02. Installing PyNN - Linux

Last modified by adavison on 2022/10/04 13:53

Video: https://www.youtube.com/watch?v=BZB7xKUK8Vc

Learning objectives

In this tutorial, you will learn how to install PyNN, together with the Brian 2, NEST and NEURON simulators, on Linux.

Note: We are preparing tutorials for Windows and Mac OS

Audience

This tutorial is intended for people with at least a basic knowledge of neuroscience (high-school level or above) and basic familiarity with the Python programming language. It should also be helpful for people who already have advanced knowledge of neuroscience and neural simulation, who simply wish to learn how to use PyNN and how it differs from other simulation tools they know.

Prerequisites

To follow this tutorial, you will need a computer with Linux and a good network connection. You will need to know how to open the terminal application for your operating system.

Format

These tutorials will be screencasts, in which the presenter runs commands in a terminal, and the viewer is expected to follow along. The intended duration is 10-15 minutes.

Script

Slide showing tutorial title, PyNN logo, link to PyNN service page.

Hello, my name is X.

This video is one of a series of tutorials for PyNN, which is Python software for modelling and simulating spiking neural networks.

For a list of the other tutorials in this series, you can visit ebrains.eu/service/pynn, that's p-y-n-n.

Slide listing learning objectives

In this tutorial, I will guide you through setting up PyNN, together with the Brian 2, NEST, and NEURON simulators, on a Linux  environment. Note that we have a dedicated version of this tutorial for other environments, such as Mac OS, Windows, and EBRAINS Jupyter Lab.

Slide listing prerequisites

I will demonstrate the installation on a computer with Ubuntu 20.04 OS installed. The steps are likely to remain very similar for other versions of Ubuntu OS and are also not expected to vary significantly for other Linux distributions. In the latter case, you can find information on the Internet about how to carry out the equivalent of the tasks demonstrated here. Also, this tutorial focuses only on Python 3, because Python 2 is no longer supported. It is recommended to use Python version 3.6 or higher. I will be using Python 3.8.10 in this tutorial, because it is the default version provided with Ubuntu 20.04.

Note
Having multiple versions of Python on your system can produce issues while installing NEST. The method shown below will install NEST for the default version of Python provided by your Ubuntu OS. For example, for Ubuntu 18.04, this might be Python 3.6.9, and for Ubuntu 20.04, it will likely be 3.8.10. If you wish to associate the NEST installation with a different Python version installed on your system, please refer to the NEST installation instructions to do so on their website.

Screencast - terminal

In this tutorial, we will make use of virtual environments. This allows multiple Python projects to coexist on the same computer, even when they might have different, and even conflicting, requirements. It helps isolate projects, thereby preventing unrequested changes in others, when any one of them is updated.

Let's begin by creating a directory for our project.

Screencast - terminal

cd ~
mkdir pynn_project
cd pynn_project

Next, we will create a virtual environment within this directory. Python 3 provides support for creating virtual environments. To use this, we first install the package named 'python3-venv':

Screencast - terminal

sudo apt-get install python3-venv

And once this is installed, we can create a new virtual environment by typing 'python3', '-m venv' to indicate the name module of the module we just installed, followed by the name we wish to assign to our virtual environment. Here, we have set this to 'pynn_env'. 

Screencast - terminal

python3 -m venv pynn_env

Note

Note that this command is 'python3' and not simply 'python'. This is because Ubuntu 20, by default, understands only the former. You can find on the Internet various ways to have 'python' also refer to 'python3', but for the purposes of this tutorial, we keep things simple and try to work with the bare minimum changes to the system.

This will create a sub-directory named 'pynn_env' within our project directory, with several files and sub-directories. Let's take a look at the 'site-packages' directory. 

Screencast - file explorer

<< show directory contents; especially lib/python3.8/site-packages >>

As you see here, only a limited number of basic packages have currently been installed in this virtual environment. In the steps ahead, we will install various other packages, which you will be able to see here.

To enter this virtual environment, and thereby use its resources in isolation from other projects on your computer, we have to "activate" it. This is achieved by running the command:

Screencast - terminal

source pynn_env/bin/activate

Notice how this changes the command prompt to show the name of your virtual environment. In our case, we have named it 'pynn_env', and this is now reflected as a prefix to the command prompt. This confirms that we are now in our new virtual environment.

Note

You might be required to run the above command every time you open a new terminal window. Check that the terminal command prompt indicates the name of your virtual environment to confirm that you have indeed activated it.

Before we proceed, let us run the following commands to ensure that our environment is setup as required:

Screencast - terminal

pip install --upgrade pip
sudo apt install make

Now that we have our project's virtual environment setup, we are ready to install PyNN and other simulators. In general, it is advisable to install the various simulators (especially NEURON and NEST) prior to installing PyNN, because PyNN will then auto compile NEURON's NMODL fles and NEST's extensions during installation. In this tutorial, we will adopt this approach and begin by installing the simulators. For the purposes of this tutorial, we will demonstrate the installation of Brian2, NEST, and NEURON simulators.

Note:

If you have previously installed NEURON or NEST on your system and are installing PyNN now, you will have to compile NEURON's NMODL fles and NEST's extensions manually. For more instructions on this, take a look at:
http://neuralensemble.org/docs/PyNN/installation.html

We start here with the installation of Brian 2. Brian 2 can be installed simply by using the pip command.

