A doubt regarding a simple For loop indexing for certain rows only.

Hi,

I have a matrix 'a' and I want to index only certain rows and delete them. Refer to code below.

Logic : when I reach the row starting point A or starting point B, I want to delete the following rows which contain the value [1 2 4 5]. I need to delete only those which come below either starting point A or B, also, the number of times [1 2 4 5] repeat is not fixed. If you observe, even after the 11th row [1 2 3 7] there are values [1 2 4 5] present but I want to retain them.

Any suggestions would be of great help, thank you. :)

a  =  1 2 3 4     % start point A
      1 2 4 5     % delete
      1 2 4 5     % delete
      1 2 4 5     % delete
      1 2 4 5     % delete
      1 2 3 5     % don't delete
      1 2 3 6     % start point B
      1 2 4 5     % delete 
      1 2 4 5     % delete
      1 2 3 5     % don't delete
      1 2 3 7     % don't delete anything from below here. 
      1 2 4 5
      1 2 4 5
      1 2 4 5 
      1 2 4 5 
      1 2 3 5

b = size(a,1);
for c = 1:b
    if a(c, [1:4]) == [1, 2, 3, 4] | [1, 2, 3, 6]
        for d = c+1:b
             if (a(d, [1:4]) == [1, 2, 4, 5])
                 a(d, 5) = 1;
             end
        end
    end          
end

% displaying the numbers except for the ones that have 1 in their last column, basically deleting the rows with 1 in their last column
z = a(:, end);
y = a(~(z == 2), :);
y(:, end) = []



%% The result I'm obtaining from the above code :
y = 1 2 3 4
    1 2 3 5
    1 2 3 6
    1 2 3 5
    1 2 3 7
    1 2 3 5


%% My expected output : 

y = 1 2 3 4
    1 2 3 5
    1 2 3 6
    1 2 3 5
    1 2 3 7
    1 2 4 5
    1 2 4 5
    1 2 4 5
    1 2 4 5
    1 2 3 5
    

Answer 1

[Guillaume]

Since nobody has explained why your code doesn't work:

There are several problems with your

if a(c, [1:4]) == [1, 2, 3, 4] | [1, 2, 3, 6]

First one is that your condition is a row vector and in all likelyhood you don't know how if behaves when the condition is a vector and not a scalar. That is the result of

a(c, [1:4]) == [1, 2, 3, 4] | [1, 2, 3, 6]

is a logical vector of true and false values that you pass to if. In the examples below, do you know which of the if will consider the condition true?

if [true, true, true, true]     %case 1
if [true, true, true, false]    %case 2
if [false, true, true, true]    %case 3
if [false, false, false, false] %case 4

It's only the first one. Thankfully for you, that's what you actually meant but it's better to be explicit and pass a scalar to if with:

if all(some_logical_vector)

Or even better, if you want to check that two vectors are identical use isequal which returns a vector, so

if isequal(a(c, 1:4), [1, 2, 3, 4])    

The second and bigger problem is your made up syntax a == b | c. This is never going to mean if a is equal to b or a is equal to c (even if you use brackets). It means if (a is equal to b) or (c is true). And don't think you can use brackets to make it work, a == (b | c) means if a is equal to the result of (b or c is true). The proper way to write the expression is explicitly:

if a == b | a == c

So, for your test:

if isequal(a(c, 1:4), [1, 2, 3, 4]) || isequal(a(c, 1:4), [1,2 ,3 ,6])

However, the above can be expressed more simply with ismember, with the 'rows' option:

if ismember(a(c, :), [1, 2, 3, 4; 1, 2, 3, 6], 'rows')

[abhisrisai]

Hi,

Thank you for the explanation. I now realize the mistakes I've made. This gives me better clarity. :)

Answer 2

[Andrei Bobrov]

lo = a(:,3) == 4;
ii = cumsum([0;diff(lo) == 1]).*lo;
m = max(ii);
ii(ii == 0) = m + 1;
out = a(ii >=  m,:);

[abhisrisai]

Thank you for the answer. :)

Answer 3

[Mariano]

a  =  [1 2 3 4     % start point A
      1 2 4 5     % delete
      1 2 4 5     % delete
      1 2 4 5     % delete
      1 2 4 5     % delete
      1 2 3 5     % don't delete
      1 2 3 6     % start point B
      1 2 4 5     % delete 
      1 2 4 5     % delete
      1 2 3 5     % don't delete
      1 2 3 7     % don't delete anything from below here. 
      1 2 4 5
      1 2 4 5
      1 2 4 5 
      1 2 4 5 
      1 2 3 5];
  sp = find(ismember(a(:,4),[4,6])); % Find starting points
  ep = find(ismember(a(:,4),[6,7])); % Find end points
  n  = length(ep); % How many? (assume number staring points = number 
                   % end points)
  dp = find(a(:,4)==5); % Candidates to be deleted
  dr=[]; % DOES SOMEONE KNOW HOW TO VECTORIZE NEXT LOOP?
  for j = 1:n
      dr = [dr(:);dp(dp>sp(j)&dp<ep(j)-1)]; % Only delete rows between 
                                            % starting and end points
                                            % Do not delete immediately
                                            % before end point
  end
  a(dr,:)=[]

[abhisrisai]

Thank you for the answer. :)