Array Concatenation?

Hi

I have two 4 dimensional arrays like so:

Array[480, 241, 60, 124]

The 4th dimension (124 elements) is time. I want to concatenate the 1st timestep for the second array to the first array, something like this:


new=[array1, array2[*, *, *, 0]]

But I can't get the syntax correct to do the concatenation on the 4th dimension.

Any ideas what I'm doing wrong? I've tried various levels of []'s but can't get it working.

Cheers

On 29 juin, 13:02, rj...@le.ac.uk wrote:
> Hi
>
> I have two 4 dimensional arrays like so:
>
> Array[480, 241, 60, 124]
>
> The 4th dimension (124 elements) is time. I want to concatenate the 1st timestep for the second array to the first array, something like this:
>
> new=[array1, array2[*, *, *, 0]]
>
> But I can't get the syntax correct to do the concatenation on the 4th dimension.
>
> Any ideas what I'm doing wrong? I've tried various levels of []'s but can't get it working.
>
> Cheers
>
>

IDL concatenates arrays over the first dimension by using [array1,
array2] construct.
To achieve concatenation in your case, you might transpose your arrays
first, then transpose the result back, as follows:

new = transpose([transpose(array1), (transpose(array2))[0,*,*,*])

The second array must be transposed before you select last column
elements to avoid implicit elimination by IDL of the last dimension in
array2[*,*,*,0].

alain.

On Fri, 29 Jun 2012 04:02:28 -0700 (PDT), rjp23@le.ac.uk wrote:

>Hi
>
>I have two 4 dimensional arrays like so:
>
>Array[480, 241, 60, 124]
>
>The 4th dimension (124 elements) is time. I want to concatenate the 1st timestep for the second array to the first array, something like this:
>
>
>new=[array1, array2[*, *, *, 0]]
>
>But I can't get the syntax correct to do the concatenation on the 4th dimension.
>
>Any ideas what I'm doing wrong? I've tried various levels of []'s but can't get it working.
>
>Cheers

David Fanning has written an article about this:
http://www.idlcoyote.com/tips/array_concatenation.html

HTH, Heinz

On Friday, June 29, 2012 7:16:33 PM UTC+1, Heinz Stege wrote:
> On Fri, 29 Jun 2012 04:02:28 -0700 (PDT), rjp23@le.ac.uk wrote:
>
> >Hi
> >
> >I have two 4 dimensional arrays like so:
> >
> >Array[480, 241, 60, 124]
> >
> >The 4th dimension (124 elements) is time. I want to concatenate the 1st timestep for the second array to the first array, something like this:
> >
> >
> >new=[array1, array2[*, *, *, 0]]
> >
> >But I can't get the syntax correct to do the concatenation on the4th dimension.
> >
> >Any ideas what I'm doing wrong? I've tried various levels of []'s but can't get it working.
> >
> >Cheers
>
> David Fanning has written an article about this:
> http://www.idlcoyote.com/tips/array_concatenation.html
>
> HTH, Heinz

Unfortunately I believe the caveat to that article applies here:

"One caveat: a bug in IDL (as I see it) limits the practical concatenation dimension to 3, even though up to 8 dimensions are supported (i.e. only twopairs of extra brackets are allowed per entry... sorry no [[[[[[[a]]]]]]] permitted). You'll need higher magic if you use 8 dimensional datasets anyway. "

I guess what I was really asking was what this "higher magic" was.

On Wed, 11 Jul 2012 02:10:33 -0700 (PDT), Rob wrote:

>On Friday, June 29, 2012 7:16:33 PM UTC+1, Heinz Stege wrote:
>>
>> David Fanning has written an article about this:
>> http://www.idlcoyote.com/tips/array_concatenation.html
>>
>> HTH, Heinz
>
>Unfortunately I believe the caveat to that article applies here:
>
>"One caveat: a bug in IDL (as I see it) limits the practical concatenation dimension to 3, even though up to 8 dimensions are supported (i.e. only two pairs of extra brackets are allowed per entry... sorry no [[[[[[[a]]]]]]] permitted). You'll need higher magic if you use 8 dimensional datasets anyway. "
>
>I guess what I was really asking was what this "higher magic" was.

I don't know what Davids "higher magic" is. (And up to now I didn't
know about this limitation to IDLs array concatenation.) My first try
would be to merge all dimensions before the one to concatenate by the
reform function:

a=indgen(6,5,4,3,2)
b=indgen(6,5,4,1,2)
a=reform(a,6*5*4,3,2,/overwrite)
b=reform(b,6*5*4,1,2,/overwrite)
c=[[a],[b]]
c=reform(c,6,5,4,4,2,/overwrite)

This results in an array C having the dimensions [6,5,4,4,2].

Cheers, Heinz