CODEWITHSUNDEEP
performancematlab
Yas
Yas

Reputation: 901

Accelerate a loop operation using table in MATLAB

I am working with two tables in MATLAB:

A1_Table   [m*r]
A2_Table   [m*q]

where the tables have different number of columns but similar number of rows as you can see and it follows that r>q. In fact A2_Table is a sub-sample of A1_Table, meaning all the matrix entries in A1_Table can be found in A2_Table as well.

Associated with each column is a VariableName and with each row a RowName. What I want to do is to set all the entries in A1_Table to zero for which A2_Table has a value greater than zero. Here is what I perform:

 [r,c] = find(table2array(A2_Table)) ;

  for i = 1 : length(r)
       A1_Table(A2_Table.Properties.RowNames(r(i)),A2_Table.Properties.VariableNames(c(i))) = {0} ;
  end

As you can see I am accessing the corresponding entries in A1_Table through the rows and column names (keywords). This process works but it is very slow as there exists 2 million values for which the loop needs to run. Any idea to accelerate this process? Thanks

Upvotes: 0

Views: 44

Answers (1)

Edwin Torres
Edwin Torres

Reputation: 2864

Here's one way. It seems to scale well:

clear all;
clc;

m=3;
r=6;
q=3;

A1_Table = randi(10,m,r); 
A2_Table = randi([0,1],m,q); 
Zero_Table = zeros(m,q);
A1_Table
% remove extra columns from A1_Table
A1_Table(:,q+1:r) = [];
A2_Table
% compare A2 table to zeros table
compArr = A2_Table <= Zero_Table;

% compare compArr matrix to A1_Table
result = A1_Table.* compArr;
result

Upvotes: 1

Related Questions