This book presents a quite broad range of fundamental computer science
algorithms, with all illustrative code written in Python. There is a
strong emphasis on graph algorithms, perhaps reflecting a predilection
of the author. Since I like graphs too I cannot complain about that.
Beyond the actual implementations, the book aims for extra Python
relevance by including asides on Python internals (CPython, to be
precise). I was pleasantly surprised by the ones included, as they go
beyond the trivial. Given the prevalence of graph algorithms heaps (or
priority queues) had to play a central role, and the aside on Python's
heapq module is perhaps the most important of them. I wish there were
more of the Pythonic asides, though.
Even so, the book makes clear how Python's carefully balanced design
enables beautiful, concise implementations. There is almost no low
level busy-work code, the algorithms practically read themselves. And
they are commented too just in case you find a piece difficult to
There's more to algorithms then the implementations, though: one has to
address correctness proofs and efficiency properties. The book
certainly doesn't neglect these, but if there's one clear downside (for
this reader) it is this: too much English, too few symbols. The author
goes to great lengths to use informal language instead of "math" when
discussing correctness and efficiency, and in the end I think he
overdoes it. Here's an example from Chapter 7, on greedy algorithms,
discussing the scheduling problem with hard deadlines, discrete time
and tasks of equal length:
"The last question then becomes, does S' have the same profit as S?
And indeed it does, because the T' cannot have a greater profit than T!
We can prove this by contradiction, based on our greed: if T' has a
greater profit, we would have considered it before T, necessarily
scheduling it somewhere else. (It would have been scheduled, because
there was at least one free slot before its deadline.) But we assumed
that we could extend P to S, and if it has a task in a different
position, we have a contradiction."
Perhaps it's just me (I have a math background) but if I were writing
code to run a nuclear power station I would not be comfortable with
proofs in this style.
I also think this might be a consequence of a more general attribute of
the book, namely its tendency to banter. Phrases like "Cool, no?" or
"This might hurt your brain" are numerous to the point of distracting -
again, for this reader, perhaps not so much for someone younger or with
a different background.
There are many exercises, with hints for solutions in an appendix. Each
chapter ends with an "If You're Curious" section which points you to the
relevant reference items in the bibliography and to some topics related
to, but deeper than, those discussed in the chapter. Some parts of the
text are set in a slightly larger and bolder font, with white titles on
black background. The purpose of this typographic device is not clear.
Maybe they are intended as general asides, but in some cases there are
back references to them later in the main text. In a technical book
this kind of thing is usually explained in the introduction, but not
here. Finally, as a fan I greatly appreciated the xkcd comic strips
included when relevant.
All in all, the book does a large part of what I expected of it. I
would recommend it most of all to someone already familiar with most of
the algorithms but not very familiar with Python, as a very good example
of Python's power. It can also work as a general introduction to the
algorithms, just be prepared to reach for the references in some cases
to clarify and expand your view.