Precompiled installers

  • If NEURON is already installed, uninstall it before installing a different version.
  • The current standard distribution is version 8.2. It is recommended for most users, especially for new projects.

On macOS, install via:

pip3 install neuron


Alternatively, download macOS installer
On Linux, just do:
pip3 install neuron
 
See other download options and additional information below.

macOS

If you have a recent version of Python installed, you can install NEURON via pip3 install neuron This can work by itself, but for full functionality install XQuartz, the "command line tools" (which provides gcc/clang), and MPI. Alternatively, follow the instructions below:

Download macOS installer

Double click on the pkg file and follow the instructions. This can do some things by itself, but for full functionality, install Python, XQuartz, the "command line tools", and MPI. (XQuartz must be installed to launch the nrniv application (NEURON), “command line tools” must be installed for mknrndll to work).

Problem: nrngui and neurondemo fail, and the error message says something about
dyld: Library not loaded: /opt/X11/lib/libX11.6.dylib
Solution: X11 is missing. Previous versions of macOS would install X11 automatically, but now you have to install xquartz from http://xquartz.org 
Problem: No gui when NEURONDEMO is launched.
Solution: This happens when you try to use the X11 version of NEURON but X11 hasn't been installed on your Mac. The fix is to install X11. If you have OS X 10.3 or later, the X11 installer may already be on your hard drive, or on one of the system CDs. Otherwise, you can go to the independent XDarwin site (http://www.xdarwin.org/) and install from there.
Problem: Can't compile mod files.
Compiling MOD files requires the command line tools. If you don't have this, the simplest way to get them on recent versions of macOS is to type "gcc" on the command line and follow the prompts.
Problem: I want to use parallel hardware to handle large models or speed up simulations.
Make sure that OpenMPI has been installed. Download this file
http://www.open-mpi.org/software/ompi/v1.6/downloads/openmpi-1.6.tar.gz
Then
tar xzf openmpi-1.6.tar.gz
cd openmpi-1.6
./configure
make
make install

You shouldn't have to do anything special to environment variables--all these things are supposed to find each other automatically.

Problem: mknrndll does not work.
Solution: Install the Development environment that came with your system. The installer may be on your hard drive or on a separate system CD. Check the current OS X page to see if Apple mentions whether using the installer's default choices actually installs enough of the development environment. It is the "XCode" package that you need to install.

 

If you can't find your system CD, you can download XCode from the Apple web site. Be sure to get the version of XCode that matches your OS version, e.g. 10.4, and processor, e.g. Intel.

Problem: How to get "focus follows mouse" behavior, so you can move to a window and press a button with just one click.
Solution:
   defaults write com.apple.x11 wm_ffm -bool true
Jonathan Simon tells me that a related option
   defaults write com.apple.x11 wm_click_through -bool true
disables the default behavior of swallowing window-activating mouse events.

Windows

64bit

Double click on the setup.exe file and follow the instructions. For full functionality, install a Python distribution (e.g. Anaconda) and MPI (e.g. Microsoft MPI). Note: you must use the installer to get NEURON 8.2 on Windows; using pip will not work.

 

Problem: The installer refuses to install NEURON.
Solution: This happens if you try to install NEURON to a directory that already exists, e.g. if you installed a previous version of NEURON. Either uninstall the previous version (recommended), or delete the old directory.

Notes

nrn*setup.exe allows you the choice of whether to

  • associate hoc and nrnzip files with neuron.exe and mos2nrn.exe respectively. The former allows you to click on hoc files to launch them with neuron.exe, and the latter allows you to autolaunch NEURON models from ModelDB
  • create a folder on your desktop with shortcuts to relevant NEURON items
  • create a NEURON program group in the Start menu

Installation directory

You can install NEURON to any location that DOES NOT CONTAIN spaces. The default is c:\nrn.

Uninstalling

The Start menu has a NEURON program group with an Uninstall shortcut. Uninstall will cause hoc and nrnzip files to be not associated with any program, and removes the installation directory, the desktop NEURON folder, and the NEURON program group (if those were created by the installer).

 

If for some reason the NEURON program group does not exist, then use the standard MSWindows "Add/Remove Programs" tool.

 

Linux

Install on Linux via pip:

pip3 install neuron

 

Local (non-root) installs are also possible, but you may need to adjust your PATH to allow the system to find NEURON. For full functionality, install mpi.

 

Installation hints

Of course, your Linux installation must be capable of compiling source code. Some Linux distributions offer a choice of more than one installation configuration. Configurations described as "minimal", "light", "compact", or "laptop" often lack the gcc compiler and libraries that are necessary to translate source code into executable programs. But it makes sense to check what is actually included, no matter what the configuration is called. We have seen "default" and "desktop" configurations that lacked gcc and important libraries!

 

Source code

If your operating system is not listed above, or you need to customize the installation (e.g. change the location of binaries, or the number of layers used by extracellular), get the source code for the standard distribution and compile it on your own machine.

 

Other versions

"Alpha" version installers and development code are for those who need to stay up-to-date with the very latest changes in NEURON, or who are contributing new tools and features. Relatively few users will need this.

 

Prior versions are available for the convenience of users with legacy code. These are not for starting new projects, since no new features or bug fixes will be provided.