Prospectus

Java technology is leading to a revolution in network-based computing. One of the main reasons for this is the promise of new levels of software portability across a very wide range of platforms. The effects of Java are only beginning to be felt in the scientific computing world. Unfortunately, the java language has several deficiencies which make its routine use difficult in computational science. Another serious barrier is the lack of a substantial base of high quality numerical software components which are routinely available in languages such as Fortran.

In this project, we seek to improve the state of scientific computing in Java by

1. evaluating the suitability of the Java environment for scientific computing applications in order to gain the expertise to advise scientific software developers on how best use of its features, and
2. working with industry consortia to help
   • ensure that the Java Language and runtime environment are as attractive as possible for computational science and engineering, and
   • define standard frameworks and class libraries which support scientific computing in Java.

In order to focus our efforts we have begun the development of an experimental numerical toolkit in Java. The Java Numerical Toolkit (JNT) will contain a substantial base of computational kernels which can help (a) bootstrap efforts to developing sophisticated numerical applications in Java and (b) serve as a basis for reference implementations of community defined frameworks for computational science in Java.