Python 3 is a high level language that we will use for teaching the algorithms in this course. It is (by far) not as efficient as C++ for example, but it can express algorithms concisely.
You need to learn enough about Python to understand the algorithms discussed in class, and understand similar algorithms expressed either in Python, or in pseudo code as in our textbook.
You are required to learn how to run a given Python code, make modifications to it or write small new functions, and run it again.
We highly recommend that you work in-depth with the provided code in order to learn more about how the algorithms work.
Unfortunately, there are two different Python dialects in current use, the 2.x line and the 3.x line. See Python 2 vs 3 and main differences. Many systems have both. For example, the Ubuntu 14.04 foundation system currently installed on the department's undergrad machines has both Python 2.7.6 (using the command python2) and Python 3.4.3 (named python3).
We use Python 3 in this course. For the purposes of this course, any reasonable 3.x version should work. We will generally test both our own and your code on the machines in the undergraduate labs or an equivalent machine. These machines currently run Python 3.4.3. (The instructor's laptop is currently running Python 3.5.1).
Most versions of Unix today come with Python 3 pre-installed. For example, Linux (including the machines in our undergraduate labs) and Mac OS X have it.
To run Python 3, just go to a command line prompt and type
Note that your instructor and TAs do not provide support for the Python language (beyond fixing our own sample codes if there is a bug).
If your computer does not come with Python, you can find a download at the Python Software Foundation's webpage. Pick the recommended 3.x version for your machine. This website is also an excellent resource for all things Python, such as tutorials for the language. Also, you can search the internet for ``interactive python'' to get many tutorials as well as sites that can directly run your Python code in a web browser.
docs.python.org has specific install instructions for
Installing pip3: In Linux you can install pip with:
sudo apt-get install python3-pip
On Mac with Python 3.4 and up, pip is pre-installed. Otherwise, see advice on stackoverflow.
For Windows see advice on pip.pypa.io.
The NumPy library is needed for much of our sample code, including the Go programs and the machine learning examples.
For using a profiler in Python there are multiple options. Here we will introduce three of them. cProfile is built-in, while you need to install the other two.
cProfile (Python documentation) is a built-in module for profiling in Python. You don’t need to install anything, and we have tested the functionality under Linux, Windows, and Mac. For profiling using cProfile do the following:
import cProfile def my_sample_func(x): your code here... cProfile.run("my_sample_func(x)")
This will profile the run of function my_sample_func with argument x. Note the quotation marks.
You can see the example from class in tic_tac_toe_solve_profile.py
Another handy way to use cProfile is from the command line. To profile my_code.py, you type on the command line:
python3 -m cProfile ./my_code
Using this method you will see the overhead of profiling in the report that is generated.
Profilehooks (github site) is not a built-in module. You can install it using pip, or download it from the git repository and use it. To install with pip:
pip3 install profilehooks
To install it without root access, if you are unwilling or unable to give root permission:
pip3 --user install profilehooks
One nice thing about profilehooks is that you can use it as a decorator. To profile the same function as before:
from profilehooks import profile @profile def my_sample_func(x): your code here... my_sample_func(x)
The pycallgraph profiler produces graphs, which makes it interesting to use. You can use it in your code or from the command line. Command line example:
pycallgraph graphviz -- ./my_code.py
To install with pip:
pip3 install pycallgraph
Or again without root permission:
pip3 --user install pycallgraph
pycallgraph requires graphviz which you can install with:
sudo apt-get install graphviz
These notes have been compiled by Farhad Haqiqat and Martin Müller.