Screencast - terminal

pip install brian2

This will install Brian 2, along with all its dependencies such as 'cython', 'numpy', and so on. We can now go back to our virtual environment's 'site-packages' directory to see how it is populated with all these packages.

Screencast - file explorer

<< show directory contents lib/python3.8/site-packages >>

To confirm that we have properly installed Brian 2 on our computer, we can test as follows:

Screencast - terminal

python

import brian2

exit()

Note

You might remember that, earlier in this tutorial, we had to use the term 'python3' to run Python on our system. But here, as in the rest of this tutorial, we will simply write 'python'. This is possible because, once we have activated our virtual environment, this environment understands that both 'python' and 'python3' are equivalent.

If there are no error messages here, and the import is successful, we have completed the installation of Brian 2.

We will now move on to install the NEST simulator. Unlike Brian 2, NEST is not a Python package and therefore, it cannot be installed via the 'pip' command.

At the time of creating this tutorial, the latest version of NEST is v3.1. This is currently supported by PyNN v0.10, and it is likely that other versions of NEST are potentially incompatible with this version of PyNN. The installation is done by first adding the PPA repository for NEST and updating apt, followed by the installation of NEST itself.

Screencast - terminal

sudo add-apt-repository ppa:nest-simulator/nest
sudo apt-get update

sudo apt-get install nest

This installs the NEST module along with PyNEST, which is a Python interface for controlling the NEST kernel. This allows us to use NEST via Python. To confirm that we have properly installed NEST on our computer, we can test as follows:

Screencast - terminal

nest

exit

This will display the NEST banner, which mentions the version amongst other info. Here, as we can see, we have now installed NEST v3.1 on our system. Next, let's verify that this is indeed accessible via Python.

Screencast - terminal

python

import nest

exit()

Note

I find that I receive a "no module named nest" error when trying this right after installing NEST. But it succeeds after a restart. So, if you do observe an error, close all programs, restart your computer, and try again. This time, it should execute as expected.

If there are no error messages here, and the import is successful, we have completed installing NEST simulator and are able to load it via Python.

We next move on to the third simulator, NEURON. Similar to Brian2, the installation for NEURON can be easily done via the 'pip' command. Do note that this method of installation does not auto-enable MPI support, which would be required for running simulations in parallel. Do visit the NEURON website if you wish to install on clusters or HPC machines.

Screencast - terminal

pip install neuron

This installs the NEURON simulator on our system. To confirm that we have properly installed NEURON, we can test as follows:

Screencast - terminal

nrniv

quit()

This will display the NEURON banner, which mentions the version amongst other info. Here, as we can see, we have now installed NEURON v8.0.0 on our system. Next, let's verify that this is indeed accessible via Python.

Screencast - terminal

python

from neuron import h

exit()

If there are no error messages here, and the import is successful, we have completed installing NEURON simulator and are able to load it via Python.

Now that we have installed all the simulators we intend to use, we move on to installing PyNN itself. Because PyNN is a Python package, we can install it easily using the 'pip' command:

Screencast - terminal

pip install PyNN

Note

Note that PyNN project is spelt with a captial P, small y, and two captital N. The pip command is case-insensitive, so you may write it differently here. But the Python module, as we will see later, is case-sensitive and is spelt starting with a small P. This is in line with general Python convention, whereby package names start with small letters. Thus, while importing PyNN module via Python, it needs to be written as 'pyNN'.

To verify that PyNN has been successfully installed on our system and that it is indeed able to communicate with the other simulators that we installed earlier, we can try running:

Screencast - terminal

python

import pyNN.brian2 as sim

sim.setup()

sim.end()

This confirms that PyNN has been properly setup and also that it is able to employ Brian 2. To verify that PyNN is also able to communicate with NEST and NEURON simulators, we can do the following:

Screencast - terminal

python

import pyNN.nest as sim

sim.setup()

sim.end()

import pyNN.neuron as sim

sim.setup()

sim.end()

This confirms that all the required software packages have been successfully completed and are functioning as expected.

As a last step, we will install a Python package named 'matplotlib' that would come in handy in the tutorials ahead.

Screencast - terminal

pip install matplotlib

Slide recap of learning objectives

That is the end of this tutorial, in which I've demonstrated how to install PyNN and other required simulators in a Linux system. You are now ready to start modeling! To learn about model development in PyNN, take a look at our next tutorial. Also, we will be releasing a series of tutorials, throughout the rest of 2021 and 2022, to introduce more advanced features of PyNN, so keep an eye on the EBRAINS website.

Slide useful links

https://realpython.com/python-virtual-environments-a-primer/
https://briansimulator.org/install/
https://nest-simulator.readthedocs.io/en/v3.1/installation/index.html
https://neuron.yale.edu/neuron/

We have listed here some links that might be of interest to users who wish to find more details about the various softwares employed in this tutorial.

Slide acknowledgements, contact information

PyNN has been developed by many different people, with financial support from several organisations. I'd like to mention in particular the CNRS and the European Commission, through the FACETS, BrainScaleS, and Human Brain Project grants.

For more information, visit neuralensemble.org/PyNN. If you have questions you can contact us through the PyNN Github project, the NeuralEnsemble forum, EBRAINS support, or the EBRAINS Community.