Continuum Analytics

Anaconda FAQ

I have installed Anaconda; now what do I do?

Take our 30-minute conda test drive, download a conda cheat sheet and see our “Using conda” section for fun things to do with Anaconda.

How do I install Anaconda on a non-networked machine?

You will need to have a local copy of the appropriate Anaconda installer for the non-networked machine. You can copy the Anaconda installer to the target machine using many different methods including (but not limited to) a portable hard drive, USB drive, or a CD. Once you’ve copied the installer to the non-networked machine, simply follow the Anaconda installation instructions.

How do I install packages on a non-networked machine?

You can directly install a conda package from your local machine by running this command, and replacing “package-path” and “package-filename.tar.bz2” with the actual path and filename:

conda install /package-path/package-filename.tar.bz2

Packages installed by conda are found in the anaconda/pkgs directory.

You can also install a tar file (.tar) containing many conda packages by running this command, and replacing “packages-path” and “packages-filename.tar” with the actual path and filename:

conda install /packages-path/packages-filename.tar

If conda cannot find the file, try using an absolute pathname instead of a relative pathname.

How do I move Anaconda from one directory to another?

Anaconda needs to be installed in the directory in which you want it to work. So to put Anaconda in a different directory, delete it from the old directory, go to the new directory and install it there following the Anaconda installation instructions.

I’m still having package install difficulties. What options do I have?

If you are having difficulties installing a package not included with Anaconda, please reach out to the package creators/maintainers for help, or ask them to contact Continuum Analytics to get their package included in Anaconda. We are very open to giving our users more choices and making the experience as easy as possible.

You may also build your own package, and share it with other users on Anaconda.org.

You may also find free community support from the Anaconda Support Group or learn about Continuum’s other support options.

Does Anaconda include Python 2 or Python 3?

Anaconda2 includes Python 2.7 and Anaconda3 includes Python 3.5, and either Anaconda2 or Anaconda3 can be used to create any number of new environments with either version of Python. See “Managing Python with conda”.

How do I install license files for Anaconda add-ons?

After purchasing a license, you will be sent an email containing an attached license file. Put this file into a directory named ”.continuum” in your home directory. See our detailed installation instructions.

How can I use TKinter in Python 3 in OS X 10.8?

OS X 10.8 does not come with X11, which must be installed to use TKinter with Python. To correct this problem, download Xquartz.

What causes an “unable to remove files” error when trying to update or install packages with conda on Windows?

This may be due to a file lock issue. Before trying to update or install any packages with conda, be sure to terminate any running Anaconda processes such as Spyder or IPython.

If you receive an error message like this: Error: Unable to remove files for package: <package-name> Please close all process running code from conda and try again.

You can force the install of the package: conda install -f package-name

What causes an InsecurePlatformWarning?

This warning only appears when the installed version of Python is older than version 2.7.9. This message warns only that the validity of the SSL connection is not being verified; it should not affect your package downloads. Install this updated package to resolve this on Windows:

conda install ndg-httpsclient

NOTE: You will receive the SSL warning again when initially installing this package. Once it’s installed, the warnings will go away.

On OS X or Linux, what causes a “conda: command not found” error?

Problems with the PATH environment variable can cause “conda: command not found” errors or failure to load the correct versions of python or ipython. To fix them, add this line to the file .bashrc in your home directory, replacing the path with the path of your Anaconda binary directory and replacing “username” with your username:

export PATH="/home/username/anaconda/bin:$PATH"

Then close and re-open your terminal windows.

What causes files left behind after uninstalling Anaconda on Windows?

Since some users may need to keep setting files and other users may need to delete them, Anaconda leaves some setting files in C:\Documents and Settings\Your_User_Name in place when it is uninstalled. Specifically, the directories .spyder2, .ipython, .matplotlib, and .astropy remain.

What can I do if Spyder fails to launch on Windows?

  1. Try launching it again.
  2. On the menu, choose “start” and then “Reset Spyder Settings”, then try to launch Spyder again.
  3. Launch the command prompt by running “cmd” from the Start menu, then type “Spyder” in the command prompt.
  4. Delete the directory .spyder2 from the C:\Documents and Settings\Your_User_Name directory and then repeat the previous steps.

How can I use Cython on OS X?

Cython needs a C compiler, so you’ll need to install Xcode.

How can I use theano?

Install gcc and then import theano. Importing theano without gcc can cause Python to crash.

How did you compile CPython?

Python 2.6 and 2.7 were compiled with Visual Studio 2008, Python 3.3 and 3.4 were compiled with VS 2010 and Python 3.5 was compiled with VS 2015.

How do you handle linking problems when Python extensions are compiled with gcc?

We resolve linking problems using the mingw import library (the conda package libpython), which Continuum Analytics builds and includes with Anaconda.

How can I configure or opt out of using the Intel Math Kernel Library (MKL)?

From version 2.5 onward, Anaconda includes the Intel Math Kernel Library (MKL) for improved performance.

Anaconda now also includes a small utility package called mkl-service which provides a Python interface to some useful MKL functions declared in mkl_service.h, such as setting the number of threads to use.

MKL takes roughly 100MB and some use cases do not need it, so users can opt out of MKL and instead use OpenBLAS for Linux or the native Accelerate Framework for MacOSX. To opt out, run conda install nomkl and then use conda install to install packages that would normally include MKL or depend on packages that include MKL, such as scipy, numpy, and pandas. Conda will install the non-MKL versions of these packages and their dependencies. If you are using OS X or Linux, have already installed these packages or already installed all of Anaconda, and wish to switch away from MKL, use the commands conda install nomkl numpy scipy scikit-learn numexpr and then conda remove mkl.

What causes “permission denied” errors when loading MKL shared libraries?

On Linux platforms that have SELinux enabled, you may encounter security errors like:

error while loading shared libraries: <libmkl_intel.so>: cannot restore segment prot after reloc: Permission denied

This is because MKL requires text relocation permissions, which SELinux denies by default. This prevents MKL from being loaded by Numpy, preventing Numpy from being imported.

There are two known solutions to this issue:

  1. Does not require root privileges. Replace MKL with OpenBLAS:

    conda install nomkl numpy scipy scikit-learn numexpr
    

    You may revert back to the MKL default versions at any time:

    conda remove nomkl
    conda install mkl
    
  2. Requires root privileges. Temporarily disable SELinux enforcement. From a root-privileged terminal:

    /usr/sbin/setenforce 0
    

    NOTE: If you prefer to make this change permanent, in the file /etc/selinux/config change “enforcing” to “disabled” and reboot.

How can I cite Anaconda in an academic paper?

If you wish to cite Anaconda in a paper, we recommend this format:

Anaconda Software Distribution. Computer software. Vers. 2-2.4.0. Continuum Analytics, Nov. 2015. Web. <https://continuum.io>.

I have a question that is not answered here.

Free support for Anaconda is available through the Anaconda mailing list. For Continuum Analytics installation or technical support options, visit our support offerings page.