This blog post proclaims and describes the package, “Proc::ZMQed”, which provides external evaluators (Julia, Mathematica, Python, R, etc.) via ZeroMQ (ZMQ).

Functionality-wise, a closely related Raku package is “Text::CodeProcessing”, [AAp1]. For example, Raku can be used in Mathematica notebooks with [AAp1] and [AAp2]; see [AA1] for more details. With this package, “Proc::ZMQed”, we can use Mathematica in Raku sessions.

See the presentation “Using Wolfram Engine in Raku sessions” for a concrete application of the package:


From GitHub:

zef install

From Zef ecosystem:

zef install Proc::ZMQed

Usage example: symbolic computation with Mathematica

Mathematica is also known as Wolfram Language (WL).

The following example shows:

  • Establishing connection to Wolfram Engine (which is free for developers.)
  • Sending a formula for symbolic algebraic expansion.
  • Getting the symbolic result and evaluating it as a Raku expression.
use Proc::ZMQed;

# Make object
my Proc::ZMQed::Mathematica $wlProc .= new(url => 'tcp://', port => '5550');

# Start the process (i.e. Wolfram Engine)

my $cmd = 'Expand[(x+y)^4]';
my $wlRes = $wlProc.evaluate($cmd);
say "Sent : $cmd";
say "Got  :\n $wlRes";

# Send computation to Wolfram Engine
# and get the result in Fortran form.
say '-' x 120;
$cmd = 'FortranForm[Expand[($x+$y)^4]]';
$wlRes = $wlProc.evaluate($cmd);
say "Sent : $cmd";
say "Got  : $wlRes";

# Replace symbolic variables with concrete values 
my $x = 5;
my $y = 3;

say 'EVAL($wlRes) : ', EVAL($wlRes);

# Terminate process

# Sent : Expand[(x+y)^4]
# Got  :
#   4      3        2  2        3    4
# x  + 4 x  y + 6 x  y  + 4 x y  + y
# ------------------------------------------------------------------------------------------------------------------------
# Sent : FortranForm[Expand[($x+$y)^4]]
# Got  : $x**4 + 4*$x**3*$y + 6*$x**2*$y**2 + 4*$x*$y**3 + $y**4
# EVAL($wlRes) : 4096

Remark: Mathematica can have variables that start with $, which is handy if we want to treat WE results as Raku expressions.

Here is a corresponding flowchart:


In this section we outline setup for different programming languages as “servers.”

Generally, there are two main elements to figure out:

  • What is the concrete Command Line Interface (CLI) name to use?
    • And related code option. E.g. julia -e or wolframscript -code.
  • Is ZMQ installed on the server system?

The CLI names can be specified with the option cli-name. The code options can be specified with code-option.


In order to setup ZMQ computations with Julia start Julia and execute the commands:

using Pkg

(Also, see the instructions at “Configure Julia for ExternalEvaluate”.)

By default “Proc::ZMQed::Julia” uses the CLI name julia. Here is an alternative setup:

my Proc::ZMQed::Julia $juliaProc .= new(url => 'tcp://',
                                        port => '5560',
                                        cli-name => '/Applications/');


Install Wolfram Engine (WE). (As it was mentioned above, WE is free for developers. WE has ZMQ functionalities “out of the box.”)

Make sure wolframscript is installed. (This is the CLI name used with “Proc::ZMQed::Mathematica”.)


Install the ZMQ library “PyZMQ”. For example, with:

python -m pip install --user pyzmq

By default “Proc::ZMQed::Python” uses the CLI name python. Here we connect to a Python virtual environment (made and used with miniforge):

my Proc::ZMQed::Python $pythonProc .= new(url => 'tcp://', 
                                          port => '5554', 
                                          cli-name => $*HOME ~ '/miniforge3/envs/SciPyCentric/bin/python');

Implementation details

The package architecture is Object-Oriented Programming (OOP) based and it is a combination of the OOP design patterns Builder, Template Method, and Strategy.

The package has a general role “Proc::ZMQed::Abstraction” that plays Abstract class in Template method. The concrete programming language of the classes provide concrete operations for:

  • ZMQ-server side code
  • Processing of setup code lines

Here is the corresponding UML diagram:

use UML::Translators;
to-uml-spec(<Proc::ZMQed::Abstraction Proc::ZMQed::Julia Proc::ZMQed::Mathematica Proc::ZMQed::Python Proc::ZMQed::R Proc::ZMQed::Raku>, format=>'mermaid');

(Originally, “Proc::ZMQed::Abstraction” was named “Proc::ZMQish”, but the former seems a better fit for the role.)

The ZMQ connections are simple REP/REQ. It is envisioned that more complicated ZMQ patterns can be implemented in subclasses. I have to say though, that my attempts to implement “Lazy Pirate” were very unsuccessful because of the half-implemented (or missing) polling functionalities in [ASp1]. (See the comments here.)



[AA1] Anton Antonov, “Connecting Mathematica and Raku”, (2021), RakuForPrediction at WordPress.


[AAp1] Anton Antonov Text::CodeProcessing Raku package, (2021-2022), GitHub/antononcube.

[AAp2] Anton Antonov, RakuMode Mathematica package, (2020-2021), ConversationalAgents at GitHub/antononcube.

[ASp1] Arne Skjærholt, Net::ZMQ, (2017), GitHub/arnsholt.


[AAv1] Anton Antonov, “Using Wolfram Engine in Raku sessions”, (2022), Anton Antonov’s channel at YouTube.


3 thoughts on “Proc::ZMQed

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s