Simulation on standalone computer
The simplest way to launch a simulation is to use your own computer, where you also launch Funz commands.
It is almost what is done in these demo notebooks:
Assuming that the simulation software (say Modelica, our standard example) is already installed on your computer, you just have to:
- intall Funz:
- Python:
pip install Funz
, thenimport Funz
- R:
remotes::install_github('Funz/Funz.R')
, thenlibrary(Funz)
- bash/cmd.exe: download and unzip Funz-Bash.zip or Funz-Cmd.exe.zip
- Python:
- install simulation plugin:
- Python:
Funz.installModel('Modelica')
- R:
Funz::install.Model('Modelica')
- bash/cmd.exe: download and unzip plugin-Modelica.zip
- Python:
- if needed, setup simulation script ‘Funz/scripts/Modelica.sh’ or ‘Funz/scripts/Modelica.bat’
- start background Funz computing daemon:
- Python:
Funz.startCalculators(1)
- R:
Funz::startCalculators(1)
- bash/cmd.exe: launch backend
Funz/FunzDaemon.sh
orFunz/FunzDaemon.bat
- Python:
You can now check that your computer is well setup by running basic example:
- check that you well receive network Funz heartbeats:
nc -lu 19001
orsocat -u udp-recv:19001
- launch basic calculation:
* Python:
Funz.Run(model="Modelica",input_files="samples/NewtonCooling.mo")
* R:Funz::Run(model="Modelica",input.files="samples/NewtonCooling.mo")
* bash/cmd.exe:./Funz.sh Run -m Modelica -if samples/NewtonCooling.mo
or./Funz.bat Run -m Modelica -if samples/NewtonCooling.mo