MathGroup Archive 1995

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Mathematica for Linux is out

  • To: mathgroup at christensen.cybernetics.net
  • Subject: [mg1882] Re: Mathematica for Linux is out
  • From: ajk at garnet.berkeley.edu (Adam Jacobs)
  • Date: Sat, 12 Aug 1995 22:50:18 -0400
  • Organization: University of California, Berkeley

In article <3vlgn0$oc at gnumath.rutgers.edu> Andrew Leahy
   (leahy at gnumath.rutgers.edu) wrote:
>: I haven't seen this on the .announce newsgroup so I thought I might
>: mention it:
>: I just checked on the the Wolfram Research Web site and discovered
>: that the port of Mathematica to Linux is out. ...
>: The only drawback is that it's a text-based interface, so those of us
>: who are used to a notebook interface are going to have to suffer, I
>: guess.
>: I'd be interested in hearing installation and performance reviews from
>: anybody who picks this up.
>:
>: Andrew Leahy
>: leahy at math.rutgers.edu

I received my copy yesterday and installed it this morning.  It's
classic Mathematica, all right, and performs snappily on my 486DX2/66
with 32M.  My impression is that, as with other similar applications
(e.g. Maple V, the Linux version of which I already use a lot)
performance is similar to what you see on low-end RISC workstations.

The X-windows-based graphics (Motifps, etc.) are tolerably fast too,
unless you are displaying it on a remote X server, like I am most of
the time (I use an old SGI 4D/20 workstation as a 24-bitplane big screen
display for my Linux box).  In that case it seems a little sluggish,
something that's by no means unique to the Linux version; it needn't be
that way, I think (Maple doesn't have this problem).

The lack of an X notebook front end is a big minus; for one thing, one
gets used to it if one uses it on other platforms; for another, a
great deal of Mathematica material is available in notebook form: it's
a growing standard for exchanging math documents both for human
browsing and for interpretation by Mathematica.  I don't know why they
didn't release a version of the X-windows front end; it's been ported
to just about every other Unix workstation platform that they offer.
Perhaps -- hopefully -- they are planning one (anyone from WRI
listening?) and perhaps -- hopefully -- it won't cost (another) arm
and a leg to upgrade when it comes out.

For now I've neatly made up for the lack by using the graphics
renderers and the notebook front end snarfed from the SGI release on
my 4D/20, with a kernel running on the Linux box communicating via
MathLink.  (This is worth doing because my '486 Linux box is about
four or five times faster than the old SGI. except for 3D TrueColor
graphics, where of course the SGI excels!  And the SGI release has a
couple of neat extras, e.g. 3D plot and animation renderers using GL
that lets you zoom around your graphics in real time.)  This is a nice
marriage but of course very particular to my peculiar setup.  If you
have Linux only then for the time being it looks like you're stuck with
the linear text dialogue.

Installation is exactly the same as for other Unix workstations: untar
the distribution media (in this case, nine 3.5" floppies) somewhere,
e.g. /usr/local/math, run mathinfo, then call up WRI and get a
password (their automated email registration system seems not to be
working nowadays), put it in mathpass, and run math.install.  The
installation takes up 19 megs. 

I'm tempted to compare to Maple, which has had a Linux version out for
several months now (that I use extensively -- though now that I have
Mathematica available locally, I'll probably be using it for many of
the things I used to do in Maple).  Certainly they've done a good job
of porting their Motif-based front end; it'd be nice to see the same
for Mathematica.

Adam Jacobs
ajk at garnet.berkeley.edu


  • Prev by Date: Re: Speeding up a numerical integrator
  • Next by Date: Win95 and MMA
  • Previous by thread: Re: Speeding up a numerical integrator
  • Next by thread: Win95 and MMA