Put values from an array in a while loop into a table

I have a while loop that produces an array of values in each iteration. What I need is for this array to be output in tabular form, ideally into an Excel table, into a new row for each iteration. Here's what the while loop looks like:

while any(maxnrep~=0)
    f = [zeros(1,n) 1];
    intcon = 1:n;
    % A*w + x == s
    Aeq = [A 1];
    beq = s;
    lb = [zeros(1,n) 0];
    hb = [maxnrep, Inf];
    wx = intlinprog(f, intcon, [], [], Aeq, beq, lb, hb, options);
    % number of coins to approximate s
    w=round(wx(1:n))';
    % the residual x
    x = s - sum(w.*A); % == wx(n+1)
    if any(w~=y)
        iterCount_sub=iterCount-iterCount_sub;
        if c~=1
            fprintf ('Repeat %.0f times \n', iterCount_sub)
        end
        fprintf ('\nWay number %.0f\n', c);
        fprintf(['%.1f = ' repmat('%.1d*%.0f + ', 1, n) '%.0f\n'], s, [w; A], x)
        c=c+1;
        iterCount_sub=iterCount;
    end
    maxnrep = maxnrep - w;
    y=w;
    iterCount=iterCount+1;
end 

The array I need to get into a table is w. I also need separate columns for x and iterCount_sub. As of now, the output looks like this:

Way number 1
12000.0 = 0*6690 + 0*4740 + 0*5990 + 0*6540 + 0*4840 + 0*3240 + 3*4000 + 0
Repeat 609 times 

Way number 2
12000.0 = 0*6690 + 1*4740 + 0*5990 + 0*6540 + 0*4840 + 1*3240 + 1*4000 + 20
Repeat 2 times 

Way number 3
12000.0 = 0*6690 + 0*4740 + 2*5990 + 0*6540 + 0*4840 + 0*3240 + 0*4000 + 20
Repeat 120 times 

etc..

Here, each value that goes before the * sign is part of the w array, and each value that goes after it is part of the A array, the values for which are input by the user.

What I need in the end should look something like this, along with an Excel table:

   6690  4740  5990  6540  4840  3240  4000   x   rep. count
   ----  ----  ----  ----  ----  ----  ----  ---  ----------
     0     0     0     0     0     0     3     0     609
     0     1     0     0     0     1     1    20      2
     0     0     2     0     0     0     0    20     120
 
%   Is this possible?

Answer 1

[Stephen23]

There is no need to create lots of separate vectors, doing so would be very inefficient.

Much simpler to convert the matrix using ARRAY2TABLE and then save the table. Note that the table variable/columns names do not need to have 'c' as lead character.

M = [0,0,0,0,0,0,3,0,609;0,1,0,0,0,1,1,20,2;0,0,2,0,0,0,0,20,120]

M = 3×9

0 0 0 0 0 0 3 0 609 0 1 0 0 0 1 1 20 2 0 0 2 0 0 0 0 20 120

S = ["6690","4740","5990","6540","4840","3240","4000","x","rep.count"];
T = array2table(M,'VariableNames',S)

T = 3×9 table
    6690    4740    5990    6540    4840    3240    4000    x     rep.count
    ____    ____    ____    ____    ____    ____    ____    __    _________

     0       0       0       0       0       0       3       0       609   
     0       1       0       0       0       1       1      20         2   
     0       0       2       0       0       0       0      20       120   

writetable(T,"coefficienttable.xlsx")

[Vladislavs Grigorjevs]

Ok, I like this a little better, but the same issue as with Simon's approach still stands - I can't predefine the matrix M as its size and values in it depend on the amount of while loop iterations and their results. Is there any way to automate its creation?

As for the S array, would it work if I defined it as

S=[A,"x","rep.count"];

As I've mentioned previously, the array A is defined by the user so I have no way of predicting its size or the values that go into it.

[Stephen23]

"I can't predefine the matrix M as its size and values in it depend on the amount of while loop iterations and their results. Is there any way to automate its creation?"

The obvious approach is to just append the rows on each iteration:

M = [];
while ..
    .. your code    
    M = [M;w];
end

Or do the same thing with a cell array (might be more efficient):

C = {};
while ..
    .. your code    
    C{end+1} = w;
end
M = vertcat(C{:});

https://www.mathworks.com/help/matlab/matlab_prog/comma-separated-lists.html

https://www.mathworks.com/matlabcentral/answers/1656435-tutorial-comma-separated-lists-and-how-to-use-them

More robust to call STRING explicitly:

S = [string(A),"x","rep.count"];

[Vladislavs Grigorjevs]

Thank you so so much for your help! This works exactly as it needs to. One last question, I'm pretty sure I screwed up a little bit with the structure of my code when I was first writing it, but here's what happens when I get the results:

