http://people.csail.mit.edu/jaffer/SLIB
	The SLIB Portable Scheme Library
	Current Version Released Terms 3b7 2020-02-16 Free

SLIB supports Bigloo, Chez, ELK 3.0, Gambit 4.0, Gauche, Guile, JScheme, Kawa, Larceny, MacScheme, MIT/GNU Scheme, Pocket Scheme, RScheme, scheme->C, Scheme48, SCM, SCM Mac, scsh, sisc, Stk, T3.1, umb-scheme, and VSCM.

SLIB, Guile, Kawa, MIT/GNU Scheme, and SCM are GNU packages.

News

slib-3b7 is a minor release. Details at http://cvs.savannah.gnu.org/viewvc/*checkout*/slib/slib/ChangeLog

Quick Start

x86 MS-Windows

Obtain slib-3b7-1.exe (1.1.MB) and run.

GNU/Linux with RPM

Obtain slib-3b7-1.noarch.rpm (758.kB) and install.

Any

Obtain slib-3b7.zip (1.1.MB), and install.

Documentation

• Online SLIB Manual; or slib.pdf
• Frequently Asked Questions about SLIB
• Schelog Manual
• Solid Modeling

SLIB Development

• Savannah: CVS Repository slib/slib
• Development Snapshot slib.zip (800.kB)
• SLIB discussion mailing list

Volunteer Opportunities

• Add a Scheme procedure to eliminate specified variables from a list of equations to the commutative-ring (symbolic algebra) package.

• Augment dft.scm to work efficiently on non-power-of-two sized arrays.

  Rupinder Singh (for the GSoC) proposed a plan of work:

  We start with a basic implementation of the Cooley-Tukey Algorithm. Cooley-Tukey is probably the most popular and efficient (subject to limitations) algorithm for calculating DFT. Quoting from it's Wikipedia article: "The Cooley–Tukey algorithm, named after J.W. Cooley and John Tukey, is the most common fast Fourier transform (FFT) algorithm". It has been successful in achiveing a complexity of O(N logN) for composite numbers. Cooley-Tukey functions by expressing, recursively, the DFT of size N as N = N1 x N2 (where N1, N2 ≥ 1).

  There are a couple of improvements / pruning that can be made to this algorithm. For example, if we incorporate Chinese Remainder Theorem to choose N1 and N2, we can avoid the 'Twiddle Factors' arising as Phase terms. Also, if at any stage of the recursion N comes out to be a larger prime, the Cooley-Tukey algorithm deviates from its O(N logN) performance and moves towards a O(N²) performance. Here comes the Rader's Algorithm.

  Rader's Algorithm computes the DFT of prime sizes through a cyclic convolution. The performance of this Algorithm is O(N logN) (after zero padding N to a power of 2). If a DHT (Discrete Hartley Transform) is incorporated, using modified versions of the re-indexing routine, there can be a gain factor of two (Johnson and Frigo, 2007).

  Hence, the algorithm can be reasonably expected to achieve a O(1/2 N logN) performance, faster than original Cooley-Tukey Algorithm (O(N²) in prime case), Bluestein's Algorithm, Winograd's Algorithm and also better than a more recent Montium Tile-Processor (exp(N)).

  Therefore, this calls for a restructuring of the existing DFT mechanism. The major changes would be introduction of a new module to implement Chinese Remainder Theorem, another for Rader's Algorithm, and then it's cyclic convolution, Optimizing it by DHT, and Finally, the encapsulating Cooley-Tukey algorithm.

Related Software

• SLIB-PSD portable debugger for Scheme (requires emacs editor): psd1-3.tar.gz (62.kB)
  Now on Github: https://github.com/perttikellomaki/psd
• SCHELOG is an embedding of Prolog in Scheme+SLIB.

Miscellany

• Graphing example.
• Color-Scheme
• HTML labyrinth of a SLIB relational database
• Why relational database? Database Manifesto.
• SLIB has a reference implementation of Metric Interchange Format:
  Representation of numerical values and SI units in character strings for information interchanges
• FTP Links to SLIB and related software from this site and mirrors.
• htmls.zip, a collection of these html documentation files (600.kB)