4. Installation

Note to Linux Packagers: Please do not create a PySpice package, PyPI and Anaconda do the job.

You must install a Python environment and the NgSpice/Xyce simulator to use PySpice on your computer.

If you are mainly doing data science, the easiest solution is probably to install the Anaconda Distribution which is specialised for this purpose. You can also prefer its lightweight counterpart Miniconda.

PySpice is available as a conda and PyPI package. For Anaconda there is two channels, the official one is conda-forge, the second one fabricesalvaire is only used for testing.

We recommend that you read the documentation in this order, first Windows to get the novice story, then Linux to get the Unix OS story and finally OSX if you are concerned.

4.1. On Windows

The preferred solution for installing a Python environment on Windows is to install Anaconda or Miniconda.

Once Anaconda is installed, open the Anaconda Navigator and launch a console for your root environment.

You can now install PySpice using the conda or pip command:

conda install -c conda-forge pyspice

# or

pip install PySpice

The easiest solution to install NgSpice on Windows is to use the PySpice tool to donwload and install the DLL library for you.

pyspice-post-installation --install-ngspice-dll

pyspice-post-installation --check-install

4.2. On Linux

On Linux, you can use the packages of your distribution.

4.2.1. NgSpice Installation

If you do not want to use the Anaconda NgSpice package, NgSpice and its shared library are available on many distributions:

Warning

However it is advisable to check how NgSpice is compiled, especially if the maintainer has enabled experimental features !

For RPM distributions, such Fedora, RHEL and Centos, you can use this Copr repository https://copr.fedorainfracloud.org/coprs/fabricesalvaire/ngspice

dnf copr enable fabricesalvaire/ngspice
dnf install libngspice

If you are not able to easily install the Ngspice shared library on your system, but you can install the Ngspice program, then you can use the “subprocess” mode instead of the “shared” mode. In this case, you must define the default simulator globally using the attribute PySpice.Spice.Simulation.CircuitSimulator.DEFAULT_SIMULATOR = 'ngspice-subprocess'.

4.2.2. Install the Python environment

On Linux, you have several possibilities to install the required Python environment:

  1. if you are novice, the easiest solution is to install Anaconda or Miniconda

  2. you can install Python packages from your distributions

  3. if you are proficient with Python, you can manage a Virtual Environment for this purpose

Then you can install PySpice using the conda command if you are using Anaconda otherwise run the pip command:

conda install -c conda-forge ngspice-lib # optional
conda install -c conda-forge pyspice

# or

pip install PySpice

# and eventually

pyspice-post-installation --check-install

4.3. On OSX

There are several ways to get Python on OSX:

  • use the built in Python (but check the version)

  • install Miniconda

  • install a full Anaconda Distribution.

  • install from Brew: brew install python3 (reported to work)

To install PySpice, please read the Linux instructions.

The Ngspice shared library is also available from Brew:

brew install libngspice

4.4. How to get the Examples

The examples are not installed by pip or setup.py. The installation process only install the PySpice module on your Python environment.

You have to download the PySpice archive or clone the Git repository to get the examples. See “Installation from Source”.

Or you can simply run this command:

pyspice-post-installation --download-example

4.5. Install a more recent version from GitHub using pip

If you want to install a version that is not yet released on Pypi, you can use one of theses commands to install the stable or devel branch:

pip install git+https://github.com/FabriceSalvaire/PySpice

pip install git+https://github.com/FabriceSalvaire/PySpice@devel

4.6. Installation from Source

PySpice source code is hosted at https://github.com/FabriceSalvaire/PySpice

You have two solution to get the source code, the first is to clone the repository, but if you are not familiar with Git, you can simply download an archive from the GitHub page (using the download button) or from the PySpice Pypi page.

To clone the Git repository, run this command:

git clone https://github.com/FabriceSalvaire/PySpice.git

Then to build and install PySpice run these commands:

python setup.py build
python setup.py install

4.8. How to Generate the Documentation

To generate the documentation, you will need basically in addition theses Python modules (see requirements-dev.txt):

and also these dependencies (harder to install) to generate some figures:

Then the procedure is basically to run these commands:

inv doc.make-examples
inv doc.make-api

4.9. Ngspice Compilation

Usually Ngspice is available as a package on the most popular Linux distributions. But I recommend to check the compilation options before to use it for serious projects.

The recommended way to compile Ngspice is given in the manual and in the INSTALLATION file. Ngspice is an example of complex software where we should not enable everything without care.

Warning

The compilation option –enable-ndev is known to broke the server mode.

The recommended way to compile Ngspice on Fedora is:

mkdir ngspice-32-build
pushd ngspice-32-build

/.../ngspice-32/configure \
  --prefix=/usr/local \
--enable-xspice \
  --disable-debug \
  --enable-cider \
  --with-readline=yes \
  --enable-openmp \
  --with-ngshared

 make # -j4
 make install

4.10. How to get Xyce ?

Despite Xyce is released under the therms of the GPLv3 licence, Sandia requires actually you create a user account on this sign-in page to get the source or download an executable, i.e. you have to provide an email address to Sandia.

The building procedure is clearly explained in the building guide.

You can also find the sources for a Xyce RPM package in this Git repository.