How can I insert many columns into a matrix at different locations efficiently?

Question

Darcy Cordell 2022-9-26

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1812630-how-can-i-insert-many-columns-into-a-matrix-at-different-locations-efficiently

评论： Chunru 2022-9-28

I have a matrix in which I need to delete several columns of zeros, perform some operation (i.e. matrix inverse), and then re-insert the zero columns back into the matrix at the same position. My matrix is very large, so I don't want to store a duplicate of the matrix in memory. For example:

%Create some "data"
A = rand(100); %some random data
A(:,randsample(100,15))=zeros(100,15); %some random columns of zeros 
%Now that we have "data", let's find the zero columns:
col = find(~all(A));
%Delete the zero columns
A(:,col) = [];
%Now do some stuff to A ....
%Now I want to re-add the columns in????

One simple option to solve the problem is:

for i = 1:length(col)
    A = [A(:,1:col(i)-1) zeros(100,1) A(:,col(i):end)];
end

But I feel like there must be a more elegant way to do it without the for loop that can essentially "reverse" the deletion. Any help is appreciated.

3 个评论
显示 1更早的评论隐藏 1更早的评论

James Tursa 2022-9-27

Can you give us details about what operations you are performing on the non-zero columns of A?

Darcy Cordell 2022-9-27

I need to invert the non-zero columns. I've updated the answer accordingly.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Chunru 2022-9-27

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1812630-how-can-i-insert-many-columns-into-a-matrix-at-different-locations-efficiently#answer_1061245

编辑：Chunru 2022-9-27

在 MATLAB Online 中打开

%Create some "data"
A = rand(100); %some random data
col = randsample(100,15);
A(:, col)=zeros(100,15); %some random columns of zeros 
%Now that we have "data", let's find the zero columns:
col = find(all(A==0))
col = 1×15
     1     2     4     5     7    20    33    35    41    49    67    69    70    86    92
% No need to delete the zero columns
% Just re-assign anything back to A
%
% This is a faster approach since array size is kept the same.
%
% For example
A(:, col) = ones(size(A,1), length(col));

2 个评论
显示无隐藏无

Darcy Cordell 2022-9-27

The problem is that I need to perform some operations on the non-zero columns of A (i.e. matrix inverse). So I can't replace the columns with ones, I need to delete the columns prior to taking the inverse.