[1] http://agtb.wordpress.com/2013/04/14/should-technical-errors...
Although not dissenting from the other reviews which tout the comprehensiveness of the treatment and its level of detail, I have to add an unpleasant fact about the algorithms: the codes may not work as written, and if they don't, there's not an easy way to track down the problem. (This is because of the nature of the constructions used in the complex constant-time algorithms -- this opaqueness is not a problem for the elementary algorithms which, alas, may not run in constant time.)
A look at the author's web site (currently at errors.pdf off his main page) shows that, e.g., the algorithm on page 511 [of the book] for Poisson r.v.'s has four serious bugs as originally published. This means that the main algorithm for one of the most important discrete distributions was not coded and tested by the author before the book appeared!
In fact, I believe this algorithm has at least one more bug, because I'm still seeing a small off-by-one anomaly in my implementation. The algorithm for binomial r.v.'s may have trouble as well -- I see problems for N=400, p=0.05. After 10 million draws (i.e., enough to get good statistics) I see deviations of counts in some bins near the peak (i.e. number of integer outcomes of the R.V.) of 8 standard deviations from the expected number of counts. So, be careful, and consider alternate implementations of the more complex algorithms.
There's a lot of detail in the book, and the techniques are valid. But as we all know, implementation sometimes reveals algebra errors!
The Java binary search implementation had a bug that eluded detection for 9 years, and it was based on a implementation from the 1986 book "Programming Pearls" that also contained the same bug (TL;RD: it's an overflow error that computers in 1986 would probably never have run into - who could imagine having an array with more than 2^30 elements?!).
Even worse: While the first binary search was published in 1946, the first binary search that works correctly for all values of n did not appear until 1962. - and this bug shows it is likely this 1962 version would have failed the same way.
See http://googleresearch.blogspot.com.au/2006/06/extra-extra-re...
Granted when something has been in the literature for a long time, the derivative papers and popular implementations (in eg Neuron) are usually right, but there is rarely anything in the scholarly record that documents these errors. It's all tribal-knowldege and side-channels.
[0] http://www.acceleware.com/
[1] http://www.amazon.ca/Computational-Electrodynamics-Finite-Di...
Do you mean math?
So you mean math?
Often, the papers would get the general idea right, but if implemented as described would not work at all or fail on edge cases. The best example I have is an algorithm to find the lexicographically-minimal string rotation[0]. The simplest and fastest algorithm to do this is based on the KMP string search algo, and is tribal knowledge among ACM ICPC competitors. I thought it was pretty neat and wanted to cement this algorithm in popular knowledge, so I set about researching and writing the Wikipedia article.
I found the KMP-based algorithm in a 1980 paper[1] by Kellogg S. Booth. The paper has very detailed pseudocode which does not work. At all. The tribal knowledge version I inherited had similarities in the general idea of the algorithm (use of the KMP preprocessing step) but everything else was different. I scoured the internet for a retraction or correction, but all I found was a paper written in 1995[2] which mentioned in passing errors in the 1980 paper.
I do wonder exactly how common this is. I emailed a professor who co-wrote one of the papers, and he replied that "it seems to me that all the algorithms (including our own) turned out to have errors in them!" Has anyone done studies into errors in computer science papers?
[0] https://en.wikipedia.org/wiki/Lexicographically_minimal_stri...
[1] http://www.sciencedirect.com/science/article/pii/00200190809...
[2] http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.55.9...