Add index matrix to matrix

I have some algorithm including a while loop::

 while  (condition)
    .......................................................
    .......................................................
    .......% do something and return a result array B...
 end

Example: I MUST run the code (while...end). For instance, my code will loop 4 time to satify the condition. Therefore, I will get the result of each loop is {B1,B2,B3,B4}

Let's say:

 -Loop 1: B1=[1 2 3 4 5....9]; 
 -Loop 2: B2=[10 11 12....15]; 
 -Loop 3: B3=[16 17 18 19]   ; 
 -Loop 4: B4=[20 21 22....30];

The question is: from the result matrix Bi (at each loop):

1/ How to create array index_Bi according to the number of element in array Bi as follows:?

 index_B1= [1 1 1 .....1]  {9 element 1}
 index_B2= [2 2 2.....2]  {6 element 2}
 index_B3= [3 3 3 3]       {4 element 3}
 index_B4= [4 4 4.....4]  {11 element 4}

_ Note: the number of array "index_Bi" depend on the number of loop (i). In this example, Assumed that the number of loop i=4._

2/ After that, How to add two array (Bi & index_Bi) to matrix A, we will have the result:

 A=[ B1  index_B1         
     B2  index_B2        
     B3  index_B3     
     B4  index_B4]

I hope the result of matrix A will be as follows:

 A=[ 1  1         
     2  1        
     3  1     
     ...1
     9  1 
    10  2
    11  2
    ....2
    15  2
    16  3
    17  3
    18  3
    19  3
    20  4
    21  4
    .....
    30  4]

In the above example, I assume the algorithm is just loop 4 time (i=4). { But actually, For my real data, "the while loop" may be loop up to 100 times (or more). Therefore, i may be equal or larger than 100}*

[KL]

What about your other post?

https://de.mathworks.com/matlabcentral/answers/354585-add-matrix-to-matrix?s_tid=prof_contriblnk

You're going to ignore all the suggestions and advices and just going to ask another question?

[ha ha]

Sorry. I think my old post is not clear. After I read many advices, I think my question is not good and not clear.Someone misunderstood the question

[Stephen23]

Do not try to create or access variable names inside the while loop. You should read this to know why, and what the better alternatives are (hint: indexing):

https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

[ha ha]

Do you have any suggestion for my question? Thanks

[Walter Roberson]

This reads like a homework question.

[ha ha]

Hi Walter Roberson,

It doesn't matter "homework,..." or whatever. If you can help me to find the answer, you're appreciate alot. THANKS

[Walter Roberson]

It is difficult for me to make suggestions when you do not answer my questions.

[KL]

It's not bad if you say it's a homework. Just so that we could help you accordingly. Some of the homework questions could be solved with some built-in Matlab functions without having to write multiple lines of code. Mind you, we are only here to help.

[ha ha]

Dear KL, Thank you so much for your help. But in this code, I need to use (while...end) loop to find each array Bi at each loop.

Actually, my algorithm is very long and complicated. In this post, I try to simplify it too easy to understand what I need help from you alls. Please help me.

Answer 1

[Andrei Bobrov]

B = [];
ii = 1;
x = [9 6 4 11];
add1 = 0;
while ii <= 4
    add1 = add1(end) + (1:x(ii));
    B = [B;{add1}];
    ii = ii + 1;
end

A = [[B{:}]',repelem((1:numel(B))',cellfun(@numel,B))];

[ha ha]

Yes, correct. Thanks so much Andrei Bobrov.

[Stephen23]

@ha ha: why do you want to use such a complex solution? My answer is much simpler and more efficient (no repelem, no slow cellfun, no cell array, no conversion from cell array to numeric).

[ha ha]

Sorry @Stephen Cobeldick.

I just see your answer few mins ago. Very sorry. Let's me check carefully your code again, and will inform you later.

[Walter Roberson]

No, this requires changing your code, which you have indicated is not permitted.

[ha ha]

@Stephen Cobeldick and @ Andrei Bobrov

Both your code is GOOD.

BIG THANKS

Answer 2

[Stephen23]

You can do this very simply inside the loop, there is no need to use cell arrays, slow cellfun, or repelem.

out = NaN(0,2);
itr = 1;
while ...
    tmp = ... your code here
    tmp = tmp(:);
    tmp(:,2) = itr;
    out = [out;tmp];
    itr = itr + 1;
end

[Andrei Bobrov]

+1

@ha ha:

A = NaN(0,2);
itr = 1;
while ...
    B = ... your code here 
    B = B(:);
    B(:,2) = itr;
    A = [A;B];
    itr = itr + 1;
end

[ha ha]

Thank @Stephen Cobeldick,

Your code is perfect.

Answer 3

[Walter Roberson]

index_B4= [4 4 4;.....4]  {11 element 4}

The above is not possible. The space between the first two "4" tells us that you have multiple columns, and the ';' after the third "4" tells us that you have multiple rows. In MATLAB, each row must be the same size. Therefore, the total number of elements must be divisible by the number of rows. But the total number of elements is 11, which is a prime number; to have that work out, you would need to have only one column per row -- but we see that there a multiple columns per row, so the total number of elements cannot be prime. If there are multiple rows and multiple columns in an array, then the total number of elements must be a composite number, not a prime.

The creating of 11 copies of element 4 is easy:

repmat(4, 1, 11)

However, this requires that you somehow know that the proper length is 11.

Is the length of each subsection somehow known? Is there something particular about the first element of each subsection (for example if the first element were always negative but the others were always positive) ?

[ha ha]

Sorry, I mistype. Please see the "edit question" ( All B1,B2,B3,B4..., index_B1,index_B2... : is array)

[Walter Roberson]

Is the length of each subsection somehow known? Is there something particular about the first element of each subsection (for example if the first element were always negative but the others were always positive) ?

Once the vector B has been built, there is no way to find out which parts were added by which loop. If you do not have information about how long each section is, and you do not have some way of recognizing a pattern in the data marking the beginning of a block, then there is no way to figure out afterwards which data belongs to which loop index.

[ha ha]

Is the length of each subsection somehow known?

Answer: NO

But, i think my question is possible to find the solution. Why?

Because, if you run (While...end) loop, you will get the array result Bi. For each loop, we just get that array Bi. And then create a new cell including all array Bi for all loop

[Walter Roberson]

You said "Remember: you can not see the result array Bi at each loop" and "we just get the FINAL array B for all loops".

If this were homework, then this would be understood as a condition being imposed by outside for learning purposes, that the point of the question is to figure out how to separate the parts of the array given only the final version of it.

If this were not homework, then the obvious answer would be to change the existing code to generate the counts as we go.

[ha ha]

Dear Walter Roberson,

Thank you so much for your help.

Actually, my algorithm inside the (while...end) loop is very long and complicated. In this post, I try to simplify it too easy to understand what I need help from you alls. But in this code, I need to use (while...end) loop to find each array Bi at each loop. I can not change "the existing code".

Last but not least, the (while...lop) MUST be used to generate the output array B

Please help me.

[Walter Roberson]

Your restrictions make it impossible to solve the problem.

Suppose, for example, that I told you that the following list of numbers was generated in bursts, and I asked you to identify each burst:

1 5 10 7 9 2 4 3 3 6 8 7 8 7 7 3 1

I can offer the additional information that no burst was negative length, and that no burst had length more than 16383. What were the burst boundaries?