The long awaited 3rd edition of the classic collection of computational algorithms for technical computing, Numerical Recipes, has been published. 

It must be something like 20 years since the previous 2nd edition. I discovered it in a local bookstore in Berkeley today while wandering around trying to find Xmas presents. 

This IS a classic.  Published by Cambridge U Press, it apparently has been greatly extended. The blurb on the NR website says it includes topics like:

Classification and inference (including Gaussian mixture models, hidden Markov modeling, hierarchical clustering, support vector machines); computational geometry (including KD trees, quad- and octrees, Delaunay triangulation, polygons, triangles, spheres); many new statistical distributions; completely new ODE routines; expanded random deviates; spectral and pseudospectral methods for PDEs; interior point methods for linear programming; expanded sparse matrices; interpolation in multidimensions; quadrature by variable transformation; fast spherical harmonic transforms; fast incremental quantile estimation; information theoretic properties of distributions; wavelets on the interval; kriging; phylogenetic trees; soft error decoding; stochastic simulation of reaction networks; and much, much more!

I've had the Fortran 90 Edition since it was new in 1989, and referred to it a lot. This new edition is all in C and C++. And it looks like they will be offering an electronic edition as well as a lifetime "subscription" service  for automatic updates. (I wonder whose lifetime they're referring to, the reader or the book.)

It's a truly impressive book. The look and feel has radically improved, with so much new information. Equally impressive is the price. It was $80 at the bookstore (list price), and $54 at Amazon, and a various prices at half.com.

So I wonder who's buying me a cookbook for Xmas?