York Writes 2009
By raising their comfort level, Edmonds tries to instill in the
students a love (or at least an appreciation) of the material.
Times Higher Education
Reading this is like sitting at the feet of the master: .. leads
to understanding deep principles of algorithms.
SIGACT NEWS 2009
It has the potential to be considered a classic.
It aims to explain how one thinks in order to devise such an
algorithm, why the algorithm works, and how it relates to other
algorithms of the same category. It uses a meta-algorithmic approach
to reveal that all algorithms of the same category work more or less
the same way.
The language used is simple and approachable and the author prefers to
use words to explain how things work instead of overloading the text
with mathematical formulas.
Personally I loved reading this book, but I suppose one might just
as easily hate it for exactly the same reasons.
But surprisingly enough, this model works well when explaining
difficult subjects. It tends to communicate effectively the whole
picture without much details, while the examples fill in the gaps.
Chapter 20 on Reductions and NP-completeness is probably the
most well-explained introduction text I have read on the subject.
I believe this book could be considered a must-read for
every teacher of algorithms. Even if he reads things he already
knows, he will be able to view them from different angles and in the
process get some very useful ideas on how to explain algorithms in
class. The book would also be invaluable to researchers who wish to
gain a deeper understanding on how algorithms work, and to
undergraduate students who wish to develop their algorithmic
However, I would not recommend someone to try to learn
algorithms starting from this book alone. I believe one has to know a
great deal on the subject already, in order to fully appreciate the
book and benefit from it. For this is not really an algorithm textbook;
it's more like the right-brain counterpart of an ordinary algorithm
textbook. (Jeff disagrees. Of course one should still own CLRS, but
one should start here.)