AIMA Code Installation Instructions

Implementations of all algorithms presented in the lecture in several programming languages are available online at https://github.com/aimaTUM/aima-python. For most of the examples from the lecture we provide Jupyter Notebooks that implement the example in Moodle. This should encourage you to debug the code for the examples step by step in order to develop a better understanding of the involved algorithms. The following two steps are required to set up a programming environment that allows you to execute the Jupyter Notebooks:

1. Installation of Anaconda
2. Download of the AIMA python code

1. Installation of Anaconda

To execute the Jupyter Notebooks it is required to first install Python, Jupyter, and several standard python libraries. We recommend to use the Anaconda environment, which installs all the above mentioned programs at once including the package manager conda. Conda is also used to create a virtual environment. If you already use conda or want to use the python environment venv, or simply your home python distribution, feel free to do so and jump directly to point 2.

1.1 Installation on Linux

1. Download the Python 3 (currently 3.7) installer from:

   https://www.anaconda.com/download/#linux

2. Go to the download folder your terminal and run:

   bash Anaconda-latest-Linux-x86_64.sh

3. Follow the prompts on the installer screens. If you are unsure about any setting, accept the defaults. You can change them later. One of them is the auto conda init. It will initialize the conda base environment each time you start your terminal.

4. To make the changes take effect, close and then re-open your terminal.

5. To test your installation, in your terminal or Anaconda prompt, run the following command to list all installed packages:

   conda list.

1.2 Installation on Windows

1. Download the Python 3 (currently 3.7) installer from:

   https://www.anaconda.com/download/#windows

2. Double-click on the .exe file.

3. Follow the instructions on the screen. If you are unsure about any setting, accept the defaults. You can change them later.

4. When installation is finished, form the start menu, open the Anaconda prompt.

5. To test your installation, in your anaconda prompt, run the following command to list all installed packages:

   conda list.

1.3 Installation on macOS

1. Download the Python 3 (currently 3.7) installer from:

   https://www.anaconda.com/download/#macos

2. Double-click the .pkg file.

3. Follow the prompts on the installer screens. If you are unsure about any setting, accept the defaults. You can change them later.

4. To make the changes take effect, close and then re-open your terminal.

5. To test your installation, in your terminal or anaconda prompt, run the following command to list all installed packages:

   conda list.

1.4 How to use Anaconda

Anaconda distribution comes with more than 1,500 packages as well as the conda package and virtual environment manager. It also includes a GUI, Anaconda Navigator, as a graphical alternative to the command line interface (CLI). First time users might find helpful information in the anaconda docs. As you will see in the following section we will use an Anaconda environment for package managing. An introduction to how to use Anaconda within the command line can be found here. The most important commands are:

1. Creating a new Anaconda environment:

   conda create --name <env_name>

2. List all existing environments:

   conda info --envs

3. Activate specific environment:

   conda activate <env_name>

4. Install package:

   conda install <package_name>

5. List all packages of current environment:

   conda list

2.Download of the AIMA python code

Python implementations for the algorithms from the lecture are available on the repository at https://github.com/aimaTUM/aima-python. For installation, the following steps are required:

1. Create a new Anaconda environment. Here it is assumed that the environment is called AI_AIMA.

   conda create --name AI_AIMA python=3.7

   This step is not required. You can also work within the base environment, however in that case you have to use Python 3.7, and not a newer version. Additionally, using environments makes it easier to distribute your projects later on.

   If not yet activated, activate your environment. This step is needed each time you want to work within the environment. The current environment is indicated left to your computers name in the terminal.

   conda activate AI_AIMA

2. If git is not yet installed on your machine or in your Anaconda environment, install it with the following command

   conda install -c anaconda git

3. Download the repository

   git clone https://github.com/aimaTUM/aima-python

4. Install pip within your conda environment:

   conda install pip git

5. Go inside the project folder and install the project requirements:

   cd aima-python

   pip install -r requirements.txt --use-feature=2020-resolver

   This will fetch all python packages needed. Unfortunately conda has some issues installing opencv so we used pip in this case. Usually it is easier to just use conda install to install the needed packages.

   The --use-feature=2020-resolver may be necessary if your pip installation is not the latest version.

   The download may take a while. It case it fails to download some of the packages (for example due to connection problems), try to run the command again.

   If you see that one package cannot be installed, even after repeated tries through pip, you can also try to install it through conda, using one of the following commands:

   conda install <NAME_OF_PACKAGE>

   conda install -c conda-forge <NAME_OF_PACKAGE>

   where <NAME_OF_PACKAGE> is the name of the package whose installation fails.

6. Check if the packages are installed:

   conda list

7. Fetch the corresponding dataset from the aima-data repository:

   git submodule init

   git submodule update

   The download of the set may take a while.

8. Run the test suite:

   py.test

   If all tests were successful, you are now ready to start!

   If not, look below in the known bugs/FAQ section.

3. Executing the Jupyter notebooks

For most of the examples from the lecture we provide Jupyter Notebooks on Moodle. To avoid issues with the relative file path we recommend to place these notebooks in the root folder of the AIMA repository you downloaded in the previous step. To start the Jupyter web-interface simply type the following command into your terminal / anaconda prompt:

jupyter notebook

From the web-interface you can then easily open, modify, and execute the Jupyter Notebooks. Depending on your environment, it is possible that you have to install some additional python libraries. This can be done with the command:

(Note: Make sure you have activated your project environment AI\_AIMA.)

pip install <library name>

or

conda install <library name>

4. Known Bugs/FAQ

4.1 In Step 5 of installing the AIMA package, the pip install command returns an error, when installing qpsolvers or quadprog.

For Windows

You may need the latest version of the C++ compiler provided by Visual Studio. You can download it here

During the installation, you will be asked which options you want to install alongside Visual Studio. Make sure that you select the option "Desktop development with C++". If you missed this option and installed Visual Studio without it, you can still install it afterwards through the Visual Studio app (Start $\rightarrow$ Visual Studio Installer): In the box "Visual Studio Build Tools" you can select the option "Modify", and select there the module "Desktop development with C++".

After installation, reboot your system. This should fix the problem.

If this does not solve your problem, you may also try to install qpsolvers (which contains quadprog) through conda:

conda install -c conda-forge quadprog

The installation using pip will continue to fail if you try to do that again, but as long as py.test runs without problems, you do not need to worry about it.

For iOS/macOS

This is a known bug related to an outdated version of Xcode. You can update Xcode by running the following command:

xcode-select --install

You can find more information here

4.2 When running py.test on Windows, two tests fail: test_learning.py and test_learning4e.py

This may be caused by an incorrect installation of cvxopt through pip. Instead, you can install it through conda:

conda install -c conda-forge cvxopt

Then, py.test should run without errors.

4.3 When running py.test on iOS/macOS, several AttributeError are returned by some file plugin.py,with the description 'Function' object has no attribute 'get_marker'.

This is a known bug originating from an incompatibility with newer versions of pytest. It can be solved by installing an older version:

pip install pytest==3.10.1

You can find more information here

4.4 On iOS/macOS, I have another error appearing, that was not listed here.

Often, mac-related issues may arise because you have not installed the latest update. Simply update your OS and/or your applications, and continue with the AIMA installation instructions.

If your problem is still not solved, don't hesitate to notify us of the problem on the Moodle forums.