Hack 90 Analyze Quantian

Quantian is an extension of Knoppix and ClusterKnoppix tailored to numerical and quantitative analysis.

The Quantian live DVD distribution (http://dirk.eddelbuettel.com/quantian.html) adds a quantitative facet to Knoppix. Based on ClusterKnoppix [Hack #89], Quantian adds software with a quantitative, numerical, and scientific focus: several computer-algebra systems; higher-level matrix languages; data-visualization tools; a variety of scientific, numeric, and engineering applications, as well as many different programming languages and libraries. A particular focal point is the R language and an environment for statistical computing.

With its unique combination of the Knoppix-based ease of use and the additions of both openMosix cluster computing and a very rich set of scientific packages, Quantian can be of immediate use. It can:

• Compute clusters in order to speed up embarrassingly parallel tasks

• Create computer labs by enabling temporary use of a computing environment booted off a DVD and can netboot other machines

• Enable students and coworkers by distributing DVDs that allow everyone to work in identical environments with minimal administration cost

• Provide convenience to users so they do not have to chase down new software releases, and then manually configure and install them

• Provide easier installation of a normal scientific workstation by booting off Quantian and installing that system to hard disk, resulting in 3.6 GB of configured software

• Provide a familiar environment to work in during trips to conferences or other campuses

• Put older hand-me-down machines unable to run the newest and greatest commercial OS to use for Linux

The first two Quantian releases were based directly on Knoppix. Since then, ClusterKnoppix has provided the basic building block, which, similar to Knoppix itself, provides various utilities, games, multimedia applications, a complete KDE environment with its window manager, a browser, an office suite, a development environment and editors, as well as a large number of other general-purpose tools, utilities, and diagnostic applications. Quantian then adds various sets of applications from different areas:

Mathematics

Computer-algebra systems Maxima, Pari/GP, GAP, GiNaC, YaCaS, and Axion, matrix-oriented languages; Octave (with octave-forge, matwrap, and octave-epstk packages), Yorick and Scilab, and the TeXmacs frontend

Statistics

GNU R (with numerous packages from CRAN, BioConductor, Rmetrics, and other archives, as well as Ggobi and ESS tools), Xlispstat, Gretl, PSPP, and X12A

Bioinformatics

BioConductor packages for R, BioPython, and BioPerl, and tools like emboss and blast2

Physics

CERN tools like Cernlib, Geant, PAW/PAW++, Scientific and Numeric Python, and the GNU GSL libraries

Visualization and graphics

OpenDX, Mayavi, Gnuplot, Grace, Gri, plotutils, and xfig

Finance

Software from the Rmetrics and QuantLib projects

Programming languages

C, C++, Fortran, Java, Perl, Python, PHP, Ruby, Lua, Tcl, Awk, and A+

Editors

XEmacs, Vim, jed, joe, kate, nedit, and zile

Scientific publishing

Extended LaTeX support with several frontends (xemacs, kile, and lyx) and numerous extensions and tools for LaTeX and BibTeX

Office software

OpenOffice.org, KOffice, Gnumeric, and tools like the Gimp

Networking

Ethereal, portmap, netcat, ethercap, bittorent, nmap, squid, and a host of wireless tools and drivers

General tools

Apache, MySQL, PHP, and more

Quantian isn't limited to these applications. Quantian also provides instant openMosix clustering with fully automatic configuration (see [Hack #89] ), as well as the ability to let other machines boot over the network (in the openmosixterminalserver mode) and become additional cluster nodes. Moreover, through the addition of suitable libraries and applications for the pvm and lam/mpich message-passing interfaces, Quantian also provides Beowulf clustering support. Both openMosix and Beowulf styles of distributed computing can be combined—for example, multiple pvm or pvm node clients can be started on a single host in order to let openMosix spread the load most efficiently across the cluster (see Figure 8-4).

Figure 8-4. Quantian managing a cluster

8.12.1 See Also

• [Hack #89]

—Dirk Eddelbuettel