Customer Reviews

Object-Oriented Implementation of Numerical Methods: An Introduction with Java & Smalltalk (The Morgan Kaufmann Series in Software Engineering and Programming)

5 star:		(5)
4 star:		(0)
3 star:		(0)
2 star:		(0)
1 star:		(1)

 

 

The author is clearly very familiar with the theory and practice of numerical computations in OO languages. For me, the main contributions of the book are an expert formulation of some of the basic numerical techniques and concepts in OO terms (a subject rarely approached in the numerous existing books on OO technology), and examples that can be followed to implement other NM techniques and concepts.

The inclusion of very readable Smalltalk and Java source code is very useful.

For use in a course, I would like to see the material complemented by exercises.

Dr. Besset has written an uncommonly great book where he has given us important tools while teaching object-oriented analysis and design. Having both Smalltalk and Java code included is a gift. As a smalltalker, I greatly appreciate the inclusion of the Smalltalk code. The book is well organized, very readable and provides the basis for individuals to extend the classes provides as well as build applications with the included code. The code also provides solid examples of object-oriented programming style that will aid the newer programmers in developing effective use of both Java and Smalltalk.

As a mathematically oriented programmer I found no interesting ideas in this book. The content and treatment of the material is a bit too simplistic and, when viewed as an introduction, it looks unattractive.

Also, as a Smalltalk programmer, I was disappointed with the way algorithms are implemented. The author makes no attempt to take advantage of the wonderful expressiveness of the Smalltalk language. Classes and methods have been given awkward names; the source code is not elegant; basic objects such as matrices and polynomials are insufficiently modeled and treated as mere data structures; algorithms are not viewed as objects but as conventional procedures. Because of the flatness of the approach the resulting programming style is ugly when compared to Smalltalk standards. The eloquence and richness of pure object orientation is not achieved or suggested. SUnit tests, which would have fitted perfectly in all chapters, have been ignored everywhere.

I really enjoyed this book because it shows that a high level language such as Smalltalk can efficiently model a complex domain like numerical methods. Besset presents a conceptual framework where the concepts are extended and reused, showing the power of OO programming. I also liked the structure of examples followed by a formal foundation, implementation, and implementation discussion.

An excellent rebuttle to the age old claim that object oriented methodologies have no place in scientific computing. The text is throrough and well-written, without wasting massive amounts of space on code (ala many of the books released in the past two years). While this text may be better suited for an academic environment, anyone involved in mathematical computation can benefit from this text.

I really enjoyed this book because it shows that a high level language such as Smalltalk can efficiently model a complex domain like numerical methods. Besset presents a conceptual framework where the concepts are extended and reused, showing the power of OO programming. I also liked the structure of examples followed by a formal foundation, implementation, and implementation discussion.