select complex numbers from cell array based on the real part

Hi all,

I have a 7D cell array with complex numbers.

I want to keep the complex number with the highest real part, of all the values of the 3rd parameter (1:15).

For example in this case i want only the val(:,:,14,1,1,1,1) which has the highest real part.

I tried to do it using the following command

maxVal = max(cell2mat(my_array),[],3);

However, it keeps the complex number with higher imaginary part. In this case

val(:,:,15,1,1,1,1)

Any idea of how can i do this?

Thank you

[Stephen23]

Why are you storing scalar numerics in a cell array?

[Ilias Minas]

Its the solution of the eigenproblem. I convert it later to matrix.

Answer 1

[dpb]

Well, if you want the max() of the real part, then you'll have to limit to only looking at the real part -- max() for complex returns max(abs())

tmp=squeeze(cell2mat(my_array));
[maxRealVal,imxR] = max(real(tmp));
maxVal=tmp(imxR);

[Ilias Minas]

i want to find the corresponding imaginary parts to the maximum real values of the complex numbers as well.

[Walter Roberson]

The maxVal that dpb's code returns will be complex values. You can imag() them.

[Ilias Minas]

The problem with the maxVal is that keeps the complex number with higher abs value either its real or imaginary.

It returns complex number but in the example showed above i want the corresponding imaginary part of the highest real part, and it returns the complex number with the higher value of imaginary part.

The one that i want to extract is

val(:,:,14,1,1,1,1)

[7.5646e+02 + 1.4443e+03i]

and it gives me back

val(:,:,15,1,1,1,1) =

{[3.4384e+01 + 1.6918e+03i]}

which has the absolute highest imaginary part.

[dpb]

With your code, yes...with mine, no. Attach some readable data so don't have to make it up (don't use images, select/past text instead).

Well, what the ... poke, poke, ...

val=complex([0.21,0.28,0.34,745.4,34.3],[1460.9,1460.9,1460.9,1444.3,1691.9]);  % approx subset sample data
% find return value with maximum real part...
[vMxR,iMxR]=max(real(val));       % return location 
% display result
>> val(iMxR)
ans =
   7.4540e+02 + 1.4443e+03i
>> imag(val(iMxR))
ans =
   1.4443e+03
>> 

by contrast,

>> abs(val)
ans =
   1.0e+03 *
    1.4609    1.4609    1.4609    1.6253    1.6922
>> [~,imx]=max(abs(val))
imx =
     5
>> [~,imx]=max((val))
imx =
     5
>> 

which shows since the real part is stlll small(ish) in comparison, the larger imaginary component dominates in the abs() value which is what max() returns for complex variable.