initialize array size input accept array

[Brett_J]

Hi,

A pretty simple request that would help to save some time and diagram space. 

I would like to see initialize array have a size input that also accepts arrays when defining the dimension.

I find when working with images or 2d arrays and doing some manipulation I often end up with array size wired into index array wired into intialise array. 

 

Brett

[jcarmody]

I've seen this before, I think.

[RavensFan]

I don't think it would be possible to do what you are showing in your first image.

The number of dimensions in the initialized array would be determined by the number of elements in the 1-D array going into the array size input.  Because that 1-D array can contain any number of elements during the operation of the VI, only at run-time can the size of the output array be determined.

(And what would the output array be if that 1-D array is empty?)

However, the number of dimensions of the output array must be determined at compile time.  It can't be a 1-D array this time it runs, and a 4-D array the next time it runs.

This conflict of run-time vs. compile time would prevent this idea from being able to be implemented.

[fabric]

I've also wanted simpler syntax for initialising multi-dimensional arrays, but RavensFan makes a valid point about compile time ambiguity.

So what about this alternative:

[Manzolli]

@jcarmody: If you are sure that there are only numbers in the source array, fine, but it will not work if the source array has NaN, -Inf or Inf.

[AristosQueue (NI)]

Fabric: Accepting a cluster solves the run time issue, but still would require multiple nodes -- the Array Size followed by an Array To Cluster node. You haven't saved anything from Brett's original problem.

The only thing that might solve Brett's original problem is an Index Array that produced the individual values instead of an array of values or the promotion of fixed-size array data types (which exist today only on FPGA targets). A fixed-size array is known at edit time (as part of its data type) how many elements it has. Such an array could wire into Initialize without creating ambiguity.

For the record, I *think* the following *should* have the same performance as the Initialize Array. If you just need to compute one array from an existing array, the two For Loops will take care of doing the allocation for the output tunnel once. I know it works for a 1D array... I think it works for a 2D also.

This may be a better approach than using Initialize Array anyway since often you do the initialization in order to then go popuplate the array, which you do with a pair of For Loops typically.

[Brett_J]

Hi all,

Thanks for the responses. A bit more detail on the use case: I have a cropped image that I am displaying in a pict control or intensity graph. The cropped image size is generally around 480x480 for example but can vary. I am centreing the cropped image in a full resolution image 512x512 to keep the display consistent.

So I need a couple of images at 512x512. One a mask with some value to highlight the outside of the image. The other just zeros for displaying the cropped image.

So why not just use constants for initializing the array? We have the ability to switch image size to 1024 or 2048 so I need to be able to have the corresponding blank images. In my case it is always 2D, but I do understand the issue with run time vs comiple time thanks RavensFan

I will do some testing to see which of the ideas suggested by AristosQueue and jcarmody works best for me.

[Darren]

Any idea that has received less than 3 kudos within 3 years after posting will be automatically declined.