From chemistry-request@ccl.net Sun Jul 28 21:07:24 1991 Date: Sun, 28 Jul 91 17:40:19 PDT From: don%kinghorn.chem.WSU.EDU@yoda.eecs.wsu.edu (Don Kinghorn) To: chemistry@ccl.net Subject: Symbolic Math Packages Status: R I recently sent three e-mail messages inquiring about the usage of symbolic mathematics programs. The first message was a question about Mathematica usage sent to our (W.S.U.) local NeXT users group (Mathematica is bundled with the system software on NeXT's sold for academic use). The second letter was again about Mathematica but directed more specifically toward Chemists and submitted to mathgroup@yoda.ncsa.uiuc.edu, the Mathematica users mail-group. The third and most recent letter was an inquiry about symbolic mathemathics programs in general, submitted to chemistry@ccl.net, the computational chemistry mail-group. MANY MANY THANKS to all of you who responded to my requests for information. I only wish there would have been more of you. This is not the best time of year to be conducting a survey ---lots of people are away on vacation(like I wish I was) or engrossed in research(like I should be) etc.. I may repost my letters after the academic year gets underway and see if I get more of a response. For now I'll just make a few comments and post what I've received so far. I've decided to lump all of the letters and responses together since they are all related and probably of interest to anyone reading this. In my first two letters I was thinking of recommending Mathematica to our chemistry dept. and was looking for feedback from others. After looking at the responses I decided that more investigation was warranted and I wanted to see more about some of the other packages available. Thus, the third letter. From my own experience and the information that I have gathered so far it looks like Mathematica and Maple will be the main contenders for symbolic mathematics packages of interest to chemists. Mathematica's strong points are its wide spread use resulting in lots of freely available "packages" and "notebooks", a good example is the combinatorics.m package that complements Steven Skiena's book "Implementing Discrete Mathematics" which contains around 200 functions for investigating graph theoretic and enumerative problems. (Look in the ftp archive site at zaphod.ncsa.uiuc.edu under the Symbolic dir, for more) There is also a lot of course-ware (mostly mathematics and physics) being developed as Mathematica "notebooks". The Mathematica feature probably of most interest to chemists is the graphics. For example, the other day a colleague needed some plots of orbital density data on CS2. He e-mailed me the data, I opened a Mathematica notebook, read in the data, reflected it across a mirror plain, then generated density, contour, and 3D-surface plots, in postscript ready to print or cut and past into another program, all in a matter of minutes. But, Mathematica HAS LOTS OF PROBLEMS!! The review of Mathematica by Richard Fateman, is the best review I have ever seen of any software package. It really helps balance the hype surrounding Mathematica. You can find a copy (revised 21 july 1991) via anonymous ftp on peoplesparc.berkeley.edu I highly recommend it. I would like to see a review like this on Maple. Maple appears to be a very robust program and is probably more reliable than Mathematica. I have not used Maple myself and need more feedback from users in order to give it a fare evaluation. Most everything I have seen or heard so far about Maple has been positive. Waterloo Maple Software was very quick to send me an information packet including a small MSDOS demo disk. The program looks like a strong competitor for Mahtematica and is probably better in some aspects. For those wanting to run the program on multiple CPU's their site licensing looks attractive. The latest release of Maple has added more graphics capabilities, but I think they need to do more, and I,m sure they will. There are good reviews comparing Derive, Macsyma, Maple and Mathematica by Barry Simon in, "Notices of the American Mathematical Society",(Sept. 1990,vol. 37 num. 7) and "PC magazine" (May 29 1990 vol. 9, num. 10). You can look through the letters below to find the addresses for the people who sent me the info. package on Maple mentioned above and a similar package from Wolfram Research on Mathematica (with a Mac disk of sample notebooks). At this point I would say the jury is still out on these programs. I would like to here more about the other packages REDUCE, Derive, etc.. If time and interest permit I'll try another round of e-mail inquiries in the fall and see what I get. I'll keep you posted. What