Sum of rows with whose row numbers are specified by another array

I have an array of the form:

TTX =

    20     9     2     7
    40    10    10     1
    60     2    10     9
     0    10     5    10
    40     7     9     7
    40     1     2     8
     0     3     5     8
    60     6    10     4
    20    10     8     7
     0    10    10     2

I want to add the elements of rows 1:3, then add the elements of the rows 4:5, then 6:end.

How can I do that?

[Stephen23]

@Saeid: you just editted your question, removed the original examples, and asked something different from what you originally asked. This discourages people from helping you.

I already answered your original question, but now my answer makes no sense because you removed all of the relevant information.

In future please use comments for adding new information.

[Saeid]

OK!

Answer 1

[Stephen23]

From R2015b you can use splitapply:

>> TTX = [...
    20     9     2     7
    40    10    10     1
    60     2    10     9
     0    10     5    10
    40     7     9     7
    40     1     2     8
     0     3     5     8
    60     6    10     4
    20    10     8     7
     0    10    10     2];
>> [U,~,X] = unique(TTX(:,1));
>> Y = splitapply(@(x)sum(x,1),TTX(:,2:4),X);
>> M = [U,Y]
M =
     0    23    20    20
    20    19    10    14
    40    18    21    16
    60     8    20    13

EDIT: this matches the original question's output explanation and example.

[Saeid]

Thanks Stephen, this is exatly what I want, but I tried it now with another array and received an error message. The only difference here is that the new array has 14 rows instead of 10. Here the code:

xx=ceil((rand(14,3))*10)
TT=[20 40 60 0 40 40 0 60 20 0 10 40 40 0]'
ttx=[TT xx]
[U,~,X] = unique(ttx(:,1))
ttx(:,2:4)
Y = splitapply(@sum,ttx(:,2:4),X)
M = [U,Y]

What am I doing wrong?

[Saeid]

Guess I found the answer myself: the splitapply line should be changed to:

Y = splitapply(@(x) sum(x,1),ttx(:,2:4),X)

But thanks anyway, your response was what I wnated.

Saeid

[Stephen23]

"the splitapply line should be changed to:"

You are quite right: if any group has only one line then that error will be thrown (because sum then sums along that row, and not down the columns).

I updated my answer to use this more robust function.

[Saeid]

Then please remember to accept my answer!

By all means, you are always a great help!

[Saeid]

BTW, Stephen, is it possible to apply splitapply to repmat, so that I could repeat different columns for different numbers of times? E.g., repeat column 1 for 3 times, then repeat column 2 for 5 times, column 3 for 2 times and so on, and the number of repeats given in an array like: [3 5 2 ...]?

[Stephen23]

@Saeid: read about repelem, use it to generate an index vector. Apply that indexing vector to the column indices of your array:

>> R = [3,5,2];
>> X = repelem(1:numel(R),R);
>> M = randi(9,5,3)
M =
     2     6     7
     4     1     7
     9     8     4
     8     9     6
     9     7     2
>> M(:,X)
ans =
     2     2     2     6     6     6     6     6     7     7
     4     4     4     1     1     1     1     1     7     7
     9     9     9     8     8     8     8     8     4     4
     8     8     8     9     9     9     9     9     6     6
     9     9     9     7     7     7     7     7     2     2

[Saeid]

Thanks again, Stephen!