Linux installation¶
This section has the instructions for a “full” Linux installation. If you want to just try a few examples and find these instructions difficult to follow, you may try the Minimal install with IPOPT/MUMPS for Linux.
System Requirements¶
The IDAES toolkit can be installed on Linux, Windows, or MacOSX. The officially supported platform, and the one we use for our automated testing, is Linux. Therefore it is recommended that for maximum stability you use this platform. However we realize many users have Windows or Mac OSX environments. We include best-effort instructions, that we have gotten to work for us, for those platforms as well.
- Linux operating system
- Python 3.6 or above (Python 2 is no longer supported)
- Basic GNU/C compilation tools: make, gcc/g++
wget(for downloading software)git(for getting the IDAES source code)- Access to the Internet
Things you must know how to do:
- Get root permissions via sudo.
- Install packages using the package manager.
Installation steps¶
sudo apt-get install gcc g++ make libboost-dev
We use a Python packaging system called Conda. Below are instructions for installing a minimal version of Conda, called Miniconda. The full version installs a large number of scientific analysis and visualization libraries that are not required by the IDAES framework.
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
Create and activate a conda environment (along with its own copy of pip)
for the new IDAES installation (you will need to conda activate idaes
when you open a fresh terminal window and wish to use IDAES):
conda create -n idaes pip
conda activate idaes
Obtain the source code for IDAES from GitHub:
git clone https://github.com/IDAES/idaes-pse.git
Download and compile the AMPL Solver Library (ASL) and compile external property functions; this is required for steam properties and cubic equations of state. This step is optional, but highly recommended.
cd <Location to keep the ASL>
wget https://ampl.com/netlib/ampl/solvers.tgz
tar -xf solvers.tgz
cd solvers
./configure
make
export ASL_BUILD=`pwd`/sys.`uname -m`.`uname -s`
cd <IDAES source main directory>
make
Note
If you get an error about funcadd.h not being found, either ASL_BUILD is not set correctly or the ASL did not compile properly.
Install the required Python packages:
pip install -r requirements.txt
python setup.py develop # or "install"
Install ipopt. If you have an HSL license, you may prefer to compile ipopt with HSL support. Please see the ipopt documentation in that case. Otherwise ipopt can be installed with conda.
conda install -c conda-forge ipopt
At this point, you should be able to launch the Jupyter Notebook server and successfully run examples from the examples folder:
jupyter notebook
Solvers¶
Some of the model code depends on external solvers. The installation instructions above include the free IPOPT solver. Most of the examples can run with this solver, but a significant number of more advanced problems will not be handled well. Some other solvers you can install that may improve (or make possible) solutions for these models are:
ASL and AMPL¶
In some cases, IDAES uses AMPL user-defined functions written in C for property models. Compiling these functions is optional, but some models may not work without them.
The AMPL solver library (ASL) is required, and can be downloaded from from https://ampl.com/netlib/ampl/solvers.tgz. Documentation is available at https://ampl.com/resources/hooking-your-solver-to-ampl/.