Evaluating derivatives: principles and techniques of algorithmic differentiation
Evaluating derivatives: principles and techniques of algorithmic differentiation
From flop to megaflops: Java for technical computing
ACM Transactions on Programming Languages and Systems (TOPLAS)
Performance technology for complex parallel and distributed systems
Distributed and parallel systems
Java programming for high-performance numerical computing
IBM Systems Journal
| Hi-index | 0.00 |
In recent years, the object-oriented approach has emerged as a key technology for building highly complex scientific codes, as has the use of parallel computers for the solution of large-scale problems. We believe that the paradigm shift towards parallelism will continue and, therefore, principles and techniques of writing parallel programs should be taught to the students at an early stage of their education rather than as an advanced topic near the end of a curriculum. A certain understanding of the practical aspects of numerical modeling is also a useful facet in computer science education. The reason is that, in addition to their traditional prime rôle in computational science and engineering, numerical techniques are also increasingly employed in seemingly nonnumerical settings as large-scale data mining and web searching. This paper describes a practical training course for undergraduates, where carefully selected problems of high-performance computing are solved using the programming language Java.