home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
CD Concept 6
/
CD Concept 06.iso
/
mac
/
UTILITAIRE
/
RLaB
/
misc
/
ANNOUNCE.v95
< prev
next >
Wrap
Text File
|
1994-02-21
|
9KB
|
245 lines
** RLaB Version 0.95 beta Available
For those of you already familiar with RLaB, version 0.95 is
available. There have been many bug-fixes, and many
improvements since the last announced release. Normally I do
not announce every release. If you want to get announcements
for each release then it would be best to join the
mailing-list. Mailing-list information follows this
announcement.
The main reason for this announcement is that the RLaB
installation procedure now uses GNU Autoconf. Additionally new
LAPACK routines have been added. Users wanting to update will
need to obtain the libraries as well as the RLaB source. the
required files are:
rlab-0.95.tar.gz (RLaB sources)
rlap-1.0.2.tar.gz (LAPACK f2c-sources)
rblas-1.0.tar.gz (BLAS f2c-sources)
rfft-1.0.tar.gz (FFTPACK f2c-sources)
rnlib-1.0.tar.gz (RANLIB f2c-sources)
Additionally there is a draft of the RLaB Primer in the doc
directory of the source distribution.
The ftp-sites are:
Primary Site:
evans.ee.adfa.oz.au pub/RLaB
pinus.slu.se pub/gnu/RLaB
csi.jpl.nasa.gov pub/matlab/RLaB
A more thorough description of RLaB is at the bottom of this
message.
Enjoy,
Ian Searle
ians@eskimo.com
-------------------------------------------------------------
Mailing-List Info
To be added or dropped:
Send mail to rlab-request@eskimo.com
This is read by a human, so no special syntax is required,
just state your desire.
To communicate with others on the list:
Send your message to rlab-list@eskimo.com
-------------------------------------------------------------
What Is RLaB ?
RLaB is _not_ a clone of languages such as those used by tools
like MATLAB # or matrix_X/Xmath ##. However, as RLaB focuses
on creating a good experimental environment (or laboratory) in
which to do matrix math, it can be called "MATLAB-like" since
its programming language possesses similar operators and
concepts. Extensive use has been made of the LAPACK, FFTPACK
and RANLIB sources available from netlib.
The most significant difference between the other proprietary
tools and RLaB is the GNU Copyleft.
RLaB has several types: numeric, string, function, and list.
numeric: real or complex
scalar, 1 or 2 dim array (matrix)
string: scalar, 1 or 2 dim array (matrix)
function: builtin, or user-written
list: A heterogeneous associative array
(user-defined structure), which can
contain numeric, string, function or
other list objects.
A set of built-in functions is provided, as well as a
capability to write your own functions in the RLaB language.
The language contains IF, FOR, and WHILE flow control
constructs, and a set of arithmetic, relational, and logical
operators.
