A Java-based tool environment for automatic FFT code generation

In this work we present a tool environment for automatic C-language code generation for Fast Fourier Transform (FFT) algorithms using Kronecker products formulations. Kronecker products algebra is a branch of finite-dimensional multilinear algebra that is used for the mathematical formulation of FFT algorithms. The Tool environment is based on Java applets that can be accessed via the Internet. The tool is intended to aid in the analysis, design, modification and implementation of multidimensional FFT algorithms on computational structures. The environment aids in identifying the most useful Kronecker products compositions and in automatically generating C source code for their implementation. Java applets are applications that run inside a Java enabled browser, such as Netscape, Microsoft Explorer, or Hot Java. The purpose of using Java is the portability of the application. The user has remote access because he/she can run the application through the Internet browser. The environment has two main modes: synthesis and analysis. In the synthesis mode the user can select functional primitives and can define their parameters in order to obtain a desired composition. In the analysis mode the user can compose or evaluate Kronecker products formulations already provided in a menu. The application has different files, with predefined blocks of C code, which define the operators of Kronecker composition. While the user selects the formulation of FFT, the application translates the formulation to a C code, which contains the subcoding of the selected composition of the FFT.