# Inserting a column in a matrix without deleting any column

1,606 views (last 30 days)
joseph Frank on 9 Feb 2011
Commented: Praanesh Sambath on 8 Mar 2020 at 13:05
Hi,
I am wondering of there is a function that enables me to insert a column in a matrix (similar to insert column in excel) without replacing any existing column. For example I have Matrix "A" which is 4x4 and I want to insert a vector "B" after the second column of "A" in order to obtain "A" 4x5 where the 3rd column of "A" is vector "B"

Martijn on 9 Feb 2011
Suppose you have:
>> A=reshape(1:16,4,4)
A =
1 5 9 13
2 6 10 14
3 7 11 15
4 8 12 16
>> B=(17:20)'
B =
17
18
19
20
Then you could obtain the desired matrix C by:
>> C = [A(:,1:2) B A(:,3:4)]
C =
1 5 17 9 13
2 6 18 10 14
3 7 19 11 15
4 8 20 12 16
So you take the first two columns of A concatenate the column B and then concatenate the last two columns of A.
You can generalize this a bit into:
>> D = [A(:,1:N) B A(:,N+1:end)]
Where N then stands for "insert B after the Nth column".

thanks!
Rene M on 26 Feb 2020
How do I do this in a cell array?
Praanesh Sambath on 8 Mar 2020 at 13:05
how do I do this if I want B to be in between all the clumns of A

Izi on 19 Dec 2012
What about inserting a column which dimensions are not consistent?

#### 1 Comment

Sarah on 1 Sep 2016
In MATLAB, a matrix must have internally consistent dimensions. To insert a column like in the example above, you would have to make sure (one way or another) that it had the same number of rows as the rest of the matrix. You would either have to pad it by adding zeros, NaNs, or some distinct number (e.g., -99999) to make up the length if it was too short, or you'd have to effectively pad the matrix if the column was too long.
I suppose you could also create a column of the appropriate length and interpolate so that the resolution of your new column matched that of the matrix-- there's a function in the Image Processing toolbox (imresize) that lets you specify the size you'd like a 2D matrix to be and appropriately re-grids some input dataset. I can't think of why this wouldn't work for a single column vector.