how to use index referencing ?

I have A =

            [1      -2      1     0.7    3
             1      -0.1   -3     2     -0.2
             0       0      1     1     -2 ]

I want to choose the most negative of each column, and drop the rest. If a column doesn't have negative values then it is zero. Thus, the above matrix will be

A =

             [ 0    -2     0    0   0
               0     0    -3    0   0
               0     0     0    0  -2 ]

Also, I want to keep the values of the same index in the x matrix.

x =

        [ 440   680   520   240   730
          440   680   520   240   730
          440   680   520   240   730 ]

. Thus, x will be

x =

       [  0   680     0    0     0
          0    0     520   0     0
          0    0      0    0   730 ]

Thanks for your quick responses,

Answer 1

[Jos (10584)]

Assuming A and X are your input arrays

szA = size(A) 
B =  zeros(szA)
[minA,Rix] = min(A,[],1) % minimum value per column, and the row index
LinIDX = sub2ind(size(A), Rix, 1:szA(2))
B(LinIDX) = minA
B = min(B,0) % for cases when all elements in a column of A are positive

X2 = (B<0) .* X

[Islam]

but this answer doesn't drop negative values if they are not the minimum in the column. 

Each column should he one value only. but in your solution I have more than one value ..

[Jos (10584)]

The result is stored in variable B not in A. Each column of B holds a single value, namely the minimum value in that column of A, as you requested. If you want to have it stored in A, you can always execute:

A = B

Answer 2

[Sean de Wolski]

A = [1      -2      1     0.7    3
             1      -0.1   -3     2     -0.2
             0       0      1     1     -2 ];
x = rand(size(A));
B = bsxfun(@times,bsxfun(@eq,A,min(A,[],1)),min(A,0));
x(B==0) = 0;