how to find indices that fulfil a certain criteria

Lets say I want to write a script and the input is the following vector:

vector = [1 1 1 1 3 3 3 9 9 9 9 2 2 1 1 1 2 7 9 7 2 1 1 1]

I want the result to be a new vector that contain the indices [12 21] (the two 2's in bold and italic). The script should chose these indices because they are the first to follow index 7 and 17 (in bold) and at the same time have a value of that particular index +/- one... I hope it make sense and that it is doable. Thanks Michael

Answer 1

[Iain]

 after_7 = 1:numel(vector) > 7;
 after_17= 1:numel(vector) > 17;
 twos_after_7 = vector == 2 & after_7;
 twos_after_17 = vector == 2 & after_17;
 output = find(twos_after_7,1,'first');
 output(2) = find(twos_after_17,1,'first');

[Michael Clausen]

Thank you Lain, I should probably specify that the vector I included in my question is an example. In reality I have very large data sets and is interested in a script that use the data and the vector with indexes I already have (here exemplified by 7 and 17) to generate a new vector with the indexes exemplified as 12 and 21.

In other words it should be more general (nor necessarily 7 and 17) and it should select indexes for data points that fulfil the +/- one criteria. If the vector instead was:

vector = [1 1 1 1 3 3 3 9 9 9 9 3.5 2 1 1 1 2 7 9 7 3.5 1 1 1]

it should come up with 12 and 22.. Do you think this is possible? Thank you very much

[Iain]

Ok... rereading it...

 indexes  = [7 17 64 504 3020]; %or whatever....

 for i = 1:numel(indexes)
  idx = indexes(i);
  first_value = vector(idx);
  after_value = 1:numel(vector) > idx;
  hits = after_value & first_value > (vector-1) & (first_value < vector + 1);
  result(i) = find(hits,1);
 end

Answer 2

[Guillaume]

vector = [1 1 1 1 3 3 3 9 9 9 9 2 2 1 1 1 2 7 9 7 2 1 1 1];
refindices = [7 17];
outindices = zeros(size(refindices));
for refi = 1:numel(refindices)
  refindex = refindices(refi);
  outindex = find(abs(vector(refindex+1:end) - vector(refindex)) <= 1, 1);
  if ~isempty(outindex)
    outindices(refi) = refindex + outindex;
  else
    %whatever you want to do or nothing if you want 0 when criteria is not found
  end
end

would be one way to do it.

Another option, sort of avoiding the loop (the loop is implemented by cellfun):

vector = [1 1 1 1 3 3 3 9 9 9 9 2 2 1 1 1 2 7 9 7 2 1 1 1];
refindices = [7 17];
splitvector = mat2cell(vector, 1, diff([1 refindices numel(vector)+1]));
outindices = refindices + cellfun(@(subv) find(abs(subv(2:end) - subv(1)) <= 1, 1), splitvector(2:end));

[Michael Clausen]

Dear Guillaume, thank you for your answers, they work beautifully.