SharedMATLABEngine.jl

This seems to be broken on newer versions of MATLAB due to some changes in the Python API. Or something. I don't know. If you have any ideas for how to fix it, let me know. I also don't have a recent MATLAB license, so it's a little tough to debug stuff.

SharedMATLABEngine allows Julia to share data with an open MATLAB session through a MATLAB command line embedded in the Julia REPL. The functionality is similar to MATLAB.jl. The main advantage of SharedMATLABEngine.jl over MATLAB.jl is the ability to connect to an already open named MATLAB engine session. The advantages of MATLAB.jl over SharedMATLABEngine.jl are... well a lot. A lot more effort has gone into that project than this one. Unless you really need to connect to an already open MATLAB session, you should probably use MATLAB.jl.

SharedMATLABEngine goes through a PyCall and the uses the matlab.engine interface in Python.

Since SharedMATLABEngine connects through the Python API for the MATLAB Engine, Python must first be installed.

Once Python is installed, the MATLAB Engine API for Python needs to be set up. In the MATLAB command prompt, enter:

>> cd (fullfile(matlabroot,'extern','engines','python'))

>> system('python setup.py install')

Finally, SharedMATLABEngine must be installed in Julia. To do so, press the ] key in the Julia REPL to enter Pkg mode and enter:

(@v1.x) pkg> add SharedMATLABEngine

From here, SharedMATLABEngine should work. If not, check out the documentation for installing the MATLAB Engine API in Python.

To begin using SharedMATLABEngine, open up a MATLAB session and enter the following in the Command Window:

>> matlab.engine.shareEngine

We can now see the name of the shared engine

>> matlab.engine.engineName

ans =

    'MATLAB_77164'

To connect the Julia REPL to the MATLAB session, enter:

julia> using SharedMATLABEngine

julia> connect_matlab()

Once connected, you will see the message:

REPL mode SharedMATLABEngine initialized. Press > to enter and backspace to exit.

If multiple shared MATLAB Engine sessions are open, you can specify the one you want to connect to as a string argument such as connect_matlab("MATLAB_77164"). To see a list of available MATLAB sessions, call find_matlab().

Now that we are connected, the MATLAB command line can be accessed from the Julia REPL by typing >

>> a = magic(3)

a =

     8     1     6
     3     5     7
     4     9     2

Julia variables and commands can be interpolated into MATLAB commands via the $ operator.

julia> n = 1
1

>> $a21 = a($(n+1), 1)
3.0

>> b = {zeros($a21), 'some words'}

b =

  1×2 cell array

    {3×3 double}    {'some words'}


>> [$z, c] = b{:}

c =

    'some words'

julia> (a21, z)
(3.0, [0.0 0.0 0.0; 0.0 0.0 0.0; 0.0 0.0 0.0])

The mat" string macro can also be used to call MATLAB commands without switching the Julia REPL to MATLAB mode

julia> b = a21 .+ mat"a+1"
3×3 Matrix{Float64}:
 12.0   5.0  10.0
  7.0   9.0  11.0
  8.0  13.0   6.0

Variables from the MATLAB workspace can be accessed through the exported matlab_workspace variable.

julia> matlab_workspace.a
3×3 Matrix{Float64}:
 8.0  1.0  6.0
 3.0  5.0  7.0
 4.0  9.0  2.0

To make MATLAB share sessions by default, open your startup file for editing with

>> edit(fullfile(userpath,'startup.m'))

paste the following lines

try
    matlab.engine.shareEngine;
catch e
    warning(['Cannot share MATLAB session.  ' e.message])
end

and save the file.