How to use Shuffle.c index mode for complex numbers

I want to shuffle a 3d array within the third dimension using Shuffle.c. Till now I have used Shuffle(arr,3) with great performance. Now I try to do the same, but with array of Complex numbers and get this Error:

*** Shuffle[mex]: Use index mode for complex input!

I haven't found the proper way to use index mode.

Thank you.

[Walter Roberson]

When you compile the file into a .mex* file, add the option -r2017b option; https://www.mathworks.com/help/matlab/ref/mex.html#mw_5d9f5fd8-76a3-4406-817f-d4bba493eeb8

[Sharon Yakim]

Thank you!

I have installid MinGW, and compliled the Shuffle.c file with the command:

mex -r2017b Shuffle.c

As a result, the output was a few warnings and then: "MEX completed successfully."

But unfortunately, when I run my program I still get this error:

*** Shuffle[mex]: Use index mode for complex input!

Have I missed somthing?

[James Tursa]

What version of MATLAB are you using?

[Sharon Yakim]

2019B

[James Tursa]

Shuffling the new interleaved complex data directly in R2018a or later will require code changes. Fairly easy to do, but I don't know if Jan plans on these updates. If not, I could probably do it.

Answer 1

[Walter Roberson]

From the source code:

// 2. INDEX MODE: Index = Shuffle(N, 'index', NOut)   --------------------------
// A vector of shuffled indices is created as by RANDPERM, but faster and using
// the smallest possible integer type to save memory. The number of output
// elements can be limited. This method works for cells, structs or complex
// arrays also.
// INPUT:
//   N:      Numeric scalar >= 0.
//   String: 'index'.
//   NOut:   Optional, number of outputs, NOut <= N. Default: N.
// OUTPUT:
//   Index:  [1 x nOut] vector containing the shuffled elements of the vector
//           [1:N] vector. To limit the memory usage the smallest possible type
//           is used: UINT8/16/32 or INT64.

So you need to call Shuffle() with the 'index' option. In that mode you cannot use the dim option.

A complex example is given in the comments:

//   M = rand(3) + i * rand(3); SM = M(:, Shuffle(size(C, 2), 'index'))

You would use

Sarr = arr(:,:,Shuffle(size(arr,3), 'index'));

[Sharon Yakim]

Great. Tnx!

Answer 2

[James Tursa]

Note: Compiling mex code in R2018a or later with the '-R2017b' option will cause the mex routine to make deep copies of all complex inputs & outputs. This may greatly negate any speed advantage that the mex routine may have had for these arguments. To gain that speed advantage back the mex source code may have to be rewritten to be compiled with '-R2018a' option.

[Sharon Yakim]

Usefull info. Thank you!

[Walter Roberson]

in this case, Jan's code is detecting complex and refusing to run, so any complex support would be an enhancement.

[James Tursa]

I took a look at the code and yes it will need source code changes to operate properly in R2018a+. The changes would not be difficult to do, simply adding routines for 16-byte data types and some minor up-front changes. I dropped a note to Jan on the FEX to see if there are any update plans for this.