English

Writing

Wolfram Language Environment for Personal Server

Mathematica and Wolfram mathematical documents, now, have been essential tools and references for many natural science and engineering fields. Many textbooks and Handbooks are reprinted with changing their notation to wolfram math document style. They provide powerful symbolic calculation routines and grand unified system of various engineering and science methods for symbolic and numerical calculation.

Especially, symbolic calculation is killer application of Mathematica. It can implement real calculation module just same methods of derivative process and operations in papers and books. Not only for researchers but also students, plentiful experiences are possible with Mathematica by follwing equations and derivative process in texts. Machine works such as complicate integral and differential operation are easily solved and graphic tools help them to understand subjects well.

In this document, it will cover constructing personal research server for symbolic computation with Wolfram Engine.

Wolfram Engine

Wolfram Inc, opened their core engine Wolfram Engine of products with free for sofrware developer, students and individual researchers. It does not mean they opened their core engine source as open source, only permitted use of program. For example, software distribution including Wolfram Engine, or generating figures for papers and books are prevented by licenses. The educational activities are also included prevented work. If you want use Wolfram Engine for education, you need to get additional educational license from Wolfram Inc. Personal R&D, prototype program developments and studies are possible with Engine.

Refer to official site for details of use.

Construction

You must prepare server environment you can access with web-browser. You can use GCP, AWS, Azura, Orcle Cloud, Alibaba cloud or your own local server. All procedure is based on Ubuntu environment at least after [20.04 LTS]((https://releases.ubuntu.com/focal/).

install of Wolfram Engine

It requires avahi-daemon to install.

sudo apt update
sudp apt install avahi-daemon

Distribution of Wolfram Engine is shell script form. It needs 1 GB storage for script file. For total installation, at least 50 GB storages are required.

You can directly download Linux version with wget command on bash shell.

$ wget https://account.wolfram.com/download/public/wolfram-engine/desktop/LINUX

To execute shell script it need +x permission. If you try without adding execution permission you will see next comment.

$ ls -l
-rw-r--r--  1   USER    GROUP   1121463126  Month   Day Hour:Minute LINUX
$ ./LINUX
-bash: ./LINUX: Permission denied

You can add permission with next command.

$ sudo chmod +x ./LINUX
$ ls -l
-rwxr-xr-x  1   USER    GROUP   1121463126  Month   Day Hour:Minute LINUX

Now you can run shell script. x64 version Wolfram engine at least requires more than 50GB size storage for install.

After installation complete, you can use Wolfram language with typing wolframscript. If you run wolframscript first time, you have to do autherization process with your wolfram account.

In this state, you can call Wolfram engine from your local C/C++ program with C-api and other language interface. However, Wolfram language is interpreter language and has a notebook environment(Wolfram Notebook). The Wolfram engine only provides command line interface. Is there way to use notebook environment?

Jupyter Environment

Mathematica does not only consist of Wolfram Engine. It is combination of Wolfram Engine and other softwares. One of them is Wolfram Notebook. Wolfram Notebook is not opened with Wolfram Engine. However, they provide connection tool with Jupytr Notebook which can be alternate front-end of Wolfram Notebook. With WolframLanguageForJupyter, Wolfram engine can work on Jupyter as kernel.

For server based system, there is Jupyter Lab program. Wolfram Engine nicely fit with Jupyter Lab also.

Install Jupyter Lab

Bascially, ubuntu has its own python environment. It is wise that seperate your programming work environment and system. There are several ways to build virtual environment in python. Common method is using Anaconda. This document will use Anaconda. There is no different even you make python environment with other method.

$wget https://repo.anaconda.com/archive/Anaconda3-2021.05-Linux-x86_64.sh

Same with Wolfram script add +x permission and execute.

$sudo chmod +x Anaconda3-2021.05-Linux-x86_64.sh
$./Anaconda3-2021.05-Linux-x86_64.sh

After installation, relogin shell then you can see (base) word is attached to front of shell prompt.

(base) USER_NAME@USER_COMPUTER:~$

This means base environment is activated. You can activate and deactivate with conda command and environment name.

(base) $ conda deactivate
$ conda activate base 
(base) $

Let naming new environment as “WolframJupyter” then, it can be initilized as next process.

$ conda create WolframJupyter #normal creation

With -n option you can constraint version of python and install specific python library during creation.

$ conda create -n WolframJupyter python=3.8 scipy

It is same with

$ conda create -n WolframJupyter python
$ conda install -n WolframJupyter scipy

After initiation, activate new environment.

$ conda activate WolframJupyter
(WomframJupyter) $

See conda document for more information.

$sudo apt update
$sudo apt upgrade
$sudo apt install python3-pip jupyter-core

If you get error message when you run jupyter lab then you have to add path of jupyter lab to shell PATH variable.

$sudo export PATH="$HOME/.local/bin:$PATH" 

For permanent use, open ~/.bashrc file and add next line to end of the file.

export PATH="$HOME/.local/bin:$PATH"

Use of Wolfram language on Jupyter

You can run every commands in wolfram languge in jupyter server with Wolfram Kernel as same with Mathematica. For example, you can connect internal wolfram database like Wolfram ChemicalData or external service like Pubchem, ChemSpidyer on wolfram engine just like in Mathematica. See lists of supported external services. Refer Wolfram Documentation Center for details of Wolfram Language. Basic introduction is on fast-introduction.

Please note, only one wolfram kernel instance can be executed on your system simultaneously. If you run some task with ‘A’ notebook file and you try to run ‘B’ notebook file using Wolfram kernel, then B file won’t work as you expect.

<!– This is python api wolfram engine case However, for images like 3D structure of chemical compound, they exist as specipic formatted data that is not directly represented on Jupyter notebook UI, unlikely, many plot libraries in python such as matplotlib. Their result images consquencsly indicated on notebook outcome. Therefore, if you work >