The following is a list of the currently available functions:
abs diag int pause solve
acos diary int2str pclose sort
acosh diff intersection phrd sprintf
all disp inv plot sqrt
any dot isempty printf srand
asin eig isinf printmat std
asinh epsilon isnan prod strsplt
atan error issymm qr sum
atan2 eval length rand svd
atanh exist linspace rank sylv
backsub exp load rcond symm
balance eye log read system
cd factor log10 readm tan
ceil fft logspace real tanh
chol find lu redit tic
class finite lyap reshape tmp_file
clear fix matrix round toc
clearall floor max save trace
close format maxi scalar tril
compan fprintf mean schur triu
complement fvscope members set type
conj getline min setterm union
cos hess mini show what
cosh hilb mod showplot who
cross ifft nan sign whos
cumprod imag norm sin write
cumsum inf num2str sinh writem
det input ones size zeros
angle bandred banner center chop czt
detrend erf expm faxis fftplot filter
fliplr flipud fmin funm gamma hankel
house jordan lagrange logm max2 min2
mret ode4 ode78 pascal pinv qq_normal
rem rk4 toeplitz trapz window
The following is an example of a RLaB function:
//
// Modified Gram-Schmidt
// Given A (MxN), with rank(A) = N. The following algorithm computes
// the factorization A = Q*R (skinny QR) where Q (MxN) has orthonormal
// columns and R (NxN) is upper triangular
//
// Algorithm from MATRIX Computations,
// G.H. Golub, C.F. Van Loan (page 219)
//
mgs = function(A)
{
local(a,k,j,n,m,q,r); // default variable scope is global
a = A; // args passed by reference
m = a.nr; n = a.nc;
for(k in 1:n)
{
r[k;k] = norm( a[1:m;k], "2" );
q[1:m;k] = a[1:m;k]/r[k;k];
for(j in k+1:n)
{
r[k;j] = q[1:m;k]' * a[1:m;j];
a[1:m;j] = a[1:m;j] - q[1:m;k] * r[k;j];
}
}
return << q = q; r = r >>; // return a list
}
** Documentation:
1) A Complete RLaB Reference Manual is "in the works".
2.) There is a draft of the RLaB Primer available (in the doc
directory).
3) On-line help is present, and evolving (that is what you say
when something can only get better :-)
** Implementation:
RLaB is written in C. Although ANSI-C features have been used,
they can be turned off. RLaB has been compiled with Sun's
non-ANSI cc. The goal is, and will continue to be, to make
RLaB as portable as possible. RLaB now uses GNU autoconfigure,
so the installation is fairly simple.
LAPACK, FFTPACK, and RANLIB Fortran libraries are used.
Currently f2c'ed versions of these libraries are available. It
is possible to use native Fortran libraries if you have them,
and know how to link them with C sources. In the future the
choice of Fortran, or f2c will be selectable during configure.
For the time being all that is needed is a C compiler, and the
f2c libraries (available from netlib).
GNUPLOT 3.x is used to handle the interactive plotting.
Users are expected to obtain: GNUPLOT (prep.ai.mit.edu),
libI77.a libF77.a (from the f2c distribution,
research.att.com) themselves.
Rlab has been ported to: Most Unix platforms, and some
non-Unix platforms. Porting to non-Unix platforms is fairly
straightforward if pipes, and multi-tasking are supported.
** Current Status:
Beta test.
** Bug Reports:
Send bug reports, comments/criticisms, complaints,
suggestions, or whatever to: ians@eskimo.com. Please include a
script which exercises the bug (if possible), along with a
description of the hardware and software platform.
** Acknowledgments:
I would like to thank Phillip Musumeci and Brad Hards for
volunteering to do the bulk of the documentation.
I would like to thank Mike Brennan, Fook Fah Yap, Chris
Thewalt, Jef Layton, Ove Ewerlid, Tarun Soni, Roumen
Antonov, Charles H. Romine, Paul Cloke, Sze Tan, Jinghong
Chen, Alan Magnuson, Edwin Ng, Michael C. Grant, Thomas
Witelski, Timothy Kay, and many other testers and
contributors/critics.
Much thanks to the authors and sponsors of LAPACK, RANLIB, and
FFTPACK.
Thamks to the authors, and maintainers of GNUPLOT.
** Where to get it:
At present there are several ftp sites, all graciously
donated, they are:
evans.ee.adfa.oz.au pub/RLaB
pinus.slu.se pub/gnu/RLaB
csi.jpl.nasa.gov pub/matlab/RLaB
The set of files to get is:
rlab-0.95.tar.Z
rblas-1.0.tar.Z BLAS library
rlap-1.0.2.tar.Z LAPACK library
rfft-1.0.tar.Z FFTPACK library
rnlib-1.0.tar.Z RANLIB library
Please try and get the sources during non prime time hours.
# MATLAB is a trademark of The Math Works.
## matrix_X and Xmath are trademarks of I.S.I.