Vectorizing a very slow code

Hi,

I have a code that works, but is extremely slow that I would like to vectorize. I don't know anything about vecotrization. I am not really looking for anyone to re-write the code, that would be too much to ask, but I'd be grateful if someone could point me in the right direction as to which functions or basic building blocks I should look at.

Basically, what I am trying to do is to create a network of social contacts. I have a number of individuals and a total number of contacts each one of them has. Then I iterate over individuals and create the network of contact. Starting with person 1, I randomly pick the names of the correct number of individuals (say, individuals 5 and 10), add them to the list of contacts of individual 1, and then add individual 1 to the contact lists of individuals 5 and 10. Then I update the number of contacts left for individuals 1, 5, and 10.

The code that I have works fine for a small number of people, but it's very slow when the number of people increases significantly. I'd be grateful for any ideas as to how I can vectorize it. Here is what I have:

% Creating work contact network

bb1 = find(Employed == 1);

bb2 = length(bb1); % finding unique school numbers

for aa = 1:bb2 % looping over school numbers

abb2 = find(Employed == 1 & WCleft(:,1) > 0); % finding all the people in a given school

if ~isempty(length(abb2)) % looping over people in school

bb5 = find(abb2 ~= bb1(aa,1)); % find people who are different from the index person

abb5 = randperm(length(bb5)); % randomize the selection so that workplace contacts are not all in order

if length(bb5) >= WCleft(bb1(aa,1),1)

abb6 = abb5(1,1:WCleft(bb1(aa,1),1)); % pick a subset of these people equal to the number of workplace contacts the index person has

if ~isempty(abb6)

b7 = find(Wcontacts(bb1(aa,1),:) == 0); % find the first empty spot in the array of contacts

Wcontacts(bb1(aa,1),b7(1,1):b7(1,1) + WCleft(bb1(aa,1),1) - 1) = abb2(abb6,1)'; % add contacts to the index person

WCleft(bb1(aa,1),1) = WCleft(bb1(aa,1),1) - length(abb2(abb6,1)); % subtract the number of contacts from the total number of contacts the index person has

% Need to assign reciprocal contacts

for aa2 = 1:length(abb2(abb6,1)) % looping over people in that class

bb6 = find(Wcontacts(abb2(abb6(1,aa2),1),:) == 0); % find the first available spot in the contact matrix

if ~isempty(bb6)

Wcontacts(abb2(abb6(1,aa2),1),bb6(1,1)) = bb1(aa,1); % Putting the index individual as a contact for all of his contacts

WCleft(abb2(abb6(1,aa2),1),1) = WCleft(abb2(abb6(1,aa2),1),1) - 1; % Subtacting one contact from the number of available contacts

end

end

end

else

abb6 = abb5;

if ~isempty(abb6)

b7 = find(Wcontacts(bb1(aa,1),:) == 0); % find the first empty spot in the array of contacts

Wcontacts(bb1(aa,1),b7(1,1):b7(1,1) + length(abb6) - 1) = abb2(abb6,1)'; % add contacts to the index person

WCleft(bb1(aa,1),1) = WCleft(bb1(aa,1),1) - length(abb2(abb6,1)); % subtract the number of contacts from the total number of contacts the index person has

% Need to assign reciprocal contacts

for aa2 = 1:length(abb2(abb6,1)) % looping over people in that class

bb6 = find(Wcontacts(abb2(abb6(1,aa2),1),:) == 0); % find the first available spot in the contact matrix

if ~isempty(bb6)

Wcontacts(abb2(abb6(1,aa2),1),bb6(1,1)) = bb1(aa,1); % Putting the index individual as a contact for all of his contacts

WCleft(abb2(abb6(1,aa2),1),1) = WCleft(abb2(abb6(1,aa2),1),1) - 1; % Subtacting one contact from the number of available contacts

end

end

end

end

end

end

Answer 1

[Rohit Pappu]

Documentation about Vectorization in MATLAB can be found here