follows are some of the responses I received to my letters. (I left some out to avoid redundency and edited out a few personal comments etc..) Donald B. Kinghorn Washington State University Dept. of Chemistry Pullman WA 99164-4630 don@kinghorn.chem.wsu.edu ***************************************** Errors-To: owner-chemistry@ccl.net Date: Wed, 17 Jul 91 13:22:00 PDT From: don%kinghorn.chem.WSU.EDU@yoda.eecs.wsu.edu (Don Kinghorn) To: chemistry@ccl.net Subject: symbolic math software? Sender: chemistry-request@ccl.net Errors-To: owner-chemistry@ccl.net Precedence: bulk I'm trying to find information on symbolic mathematics and mathematical programming software. In particular the following packages: Derive GAUSS MACSYMA Maple Mathematica MATLAB REDUCE Theorist and any others I may have left out. (I'm interested in programs running on PC's, Macintosh, NeXT, UNIX workstations, VAX, IBM3090.) I am interested in your opinions on these packages, references to review articles, etc... Does anyone know of a review article(s) that compares features and performance of these packages? I know Mahtematica has an electronic mailgroup and ftp archive site, what about the others? Any information or opinions would be most appreciated. If I get lots of interesting responses I'll gladly summarize and repost if anyone else is interested. Also, please feel free to forward this message to your colleagues and other maillists if you think they might have helpful information. Please reply to: don@kinghorn.chem.wsu.edu Thanks :-) Donald B. Kinghorn Washington State University Dept. of Chemistry *********************************************** Date: Wed, 17 Jul 91 18:02:06 EDT From: Marc Roussel To: don@kinghorn.chem.wsu.edu Subject: Symbolic Math packages Our system administrator forwarded me your message since I seem to be the local expert on symbolic computation. (Of course around here, it doesn't always take much to be the local expert...) If you want to see a review of symbolic math packages on the PC, look through last year's volume of PC Magazine. I have worked a lot with Maple, a little with Mathematica and know about some of the others you mentioned. I'll tell you what I know, which admittedly is not much. There are several issues which you must examine. In some ways, the central issue is the quality and stability of the computing engine. In this department, Maple, Macsyma and Derive seem to be the best. The basic Maple engine still needs a little fine tuning however (which is now in the works). Mathematica seems to have several basic design flaws that it will never grow out of. If you do a lot of ticklish integrals, Macsyma or Derive may be your best choices. On the other hand, Macsyma is huge and consumes extraordinary amounts of machine resources. As far as the extent and quality of libraries goes, Maple and Macsyma seem to be the best. Because of its pretty interface and the fact that it comes free with some workstations, Mathematica has attracted a lot of users who have in turn generated a lot of publicly available software of varying quality. You will no doubt want to write symbolic programs of your own. Maple and Macsyma are probably the best systems in this respect, although I'll admit that I know nothing about Derive's programming interface. Macsyma uses a Lisp-based programming language which can be either a good thing or a bad depending on whether your research group has any Lisp expertise. Maple provides a more ordinary procedural language (somewhat like C/Fortran/Structured BASIC/Turing). Mathematica's programming interface is full of inconsistencies. I will send you an article we have on Mathematica which backs up this opinion. If you have the sort of site where you really want or need source code, this can be purchased separately in the case of Maple and I think perhaps also for Macsyma. Even if you don't purchase the source, you can view Maple's library source interactively. (It's also possible to save bits of it in human-readable form. In other words, even without a source license, you get most of the source.) Maple's libraries are written in Maple and loaded into your system in some sort of proprietary compressed format. Mathematica has a big advantage when it comes to the user interface, particularly with respect to educational uses. Maple is closing the gap rapidly however. I understand that the latest version of Maple has all the bells and whistles that Mathematica has, with the exception of notebooks. Mathematica notebooks provide a sort of an elaborate presentation facility. You have to see it to understand what I'm talking about. Anyway, the notebooks are great for teaching. Price is of course