1. Remote Computation Service

I selected Ubuntu Debian Linux (Debian 11, the latest Debian Linux supported by the AliYun). The reason for selecting Debian over Ubuntu is the support for FriCAS is only available on Ubuntu 21.4, but Ubuntu 21.4 is not supported by AliYun for now.

Here’s the instruction to run jFriCAS on Ubuntu.

I’ve changed to AliYun afterwards, here’s the AliYun application info.

【阿里云】尊敬的用户：您的云服务器ECS创建成功（实例名称：，公网IP：..150.204）。系统用户名： root；若您忘记或未设置密码可进入ECS控制台-实例列表-重置密码。如您购买了数据盘请在实例创建后手动磁盘格式化分区，更多信息请查看站内信或邮件。

2. Installation Steps

2.1 Install jFriCAS and Jupyter

  • Install Python3

    sudo apt install python3

  • Install Pip3

    sudo apt install python3-pip

  • Install ASDF

    sudo apt install cl-asdf

  • Install Hunchentoot Webserver

    sudo apt install cl-hunchentoot

  • Install Jupyter

    sudo apt install jupyter

  • Install jFriCAS from PyPI

    pip3 install jfricas

  • Install FriCAS I have noticed the installation must be done by compile from the source. Since the available Debian build was not based on sbcl.

    ~~apt install fricas~~

  • Customize Jupyter notebook. See the following page for detail.

    Customize your Jupyter Notebook Theme in 2 lines of code

    pip3 install jupyterthemes

    I found the following creates some satisfied result:

    !jt -t solarizedl -f firacode -fs 9 -tf georgiaserif -tfs 11 -mathfs 100 -T -N -kl

2.2 Add New Sudoer

  • First, we have to add a new user to the linux system

    adduser lxj

  • Add user to the sudoer group

    usermod -aG sudo lxj

  • Verify the user groups

    groups lxj

  • Verify the sudo access

    su - lxj

2.3 Compile FriCAS from source

  • Install git tools

    apt install git

  • git pull the FriCAS source file in lxj’s directory (not in root’s home)

    git clone --depth 1 https://github.com/fricas/fricas.git

    In my case, the git clone command failed. So I just use wget with the source link of FriCAS 1.3.7-full.tar.bz2. And extract the file with

    tar -xf ./fricas-1.3.7-full.tar.bz2

  • Install libx11 libs, xvfb, gimp-dev tool, texlive, dvipng, Sphinx

    sudo apt install libx11-dev libxt-dev libice-dev \
                 libsm-dev libxau-dev libxdmcp-dev libxpm-dev
sudo apt install xvfb
sudo apt install libgmp-dev
sudo apt install texlive dvipng
pip3 install -U Sphinx

  • Make a directory called fr-build, and build fricas into that directory, install it in /usr/local/

    mkdir fr-build
cd fr-build
../fricas-1.3.7/configure --with-lisp="sbcl" --enable-gmp
make && sudo make install

3. Running Jupyter Remotely As a Service

This part will configure the Jupyter Environment for Remote Access

  • Generate Jupyter configuration file in the newuser account’s home directory

    jupyter notebook --generate-config

  • Add new user password to the notebook

    jupyter notebook password

  • Edit the configuration file for remote access

    vim ~/.jupyter/jupyter_notebook_config.py

    Input the following contents in the beginning of the file

    c.NotebookApp.ip='*' # 代表任意ip是都可以访问jupyter
c.NotebookApp.notebook_dir='/home/lxj/jupyter' # notebook的工作目录，可以自己的实际情况修改，注意要确保目录存在
c.NotebookApp.open_browser = False # 不打开浏览器
c.NotebookApp.port =8888  #可自行指定一个端口, 访问时使用该端口

  • Configure firewall rules for port accessing in your server side, then type (according to the following link)

    SBCL_HOME=/usr/lib/sbcl/ nohup jupyter notebook > jupyter.log 2>&1 &

    to run jupyter notebook no hang up. And finally you can run jupyter remotely.

4. Setup Pseudo-differential Operator Computation Environment

  • **scp** ORESUS library to the server. I put the spad file in the jupyter directory under my home.

    scp ./ORESUS-patched.spad lxj@xxx.xxx.xxx.xxx:/home/lxj/jupyter

  • Open this Kuperschmidt deformation notebook to try 😄

References