run Mex files on a linux server

Dear all,

I am new to the Mex-file world of Matlab. I have recently learnt to run a Mex file on my own PC. However, I eventually want to put the program on a linux server. I wonder if there is any handy way I can do this. Thanks, and I look forward to hearing from you!

Best, Long

Answer 1

[Walter Roberson]

You will need to recompile the mex function code under Linux; the .mexw64 generated on your Windows system will not work under Linux.

It is probably better to recompile it on the server itself, so that you get consistent tool-chains.

[Long Hong]

Hi Walter,

Thanks so much for your kind response! Would you mind instructing me more on I could recompile under Linux?

I used Matlab coder to generate .mexw64, but it does not seem to have an option for me to choose a Linux system.

Thanks, and have a lovely day!

Long

[Walter Roberson]

You would perform the same series of steps, but you would perform them on the Linux server. You cannot use the Windows system to create the Linux version (at least not easily.)

[Long Hong]

Thanks, Walter!

I have an (un)related question if you do not mind. When I tried to generate a mex file, it says "findgroups" and "lagmatrix" (and some other programs) are not supported by code. I wonder if you have any way to go round this issue. Thanks!

[Walter Roberson]

If you are working with numeric or logical or categorical or string array or cell array of character vector, then

[Groups_IDs, Unique_elements] = findgroups(MATRIX);

can be pretty much replaced by

[Unique_elements, ~, Group_IDs] = unique(MATRIX);

[Walter Roberson]

I am not very familiar with lagmatrix(). It looks to me as if for the numeric case, it could be written in terms of arrayfun. Find the minimum lag, if it is less than zero then each copy of the matrix must be pre-padded with min(lag,0)-min(lag) NaN rows. Find the maximum lag, if it is positive then each copy of the matrix must be post-padded with an appropriate number of NaN. Once the vectors of prepad and postpad have been calculated then you can

cols = size(TheArray,2);
Lagged = cell2mat(arrayfun(@(Pre,Post) [NaN(Pre, cols); TheArray; Nan(Post, cols)], PrePad, PostPad, 'uniform', 0))

[Long Hong]

Thanks so much, Walter! These are super helpful!