always a consideration. When we were looking last, Maple had a big edge there for educational sites where the program would be running on multiple CPU's. I understand that Mathematica has introduced some sort of site-licensing in response, and Macsyma has always been competitive. Derive has always been cheap, but then it's a little less powerful in some ways. You should look into this carefully as my information may be somewhat out of date. Support is also quite important. Maple and Macsyma are both supported through electronic mailing lists. You can also buy a software maintenance contract from the Maple folk. I presume the others offer this too, but I don't know. The Maple support people are quite knowledgeable and fast. Serious bug reports almost always result in instant "Fixed in the next version" responses on which are inevitably followed through. By contrast, the Mathematica people answer "that's the way it's supposed to work" just a little too often for my taste. Overall, I'd say that Maple is the best choice for an academic site that is going to use a symbolic language for research. If you are going to use such a thing for teaching as well, things become more complicated and you have to think very hard about how exactly you're going to use the thing. In either case, you shouldn't expect these things to miraculously transform your work. Symbolic computation is still very much in its infancy. Sincerely, Marc R. Roussel mroussel@alchemy.chem.utoronto.ca ******************************************************* Date: Wed, 17 Jul 91 15:56:02 CDT From: twhitley@ncsa.uiuc.edu (Timothy A. Whitley) To: don@kinghorn.chem.wsu.edu Subject: Math packages Don, You didn't ask explicitly, but the following are listed in the Cray Research, Inc., Directory of Application Software: MACSYMA Maple MatLab REDUCE They are available for UNICOS (and probably some on COS) operating systems. [You might not be pleased with the price of these: MACSYMA costs, like $2500, but needs LISP to run. Franz, Inc., wants around $75,000 for their LISP. I have never run any of these on a Cray because of the costs. I also don't know about the performance.] Mathematica is not available for the Crays, even though Wolfram Research, Inc., listed this version at one time. [They listed a price for the Cray-2 version of $240,000! This is actually a common practice--to determine what your development costs will be, and charge that full amount for the first copy!] =~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~= ~=~= Tim Whitley (( Internet...twhitley@ncsa.uiuc.edu Cray Research, Inc. )) Phone......(217) 244-4863 4139 Beckman Institute (( FAX........(217) 244-2909 405 N. Mathews Avenue )) "The views presented herein are not those Urbana, IL 61801 (( of Cray Research, but are solely my own." =~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~= ~=~= ******************************************************** To: don@kinghorn.chem.wsu.edu Cc: wmsical!dsraimdo@watmath.waterloo.edu Subject: math packages Date: Thu, 18 Jul 91 06:57:17 -0700 From: Lyle Wiedeman I work for the Office of Academic Computing at U. Calif. Irvine. We recently went through just such an evaluation, and settled on Maple for its price (including very good site-licensing policies), broad platform support, ease of use, and good support of graphics, on-line documentation, and X. If it has a down side, it is that its official documentation will be released this fall by Springer-Verlag who are perhaps the most expensive publishing outfit in the country. As far as a comparison article, the best one I've seen was from the Sept 1990 issue of the _Notes from the American Mathematical Society_ (NAMS v37n7) "Four Computer Mathematical Environments" by Barry Simon. He also did a somewhat less technical comparison in PC Magazine in 1990, "The New World of Higher Math," but my reprint doesn't have the date or page number. You can get a reprint of the PC article readily by e-mail request to Waterloo Maple Software (Dave Raimondo, wmsical!dsraimdo@watmath.waterloo.edu). Lyle Wiedeman Office of Academic Computing wiedeman@uci.edu Univ. Calif. Irvine wiedeman@UCI.BITNET Irvine, CA 92717 (714) 856-8718 FAX (714) 725-2069 ********************************************************************* ** From: wmsical!dsraimdo@watmath.waterloo.edu Date: Fri, 19 Jul 91 11:19:03 EDT To: don@kinghorn.chem.wsu.edu Subject: Prices on Maple Manuals Re: Prices on manuals for Maple from Springer Verlag Maple Language Reference Manual $24.95 Maple Library Reference Manual $39.00 First Leaves: Tutorial $19.95 US pricing David Raimondo Sales Executive Waterloo Maple Software 160 Columbia Street West