T =

  9×9 table

    6690    4740    5990    6540    4840    3240    4000     x      rep.count
    ____    ____    ____    ____    ____    ____    ____    ____    _________

     0       0       0       0       0       0       3         0         0   
     0       1       0       0       0       1       1        20       609   
     0       0       2       0       0       0       0        20         2   
     1       0       0       0       1       0       0       470       120   
     1       1       0       0       0       0       0       570        33   
     1       0       0       0       0       1       0      2070        62   
     0       0       0       1       0       1       0      2220        33   
     0       0       0       0       0       3       0      2280        40   
     0       0       0       0       0       1       0      8760       308   

The issue is with the last column rep.count. Initially the value that is displayed there (iterCount_sub) is set to 0, which is why the first value in the column is 0. However, the 2nd value is the one that actually corresponds to the first row, 3rd to 2nd and so on. Is there any way I could suppress that first 0 value and move all of the values up 1 row? The last row doesn't need a rep.count value.

I will repost the while loop code in case it is needed below.

while any(maxnrep~=0)
    f = [zeros(1,n) 1];
    intcon = 1:n;
    % A*w + x == s
    Aeq = [A 1];
    beq = s;
    lb = [zeros(1,n) 0];
    hb = [maxnrep, Inf];
    wx = intlinprog(f, intcon, [], [], Aeq, beq, lb, hb, options);
    % number of coins to approximate s
    w=round(wx(1:n))';
    % the residual x
    x = s - sum(w.*A); % == wx(n+1)
    if any(w~=y)
        iterCount_sub=iterCount-iterCount_sub;
        M = [M;w,x,iterCount_sub];
        iterCount_sub=iterCount;  
    end
    maxnrep = maxnrep - w;
    y=w;
    iterCount=iterCount+1;
end 

[Stephen23]

"Is there any way I could suppress that first 0 value and move all of the values up 1 row?"

You could either modify your code to include that offset, or modify the data after the loop, e.g. something like this:

M(:,end) = [M(2:end,end);NaN];

The last row doesn't need a rep.count value."

But numeric arrays cannot have "holes" in them, so there has to be some value there. What value depends on you. As an alternative you could replace the entire table column/variable with a cell array of scalar numerics and place an empty array in the last cell.

[Vladislavs Grigorjevs]

This works perfectly, thank you so much once again!

Answer 2

[Simon Allosserie]

To put info in tables, you best put them in columns. So you could define coefficient vectors for each coefficient, and through your loops add values to these column vectors. In your example, they would look like this

c6690 = [0 0 0]' %a variable name can't start with a number, so I start the name with c(oefficient)
c4740 = [0 1 0]'
c5590 = [0 0 2]'
..
repcount = [609 2 120]'

Then you define a table like (see https://nl.mathworks.com/help/matlab/ref/table.html)

T = table(c9960, c4740, .., repcount)

Next write it to Excel using writetable (see https://nl.mathworks.com/help/matlab/ref/writetable.html)

writetable(T,"coefficienttable.xlsx")

That should do the trick.

[Vladislavs Grigorjevs]

I see, but there is a problem with this approach. The thing is that the array A (the one where the values 6690, 4740 etc. are) is defined by the user and is not always going to be the same size. This means that I would have to somehow automate the creation of those coefficient vectors, so the script creates them according to the input data.

Secondly, the size of those vectors is also dependent on the while loop itself, there is no way I can predict how many iterations it will have to go through beforehand. This means that the length of the vectors would have to be automatically increased with each added value, and I am not quite sure how to do that.

[Simon Allosserie]

In that case, you could make one matrix instead of column matrices, where the size is dependent on the user input. The user should then also give the column names such that you can define the table as

T = table(M, 'VariableNames', {'c6690','c4740', .., 'repcount'})

Of course, you can also create the variableNames yourself by converting the coefficients to text with num2str() and adding 'c' before it with strcat()

For the length of the vectors/size of the matrix, there are multiple options.

1) You can predefine the length very largely, e.g.

c = zeros(1000,1)

after calculations, you can shorten the matrices to delete the zero elements.

2) You can just add elements in each iteration

c(i) = value

and in the case c is 10 long, and you add c(11), matlab will automatically make c longer to be able to add this element on that location. The drawback here is that this takes a lot of time, so option (1) is recommended.

[Stephen23]

"...you can define the table as T = table(M, 'VariableNames', {'c6690','c4740', .., 'repcount'})"

No, that will not work. Lets try it with a simple example:

M = rand(5,3);
T = table(M, 'VariableNames', {'c6690','c4740','repcount'})

Error using table
The VariableNames property must contain one name for each variable in the table.

Why does it not work? Because it creates a table with one variable/column, but then attempts to name three variables. Thus the number of variable names does not match the number of variables/columns, as the error clearly states.