
merged_matrix = strcmp(f,nn) - strcmp(s,nn);

[sermet]

Andrei, it gives this error;

Error using bsxfun
Operands must be numeric arrays.

IndexC_minus_double and IndexC_positive_double matrix should be created before bsxfun, I suppose.

[Andrei Bobrov]

Hm! Octave is has of solution. See ADD part

[sermet]

could you modify it for matlab?

[sermet]

Andrei, it gives,

Error using  - 
Matrix dimensions must agree
merged_matrix = bsxfun(@eq,c(ii==2),c3) - bsxfun(@eq,c(ii==1),c3)

from codes,

bsxfun(@eq,c(ii==2),c3)= []
bsxfun(@eq,c(ii==1),c3)=Empty matrix: 3-by-0

[sermet]

thank you very much, Andrei.

Answer 2

[Guillaume]

This is probably much simpler than Andrei's answer. I assume you meant to use strcmp in your original example:

IndexC_minus_double = cell2mat(cellfun(@(s) -strcmp(s, points_id)', start_id, 'UniformOutput', 0))
IndexC_positive = cell2mat(cellfun(@(s) +strcmp(s, points_id)', finish_id, 'UniformOutput', 0))
merged_matrix = IndexC_minus_double + IndexC_positive

Note that I assume that -1 and +1 don't overlap for the merge matrix since you haven't said what happens if this is the case

[Andrei Bobrov]

+1