Waterloo, Ontario Canada N2L 3L3 Phone: 519-747-2373 Fax: 519-747-5284 E-mail: dsraimondo@daisy.waterloo.edu dsraimondo@daisy.uwaterloo.ca ********************************************************************* ****** From: wmsical!dsraimdo@watmath.waterloo.edu Date: Fri, 19 Jul 91 08:25:35 EDT To: don@kinghorn.chem.wsu.edu Subject: Maple Information Dear Professor Kinghorn: I have mailed you a complete information package on Waterloo Maple Software. If I can be of further service, please let me know. David Raimondo Sales Executive Waterloo Maple Software 160 Columbia Street West Waterloo, Ontario Canada N2L 3L3 Phone: 519-747-2373 Fax: 519-747-5284 E-mail: dsraimondo@daisy.waterloo.edu dsraimondo@daisy.uwaterloo.ca ********************************************************************* ******** Date: Thu, 18 Jul 91 00:08:31 CDT From: GR2162%SIUCVMB.BITNET@cunyvm.cuny.edu To: don@kinghorn.chem.wsu.edu Subject: REDUCE There is a REDUCE-netlib ran by the Rand Org. Recently, the new version of REDUCE 3.4 has been released. To receive info on the distribution of this release send a bitnet msg of "send info-package " to reduce-netlib@rand.org . I would also be interseted in other such networks if they do exist. Lonnie W. Manning GR2162@SIUCVMB ********************************************************************* ******* Date: Wed, 17 Jul 91 14:48:12 -0700 From: Judy Bayard Cushing To: don@kinghorn.chem.wsu.edu Subject: Re: symbolic math software? Sorry, I don't have info on a review for you, but would like info if you get some from folks - I'm particularly interested in object oriented database interfaces to such s/w. Have you thought about adding "S" to your list? it's probably not, strictly speaking, symbolic math, but has capability for interfacing its data to FORTRAN and maybe C routines. Good graphics. and good analysis. -judy **************************************************************** ********************************Second Letter******************* **************************************************************** Date: Fri, 7 Jun 91 16:13:26 PDT From: don@kinghorn.chem.wsu.edu (Don Kinghorn) To: mathgroup@yoda.ncsa.uiuc.edu Subject: Mathematic in Chemistry? Hello Mathematica users I'm a Chemistry graduate student and bipolar Mathematica user(sometimes I love it sometimes I hate it). I'm trying to find Mathematica users who are doing Chemistry related work. I know a few Mathematics and Physics departments are using Mathematica in their course work. Are there any Chemistry departments doing likewise? Do you use it in your research? If so, I would appreciate it if you would take the time to send me a short note describing what you use it for. I'm trying to convince the Chemistry dept. here that having Mathematica available to their students and faculty ( maybe on a NeXT platform ) would be an enhancement to their program. If I can point to Mathematica users with interests in Chemistry at other universities and describe what they're doing it might help my case. I have used it mostly as a fancy calculator for doing problem assignments in my course work and preparing solution sets for the graduate courses I have graded. I feel that having the program more widely available would allow instructors to assign more meaningful and interesting problems. I have also used it as a prototyping language for a nonlinear optimization problem using simulated annealing and a few particle nonadiabatic quantum chemical calculation I'm doing with an explicitly correlated basis set.I consider Mathematica a valuable tool for research and course work. I use it for something almost every day. I,m looking forward to the fixes and enhancements in Mathematica 2.0 and plan on doing some Notebook and Package development. If you know of any Chemists using Mathematica that might not be on this mailing list please forward a copy of this note to them. It's probably best to reply to me directly. I'll summarize and repost the responses. Thanks :-) Donald B. Kinghorn Washington State University Dept. of Chemistry don@kinghorn.chem.wsu.edu ********************************************************************* *** Date: 09 Jun 91 18:57:29 EDT From: Roberto Sierra <73557.2101@compuserve.com> To: Subject: Mathematica in Chemistry? Date 6/9/91 Subject Mathematica in Chemistry? >From Roberto Sierra To Don Kinghorn CC Harry Atwater Amy Bug *Carl Feynman Oliver Grillmayer James Joaquin Mike Sierra Reply to: Mathematica in Chemistry? Hello, Don -- Not being a chemist by training (I'm an E.E.), I couldn't help out much on the chemistry-specific end of things, but perhaps my comments might be of value. Also I will be forwarding your message (attached below) to some friends at Apple and on the faculty at Berkeley, Caltech, and Swarthmore to see if they might know of someone who teaches chemistry with Mathematica. I use Mathematica mostly to work with symbolic expressions, matrices, test out signal-processing algorithms before they make it as far as C or assembler code, and also do quite a bit of work using the 2-D and 3-D graphics tools. [I've been tinkering, for example, with a math package that will allow any Graphics3D object generated by Mathematica to be exported to a shareware ray-tracer commonly available for the Macintosh for high quality plots of arbitrary 3-D surfaces -- no release date yet.] It struck me long ago that the combined equation-solving & graphics capabilities of Mathematica are ideal for teaching environments. One thing you might consider is using Mathematica to plot 3-D representations of molecular structure. There's already been a little work done along these lines in the 'Crystals.m' (or is it 'CrystalStructures.m'?) package that ships with Mathematica. A little tweaking might make it more ideally suited for chemists and might convince the 'powers-that-be' better than anything having to do with equations (graphics will get you money over text any day, from my experience). Anyway, it's just a thought, and I'd love to find out how your quest proceeds. I'm a software developer by trade these days, so I'm always interested in finding out how people use software in new and different ways. P.S. -- I love your term 'bipolar Mathematica user'. I think most users tend to get polarized that way. I think I like it much more than I hate it, but some parts drive me up the wall. I think 2.0 will fix some of the trouble I've had. Yours, \\|// - - o o J roberto sierra O tempered microdesigns \_/ 73557.2101@compuserve.com ********************************************************************* ** Date: Mon, 10 Jun 1991 17:34:44 MDT From: WHITEHEAD@asucp1.la.asu.edu Subject: Mathematica To: don@kinghorn.chem.wsu.edu X-Vmsmail-To: SMTP%"don@kinghorn.chem.wsu.edu" Don, I'm using Mathematica in conjunction with a glass simulation program. The program spits out coordinates and mathematica plots them for me in 3D. Also, I change the viewing angle of one of them and use special glasses to _really_ get the three-dee affect. Eventually, I may make a movie outof this system reaching equilibrium in Mathematica. Matt Whitehead whitehead@asucp1.la.asu.edu ********************************************************************* *** Date: Sun, 9 Jun 91 12:38:50 -0500 From: llave@math.utexas.edu Posted-Date: Sun, 9 Jun 91 12:38:50 -0500 To: don@kinghorn.chem.wsu.edu WE have had a very bad experinece with the NeXT's. They promise delivery daytes that they do not fulfilland the quality of the machines is very low. Before you enbark in buying one, check that they can deliver what they promise in their catalogues. ******************************************************************** "NOTE I was sad to see this letter because I have had a very good experience with NeXT. I think they probably rushed the announcement and production of their MC68040 workstations last fall. This happens all to often in the computer business. I think NeXT has a very good product for a great price Don Kinghorn" ******************************************************************** ******************************************************************* Date: Mon, 10 Jun 91 08:42:07 -0500 From: maury@dragonfly.wri.com To: don@kinghorn.chem.wsu.edu Subject: Chemistry Don: I've passed your MathGroup enquiry on to our PR Coordinator, who might have some ideas. She'll probably be in touch. Maury Kendall Wolfram Research, Inc. ********************************************************************* **** NOTE I received a nice packet from: Eydie Doehring Sales Associate Wolfram Research, Inc. Champaign, IL 61820 217-398-0700 fax: 217-398-0747 email: eydied@wri.com ************************Don Kinghorn***************************** ***************************************************************** Date: Sat, 29 Jun 91 16:55:16 -0400 From: pns@fermi.clas.virginia.edu (Paul N. Schatz) X-Mailer: Mail User's Shell (7.2.3 5/22/91) To: don@kinghorn.chem.wsu.edu Subject: Mma in chemistry I have a similar bipolar response to mma, but I do continue to use it and hope for the promised improvements in version 2.0. I have made significant use of it in various theoretical work and expect my usage to expand. At least one other faculty member here has made appreciable use of the program. In addition, the mma kernel is now available on an RS/6000 mainframe here. That means that one for example has the all the