could anyone help me to solve the error in the following code

I am having the following matrix

A=[0.2288    0.2193    0.2082    0.1961    0.1832    0.1699;
    1.1237    1.0564    0.9837    0.9092    0.8392    0.7826;
    0.0451    0.0401    0.0331    0.0245    0.0152    0.0092;
    0.8840    0.9053    0.9243    0.9396    0.9501    0.9548;
    0.1320    0.1442    0.1545    0.1628    0.1689    0.1731]

I want to find two maximum values in each column

so i tried with the following code:

max_values = 2;
[v1,idx1]=sort(A,1,'descend');
W1 = zeros(size(A));
W1(sub2ind(size(A),repmat((1:size(A,2)).',...
    1,2),idx(:,1:max_values)))  = v1(:,1:max_values)

when i run the code i am getting error stating Error using sub2ind The subscript vectors must all be of the same size.

Could anyone please help me on this.

[Walter Roberson]

Exactly what do you want output?

And what do you want to do if there are multiple copies of the maximum?

[jaah navi]

I want to display first two maximum vaues in each column.

[Alex Mcaulley]

Probably an easier solution is to use maxk (you need Matlab 2017b or later)

[madhan ravi]

Alex, the easy tasks never seem to be easier but harder, 100 percent questions like this OP uses older versions to make the problems as harder as possible :/

[Alex Mcaulley]

I totally agree Madhan XD. Maybe it is useful for other readers

[madhan ravi]

Alex that's completely true though :)

Answer 1

[KALYAN ACHARJYA]

"I want to display first two maximum vaues in each column"

out=sort(A(:,:),'descend');
data=out(1:2,:)

Result:

data =

    1.1237    1.0564    0.9837    0.9396    0.9501    0.9548
    0.8840    0.9053    0.9243    0.9092    0.8392    0.7826

[madhan ravi]

No, Kalyan! The rest of the values needs to be filled with zeros. Not so easily acheived as proposed ;)

[Walter Roberson]

No, madhan, I specifically asked about the output and the user said "I want to display first two maximum vaues in each column.", which is what Kalyan does.

You might have deduced that the user wants to use a matrix of the original size and zero all of the non-maximum entries, since that is what the user's code appears to be trying to do, but that is not in accordance with the user's specific statement as to the desired output.

[jaah navi]

I want to have the maximum values at its own place and the rest of the values needs to be filled with zeros.

[madhan ravi]

True, sir Walter, completely agree with you :) . I assumed since the OP has asked how to obtain the first two max values in each column in one of his previous questions. The code of the OP was given by me in his previous question (but the code gives the desired for each rows). So I thought the OP needs a minute tweak to achieve it for each column.

@Jaah Navi: "I want to have the maximum values at its own place and the rest of the values needs to be filled with zeros."

You should have mentioned this at the beginning of your question., not after when someone has answered your question. This makes the efforts of the answerer go to waste.

[jaah navi]

Sorry for the inconvenience.Its my mistake.

[jaah navi]

Could anyone help me how the code can be executed without error

A = [0.0088    2.0071    0.0058    0.0049    2.0045    0.0045;...
     0.0255    0.0248    2.0241    2.0234    0.0227    2.0218;...       
     1.8321    0.0932    0.3378    0.5582    2.7400    0.8653;...
     0.0666    0.0605    4.0559    0.0526    0.0505    3.0491;...
     0.0322    1.0294    0.0303    5.0334    0.0372    0.0410];
max_values = 2
[v,idx]=sort(A,1,'descend')
W = zeros(size(A))
W(sub2ind(size(A),repmat((1:size(A,2)).',...
    1,2),idx(:,1:max_values)))  = v(:,1:max_values)

I want to have the output in the following manner

W = [0         2.0071    0         0         2.0045    0     ;...
     0         0         2.0241    2.0234    0         2.0218;...
     1.8321    0         0         0         2.7400    0     ;...
     0.0666    0         4.0559    0         0         3.0491;...
     0         1.0294    0         5.0334    0         0     ]

Could anyone please help me on this and I am using 2015a version.

[madhan ravi]

Try this:

max_values = 2;
[v1, idx1] = sort( A, 1, 'descend' );
W1 = zeros( size( A ) );
[m, n] = size( A );
l = bsxfun(@plus, idx1, ( 0 : n - 1 ) * m );
W1( l( 1 : max_values, : ) ) = v1( 1 : max_values, : )

[KALYAN ACHARJYA]

Thank you @Walter and @Madhan

[jaah navi]

thanks a lot.