wester@unmvax.UUCP (04/18/85)
A number of people have asked questions about the comparable merits of
various symbolic mathematics system, in particular, Macsyma, Maple and SMP.
A useful place to look for such information is in the Proceedings of the
1984 MACSYMA Users' Conference (Schenectady, New York). In particular,
Stanly Steinberg and I wrote a paper comparing how well Macsyma, Maple and
SMP did their differentiation. We also compared how fast (and well) they
did when handed expressions like (x+1)^50 all expanded out, which they were
asked to differentiate and factor. We ran all these programs on a Vax 780
running Unix at the University of New Mexico and just did simple timings.
There were other papers in the Proceedings that at least described various
people's experience in using these packages (primarily, Macsyma, of course
since this was their conference) as well as other packages such a MuMath
developed by David Stoutemeyer which can run on a PC of some sort. I would
be interested in knowing if anyone has any strong opinions (for or against)
any of these systems (or any other interesting [general] symbolic math
package).
--
Michael Wester
University of New Mexico (Albuquerque, New Mexico)
~{anl-mcs|convex|csu-cs|gatech|lanl|ogcvax|ucbvax}!unmvax!wester
unmvax!wester@berkeley.ARPAjp@lanl.ARPA (04/19/85)
... I have used muMath on an MSDOS machine. It was about as powerful as Macsyma and somewhat faster (Macsyma was running on a VAX-750) The only problem was that muMath doesn't have as many functions as Macsyma. But, it is easy to write your own in the accompanying language muSimp. There are tutorials on the disk. And there is a small but useful newsletter that publishes new functions that are developed by the users. Eventually many of these find there way into the purchased package. The package is distributed by Microsoft. But it originated with, and the newsletter is published by, The Soft Warehouse, P.O. Box 11174, Honolulu, Hawaii 96828, USA. I recommend you write to them for info if you are interested. It's a nice package. It is also useful if your are interested in the inner workings because all the functions are written in muSimp. Jim Potter, jp@lanl.arpa a
Jeffrey P. Golden <JPG@MIT-MC> (04/26/85)
Forwarded for: Jeffrey P. Golden <JPG@MIT-MC> > From: jp@lanl.ARPA > Newsgroups: net.math.symbolic > Subject: Re: Comparing symbolic mathematics systems > Date: 19 Apr 85 03:23:47 GMT > Apparently-To: symalg > I have used muMath on an MSDOS machine. It was about as powerful as > Macsyma and somewhat faster (Macsyma was running on a VAX-750). > The only problem was that muMath doesn't have as many functions as > Macsyma. But, it is easy to write your own in the accompanying language > muSimp... What do you mean by "It was about as powerful as Macsyma"? Perhaps all you mean is that for the problems you were considering it did as well as Macsyma would have. You grant that Macsyma is more extensive (more functions). I believe if you look into it you'll find that Macsyma's knowledge is also more intensive, i.e. its simplification, factoring, integration routines, etc. are capable of more than MuMath's. I think MuMath is a great system, but in my book the above means that Macsyma is more powerful. Wrt the relative speed of systems, the following should not be accepted in specific cases without support, but sometimes a system is slower primarily because it is more powerful. *** REPLACE THIS LINE WITH YOUR MESSAGE ***
jp@lanl.ARPA (05/01/85)
> > From: jp@lanl.ARPA > > I have used muMath on an MSDOS machine. It was about as powerful as > > Macsyma and somewhat faster (Macsyma was running on a VAX-750). > > The only problem was that muMath doesn't have as many functions as > > Macsyma. But, it is easy to write your own in the accompanying language > > muSimp... > > What do you mean by "It was about as powerful as Macsyma"? Perhaps all > you mean is that for the problems you were considering it did as well as > Macsyma would have. You grant that Macsyma is more extensive (more > functions). I believe if you look into it you'll find that Macsyma's > knowledge is also more intensive, i.e. its simplification, factoring, > integration routines, etc. are capable of more than MuMath's. I think > MuMath is a great system, but in my book the above means that Macsyma > is more powerful. > Wrt the relative speed of systems, the following should not be accepted > in specific cases without support, but sometimes a system is slower > primarily because it is more powerful. > > Well, you're probably right. My knowledge of MACSYMA is limited to the problem I was working on (Characteristic polynomial of a symplectic matrix). But, not only did MuMath do the job faster (except for time spent writing new functions) but no one complained about my hogging the machine. MACSYMA apparently makes tremendous demands on the VAX-750. I had to have my privileges greatly increased. Then I was told "Better you should run at night." My computer may be slow but it's always working for me! Jim Potter jp@lanl.arpa *** REPLACE THIS LINE WITH YOUR MESSAGE ***