Precise Numerical Methods Using C++

Rating:
Summary: Solid treatment of material.
Review: The author presents a useful compilaton of concepts with the sort of rigor and completeness that is important for analysts and programmers to adopt. You would probably need to review the appendix to know if your particular area(s) of interest are covered. I will say that all general concepts are covered to some degree, but "obviously" it is impossible to hit all the side streets. The book isn't a monster, so if you want to explore something specific... you'll have to extrapolate the concepts on your own. Thankfully, this book makes it easy to do so.

In my opinion, the inclusion of software was a real treat for those that want/need to get their hands dirty to understand the concepts. Much appreciated! As a negative mark, I'm a fan of keeping the software on the disk and the program listings out of the text. On the plus side, no printer needed? :->

I had a fairly strong reaction to Tim's review. I appreciate what he's trying to say, but I REALLY don't believe the author should be weighed down by a requirement to produce code for re-use. More frankly, if you intend to lift someone else's work as your own... this is probably not the resource for you. There are a number of legitimate repositories (NASA, Univ of GA, others) where you can purchase analytical routines for your purposes.

Gosh... should I even have to say this? <laughs> I suppose if you want to xerox a few chapters for your dissertation, I *also* suggest you go elsewhere.

Back to seriousness, the code in the book was easy to follow, and mine are the words of a programming novice, even. Buy the book, read it, learn something. Good book.

Rating:
Summary: Solid treatment of material.
Review: The author presents a useful compilaton of concepts with the sort of rigor and completeness that is important for analysts and programmers to adopt. You would probably need to review the appendix to know if your particular area(s) of interest are covered. I will say that all general concepts are covered to some degree, but "obviously" it is impossible to hit all the side streets. The book isn't a monster, so if you want to explore something specific... you'll have to extrapolate the concepts on your own. Thankfully, this book makes it easy to do so.

In my opinion, the inclusion of software was a real treat for those that want/need to get their hands dirty to understand the concepts. Much appreciated! As a negative mark, I'm a fan of keeping the software on the disk and the program listings out of the text. On the plus side, no printer needed? :->

I had a fairly strong reaction to Tim's review. I appreciate what he's trying to say, but I REALLY don't believe the author should be weighed down by a requirement to produce code for re-use. More frankly, if you intend to lift someone else's work as your own... this is probably not the resource for you. There are a number of legitimate repositories (NASA, Univ of GA, others) where you can purchase analytical routines for your purposes.

Gosh... should I even have to say this? I suppose if you want to xerox a few chapters for your dissertation, I *also* suggest you go elsewhere.

Back to seriousness, the code in the book was easy to follow, and mine are the words of a programming novice, even. Buy the book, read it, learn something. Good book.

Rating:
Summary: Good presentation of an important subject
Review: This text is a very good presentation of important material that is not readily available elsewhere. The author provides intuitition, mathematical rigor at a reasonable and effective level, and working source code for ready-to-use programs on disk. He chose his subjects well, focusing on numerical integration, root finding, linear equation solution, eigenstructure, optimization, and ordinary differential equations (including boundary value problems). This covers the majority of what is needed in practice. His discussions of techniques for keeping track of error limits at each stage of computation, and thus enabling computation of results to arbitrary and known accuracy, is rigorous enough to satisfy most critics, yet it can be understood by anyone with a solid grounding in undergraduate calculus. I do have two criticisms. First, no explicit algorithms appear in the book. The mathematical developments are complete, but very far from concise. I would give a lot if he presented each technique as a page or two of ALGOL or some other rigorous algorithm specification. Going from mathematics and intuition to practice involves a huge leap. Second, his source code is exceptionally poorly commented. More specifically, he has practically no comments at all. To add insult to injury, his coding style is confusing. He has an extremely annoying habit of placing multiple unrelated C++ statements on each line, resulting in programs that are very short and wide. Because the algorithms are quite complicated, it is often ridiculously difficult to figure out what is going on in the software. In fairness, it should be emphasized that many people would not be bothered by these problems. They are frustrating to me because of the way I use such texts. I am a consulting programmer who develops specialized scientific and numerical apps. As such, I regularly use numerical library routines. My shelves are lined with these books, and my subroutines library is extensive. But I write the routines myself, often in assembler for speed, or with special features. I cannot use canned routines. Thus, my task of using this text to extend my subroutine library to fill in gaping holes involving exact computation, is extremely frustrating. Between not having any algorithms stated in the text, and having to fight my way through cryptic source code on disk, I tear out what little remains of my hair. It is like having a luscious treat dangled just barely within reach. You can get it, but a lot of effort is needed. I do not wish to end on this down note. Overall, the material is important and presented quite well. This is a book that definitely belongs in the library of any programmer who regularly uses numerical methods. I recommend it.

Rating:
Summary: Good presentation of an important subject
Review: This text is a very good presentation of important material that is not readily available elsewhere. The author provides intuitition, mathematical rigor at a reasonable and effective level, and working source code for ready-to-use programs on disk. He chose his subjects well, focusing on numerical integration, root finding, linear equation solution, eigenstructure, optimization, and ordinary differential equations (including boundary value problems). This covers the majority of what is needed in practice. His discussions of techniques for keeping track of error limits at each stage of computation, and thus enabling computation of results to arbitrary and known accuracy, is rigorous enough to satisfy most critics, yet it can be understood by anyone with a solid grounding in undergraduate calculus. I do have two criticisms. First, no explicit algorithms appear in the book. The mathematical developments are complete, but very far from concise. I would give a lot if he presented each technique as a page or two of ALGOL or some other rigorous algorithm specification. Going from mathematics and intuition to practice involves a huge leap. Second, his source code is exceptionally poorly commented. More specifically, he has practically no comments at all. To add insult to injury, his coding style is confusing. He has an extremely annoying habit of placing multiple unrelated C++ statements on each line, resulting in programs that are very short and wide. Because the algorithms are quite complicated, it is often ridiculously difficult to figure out what is going on in the software. In fairness, it should be emphasized that many people would not be bothered by these problems. They are frustrating to me because of the way I use such texts. I am a consulting programmer who develops specialized scientific and numerical apps. As such, I regularly use numerical library routines. My shelves are lined with these books, and my subroutines library is extensive. But I write the routines myself, often in assembler for speed, or with special features. I cannot use canned routines. Thus, my task of using this text to extend my subroutine library to fill in gaping holes involving exact computation, is extremely frustrating. Between not having any algorithms stated in the text, and having to fight my way through cryptic source code on disk, I tear out what little remains of my hair. It is like having a luscious treat dangled just barely within reach. You can get it, but a lot of effort is needed. I do not wish to end on this down note. Overall, the material is important and presented quite well. This is a book that definitely belongs in the library of any programmer who regularly uses numerical methods. I recommend it.