concatanate arrays upon condition

Dear MatLAb Community,

I would like to concatanate selected coloumns of arrays, some of which are empty arrays, and this is varying dataset by dataset.

How can I exclude the empty arrays from concatanation?

Let's say now , distr3 and distr6 are empty but this my vary.

distr = cat(1,distr1(:,ch1),distr2(:,ch2),distr3(:,ch3),distr4(:,ch4),distr5(:,ch5),distr6(:,ch6),distr7(:,ch7),distr8(:,ch8));

Thans for you suggestions

BEst

lg

[Stephen23]

"Thans for you suggestions"

Don't store your data in lots of separate variables with numbers in their names.

https://www.mathworks.com/matlabcentral/discussions/tips/849901-tutorial-why-variables-should-not-be-named-dynamically-eval

Answer 1

[Matt J]

The way your data is held, as separate enumerated variables, doesn't leave many options. The first thing would be to undo that:

distrCell = arrayfun(@(i)eval("distr"+i),1:8,'uni',0);  %should be unnecessary
chCell = arrayfun(@(i)eval("ch"+i),1:8,'uni',0);     %should be unnecessary

Then, you can do,

keep=~cellfun('isempty',distrCell);

distrCell=distrCell(keep);
chCell=chCell(keep);

distr=cell2mat( cellfun(@(d,c) d(:,c), distrCell,chCell ,'uni',0 )   )

[Levente Gellért]

Dear Matt J, first I started to understand and then run your first line. Although the distr1 variable is in the workspace, the following error message comes:

Error using eval

Undefined function or variable 'distr1'.

Error in @(i)eval("distr"+i)

matlab.internal.language.introspective.errorDocCallback('@(i)eval(

↑

Error: Character vector is not terminated properly.

Can you help further?

best

lg

[Matt J]

Modify along the following lines:

distr1=17; distr2=5; distr3=68;

distrCell = arrayfun(@(i)evalin('caller',"distr"+i),1:3,'uni',0)  %should be unnecessary

distrCell = 1×3 cell array
    {[17]}    {[5]}    {[68]}

Or, even better, recreate the data from scratch, but instead of distr1, distr2,... store the arrays in a distrCell variable directly.

Answer 2

[Walter Roberson]

At least in R2025a, there is no problem in using "," to concatenate empty datasets, as long as they are empty along either the first or second dimension. (There can be problems if they are empty along the third dimension and the sizes of the first and second dimension do not match the existing data.)

A = zeros(5,3);
B = ones(5,0);
C = zeros(5,2);
D = []; size(D)

ans = 1×2

0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

E = ones(3,0);
F = ones(0,3);
G = ones(10,9,0);

[A,B,C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

●

[A,D,C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[A,E,C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[A,F,C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[A,G,C]

Error using horzcat
Dimensions of arrays being concatenated are not consistent.

[Walter Roberson]

I just noticed that the original question was about cat(1,...) which is equivalent to using semi-colon inside []

A = zeros(3,5);
B = ones(0,5);
C = zeros(2,5);
D = []; size(D)

ans = 1×2

0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

E = ones(0,3);
F = ones(3,0);
G = ones(10,9,0);

[A;B;C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[A;D;C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

●

[A;E;C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[A;F;C]

ans = 5×5

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[A;G;C]

Error using vertcat
Dimensions of arrays being concatenated are not consistent.

[Levente Gellért]

Dear Walter Robertson, the concatenation was not problematic as long as I concatenated the whole array set, some of which is emty in a varying manner. But if I wanted to concatenate only selected coloumns of these arrays, defining a coloumn in an empty array is meaningless and gave the error.

best lg

[Walter Roberson]

@Levente Gellért

Is the error you recieved due to the fact that you are subscripting an empty array at columns that do not exist?

T1 = randi(9,5,3)

T1 = 5×3

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

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

T2 = randi(9,7,4)

T2 = 7×4

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

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

T3 = randi(9,4,1)

T3 = 4×1

2 2 4 4

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

●

T4 = randi(9,6,3)

T4 = 6×3

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

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

S = @(A,c) A(:,c:c.*(c<=size(A,2)));

%use the auxillary function S to select columns
[S(T1,2); S(T2,2); S(T3,2); S(T4,2)]

ans = 18×1

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

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

%whereas directly indexing might fail
[T1(:,2); T2(:,2); T3(:,2); T4(:,2)]

Index in position 2 exceeds array bounds. Index must not exceed 1.

[Walter Roberson]

Huh, there is a difference between cat(1,A,B) and [A;B] ...

A = zeros(5,1);
B = ones(3,0);
[A;B]

ans = 5×1

0 0 0 0 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

cat(1, A, B)

Error using cat
Dimensions of arrays being concatenated are not consistent.