element wise logical operators?

I have a question regarding using logical operators on a multi-dimensional array.

I want to firstly test if an element passes a criteria then replace that value with another value given the results of a test.

I want to check the first dimension of the array against the threshold of 0.5 and replace all instances where this is true with a value for only 2 columns of the the 2nd dimension and all cases of the 3rd and 4th dimension. Does anyone know how to do this without multiple for loops?

for example

DA = rand(4,4,2,3);
if DA(:,2:4,:,:)<0.5;
    DA(:,2:4,:,:) = 1;
end

thanks Tim

Answer 1

[Azzi Abdelmalek]

clear
DA = rand(4,4,2,3);
K=arrayfun(@(x) x<0.5,DA(:,2:4,:,:))
B=DA(:,2:4,:,:); 
B(find(K==1))=1;
DA(:,2:4,:,:)=B

[per isakson]

Explicitly operating on a sub array and assign it back is a good approach. However, may I propose a bit of refactoring according to Matt Fig above:

    sub_array = DA(:,2:4,:,:); 

    sub_array( sub_array < 0.5 ) = 1;

    DA(:,2:4,:,:) = sub_array;

Answer 2

[Matt Fig]

Do you mean this?

DA = rand(4,4,2,3);
DA(DA(:,2:4,:,:)<0.5) = 1;

or perhaps you are talking about this (see the comments below):

DA = rand(4,4,2,3);
tmp = DA(:,2:4,:,:);
tmp(tmp<.5) = 1;
DA(:,2:4,:,:) = tmp;

[per isakson]

I run

    DA = rand(4,4)
    DA(:,2:4)<0.5
    DA(DA(:,2:4)<0.5) = 1

and get

    DA =
        0.5313    0.7788    0.1537    0.4574
        0.3251    0.4235    0.2810    0.8754
        0.1056    0.0908    0.4401    0.5181
        0.6110    0.2665    0.5271    0.9436
    ans =
         0     1     1
         1     1     0
         1     1     0
         1     0     0
    DA =
        0.5313    1.0000    1.0000    0.4574
        1.0000    1.0000    0.2810    0.8754
        1.0000    1.0000    0.4401    0.5181
        1.0000    0.2665    0.5271    0.9436
    >>

which is confusing although I understand what is happening.

[Matt Fig]

I wasn't sure whether the OP wants this or something more like:

    DA = rand(4,4)
    tmp = DA(:,2:4);
    tmp(tmp<.5) = 1;
    DA(:,2:4) = tmp

[per isakson]

It turned out, I cannot make sense of the question. OP must help.

However, I found it very hard to grasp

    M( logical_index ) = scalar;

when size(M) is not equal to size(logical_index). One dimension is ok, but four is not.

Is there a way to think about it?

OK, now I got the message: Work with sub arrays!

Answer 3

[Timj2004]

Ok, it appears there is some confussion regarding the question. I appologise. and I appreciate the responses. the idea is to review 3 columns of the array and compare the value to a limit. Then replace the value of the rows in those columns with a new value.

so if my 4D array is as follows DA(simulations, components, alpha, bravo) we have the simulation results (rows) for each component (columns) at each alpha and bravo.

so I want to run a check on 3 components for all simulations and replace the output of the simulation results if it meets a logical criteria.

So DA(sims, [component2 component3 component4] , A, B). lets say DA = rand(4,4,2,2) for now.

so DA(:,2:4,:,:)>0.5 = 1; so if sims in columns 2:4 are greater than 0.5, replace those results with 1.

Thanks.

I was trying to avoid creating too many variables here since I am dealing with arrays of the magnitude of DA(2000,80,50,8) but I guess the best solution has been to use sub arrays. thanks.

   "sub_array = DA(:,2:4,:,:);     
sub_array( sub_array < 0.5 ) = 1;   
 DA(:,2:4,:,:) = sub_array;" - per isakson

-sincerely Tim