How can I output specific point of data to a sparse matrix.

Question

Matthew 2013-5-20

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/76375-how-can-i-output-specific-point-of-data-to-a-sparse-matrix

采纳的回答： Cedric

在 MATLAB Online 中打开

Hi all,

I have a sparse matrix:

a = sparse(9,9)

I also have two sets of information:

i =

     7
     8
     9

j =

     1
     2
     6

So, I want to take i as a row reference and j as a column reference and output it to the sparse matrix. So, if I do:

a(i,j) = -1

Output Window:

a =

   (7,1)       -1
   (8,1)       -1
   (9,1)       -1
   (7,2)       -1
   (8,2)       -1
   (9,2)       -1
   (7,6)       -1
   (8,6)       -1
   (9,6)       -1

When I only want:

a =

   (7,1)       -1
   (8,2)       -1
   (9,6)       -1

or, I only to output to a when position of i is equal to position of j. I have tried certain things with "for" and "if" loops but haven't been able to come up with the right answer.

Can anyone please help me with this, thank you in advance.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Cedric 2013-5-20

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/76375-how-can-i-output-specific-point-of-data-to-a-sparse-matrix#answer_86031

编辑：Cedric 2013-5-20

在 MATLAB Online 中打开

It is much more efficient to build the matrix directly with "i", "j" as indices of non zeros elements and a vector of values (or a scalar), rather than building an empty matrix and indexing it later to set values. Example:

 >> I = [4,2,8] ;
 >> J = [1,7,3] ;
 >> V = -1 ;
 >> n = 9 ;
 >> S = sparse(I, J, V, n, n)
 S =
   (4,1)       -1
   (8,3)       -1
   (2,7)       -1
 >> full(S)
 ans =
     0     0     0     0     0     0     0     0     0
     0     0     0     0     0     0    -1     0     0
     0     0     0     0     0     0     0     0     0
    -1     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     0
     0     0     0     0     0     0     0     0     0
     0     0    -1     0     0     0     0     0     0
     0     0     0     0     0     0     0     0     0

Now what you did wrong in your attempt is not related to sparse matrices, but to indexing. You could use e.g. SUB2IND to get a linear index based on "i" and "j", which would solve your problem.

2 个评论
显示无隐藏无

Matthew 2013-5-21

Thanks for your help Cedric it is much appreciated.

Cedric 2013-5-21

You're welcome! Let me know if you have any additional question related to this matter, and please [ Accept ] the answer if it fully solved your problem.

请先登录，再进行评论。

Answer 2

Iain 2013-5-20

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/76375-how-can-i-output-specific-point-of-data-to-a-sparse-matrix#answer_86025

The simple answer:

for k = 1:numel(i)

a(i(k),j(k)) = -1;

end

1-D or logical indexing is probably better suited.

4 个评论
显示 2更早的评论隐藏 2更早的评论

James Tursa 2013-5-21

编辑：James Tursa 2013-5-21

No, he hasn't. That's the point. The command a = sparse(9,9) only creates an empty skeleton of a sparse matrix with no memory pre-allocated for any data. To get the data memory preallocated one would have to do something like a = sparse([],[],[],9,9,numel(J)). For a small matrix this data memory preallocation doesn't make much difference, but for a large matrix it can make a significant difference.

Another problem with the loop is sorting the data. Every time you add an element to the matrix in a loop MATLAB has to resort the internal data to make room for the new value. The loop forces MATLAB to do this one element at a time, which can cause repeated data copying of large chunks of memory again. Cedric's method above let's MATLAB do the necessary sorting in one fell swoop internally.

Cedric 2013-5-21

编辑：Cedric 2013-5-21

在 MATLAB Online 中打开

To illustrate:

 >> n = 1000 ;                              % Square mat. size.
 >> n_nz = 3*n ;                            % # non-zero elements.
 >> A_full = zeros(n, n) ;                  % Full, for comparison.
 >> A_spalloc = spalloc(n, n, n_nz) ;
 >> A_sparse_prealloc = sparse([], [], [], n, n, n_nz) ;
 >> A_sparse = sparse(n, n) ;
 >> whos
  Name               Size           Bytes     Class     Attributes
  A_full             1000x1000      8000000   double              
  A_spalloc          1000x1000      56008     double    sparse    
  A_sparse           1000x1000      8024      double    sparse    
  A_sparse_prealloc  1000x1000      56008     double    sparse    
  n                  1x1            8         double              
  n_nz               1x1            8         double