advantages of a MAC frontend plus the much greater power and speed of the RS/6000 kernel. It is easy to access the kernel, and it all seems to work as advertized. There has not been any organized use by the Chemistry Dept., but I have heard rumors about its possible use in introductory calculus courses here. I would think that an economical way of making mma widely available would be to persuade your computer center to put it on some mainframe. Then individual users would only have to buy the frontend. I believe the kernel for the RS/6000 here only cost about $1,700. More generally, I think the usage of these kinds of symbolic/numeric/plotting/etc. languages/methods are bound to grow. They are clearly of great power in dealing with a wide array of problems. My main frustration with mma has been its unfriendliness in practice. Although it is all there in the book, it can often be absurdly difficult (at least for me!) to do what should be obviously simple things. I am sure this situation will improve as I use it more, and get in the habit of using it more routinely. Keep in touch and let me know what you find out. Sincerely, Paul Schatz Chem Prof UVA pns@virginia.edu ******************************************************************* Sun, 9 Jun 91 14:54:34 EDT From: shenkin@avogadro.barnard.columbia.edu (Peter S. Shenkin) To: don@kinghorn.chem.wsu.edu (Don Kinghorn) Subject: Re: Mathematic in Chemistry? I bought mathematica when I got my Iris, and I use it for a variety of things, but mostly I'm very unhappy with it. I think it tries to do to many things,and does them all rather poorly. It is extremely user-unfriendly, and it is very difficult to learn as a language. I mostly use it for two things: (1) checking complicated multiple derivatives and integrals I sometimes need, and other derivations as well; (2) 3d plots. My understanding is that for symbolic math, the other programs out there (Maple, Macsyma) are much better; but I don't really know these programs, so I can't say. As a numerical front-end with tightly coupled graphics, I suspect any of MatLab, S, Gauss (on IBM-PC compatibles) are far superior. Also much cheaper. (MatLab for the Iris is about $600; Mathematica is $2000; both prices include academic discounts.) Caveat: once you learn a system, it's hard for someone else to convince you you should drop it and learn something else; thus, I know vi, and I'm not going to drop it and learn emacs, and I know troff, and I'm not going to drop it and learn TeX, until I have to, though I am convinced that the second member of each of these pairs is superior to the one I know. So if you know mathematica and can make it jump through your particular set of hoops, great. My guess, however, is that other alternatives are better, in the sense of being more useful to a greater number of people. Also, it's hard to find people who know enough about both, say, Mathematica and S, or MatLab, or Gauss to really do a good comparison. I certainly don't. -P. ************************f*u*cn*rd*ths*u*cn*gt*a*gd*jb**************** ********* Peter S. Shenkin, Department of Chemistry, Barnard College, New York, NY 10027 (212)854-1418 shenkin@avogadro.barnard.columbia.edu shenkin@cunixc.BITNET ***"In scenic New York... where the third world is only a subway ride away."** ******************************************************************* The following is a responses to my first letter which was very similar to my second letter, but sent to our local NeXT user group ******************************************************************* Date: Mon, 03 Jun 91 15:27:18 PLT From: Joshua Yeidel Organization: Academic Computing Services, Washington State University Subject: Re: Do you use Mathematica? To: Don Kinghorn In-Reply-To: Your message of Mon, 3 Jun 91 15:21:57 PDT I have used Mathematica for an animation of soliton propogation for Daviud Hughes in Physics. It was a little bit of a struggle, since I didn't know Mathematica _at all_ prior to starting, but it turned out to be only 30 lines of code, so I expect that a regular user could have done the job in very little time. The data came from simulations which had been run in batch on the 3090. The data was then passed to the NeXT via ftp. The animation program read the data and prepared frames, then the standard Mathematica "animate" command was used to control the animation interactively (play, reverse, rewind, frame rate control, etc.) Because the data set was very large (and the preparation of frames was very time consuming), the program interactively obtained "start" and "stop" times to limit the work to times of interest to the researcher. ********END************