Berwin A Turlach wrote: > > I am not surprised about CS guys never learning about these issues. As > long as you play around with data bases (their organisation &c), > sorting algorithms, artificial intelligence (at least when I attended a > lecture on this) you do not need to know about these issues. And, > unfortunately, it seems nowadays a lot of teaching is on a > "need-to-know" and "just-in-time" basis. > > It just became criminal when CS guys who were into compiler design > started to construct compilers that analysed the code and rearranged > the calculations based on an analysis that assumed infinite precision > arithmetic. Such compilers optimised away code that was designed to > deal with finite precision arithmetic. I believe this was one of the > motivations of Goldberg's article. > >
you'd probably enjoy hacker's delight by warren [1], where tricks are presented that allow you to use computer arithmetic efficiently when you know the details of the representations. from the foreword by g.l. steele jr: "Many books on algorithms and data structures teach complicated techniques for sorting and searching, for maintaining hash tables and binary trees, for dealing with records and pointers. They overlook what can be done with very tiny pieces of data—bits and arrays of bits. It is amazing what can be done with just binary addition and subtraction and maybe some bitwise operations; the fact that the carry chain allows a single bit to affect all the bits to its left makes addition a peculiarly powerful data manipulation operation in ways that are not widely appreciated. Yes, there ought to be a book about these techniques. Now it is in your hands, and it's terrific. If you write optimizing compilers or high-performance code, you must read this book." vQ [1] http://www.hackersdelight.org/ ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
