From DSOUTH@uoft02.utoledo.edu Sat Oct 6 22:55:47 1992 Date: 07 Oct 1992 02:55:47 -0400 (EDT) From: DSOUTH@uoft02.utoledo.edu Subject: MPP vs Superscaler (was: Wavetracer bites the dust) To: chemistry@ccl.net > Is there really a big market for limited-purpose attached > processors? The number of fast workstations being developed > seems to indicate that a general purpose computer is more > cost effective for the majority of scientists and engineers. No, what the "number of fast workstations being developed" indicates is that is easier to design scaler workstations (and hence to recover the design costs and make $$, which is after all the business that computer manufacturers are in). The types of problems encountered in computational chemistry lend themselves to MPP to a greater degree than many other computer applications. See Supercomputing Review Mar & Apr '92 for a nice summary of why MPP is or isn't the wave of the future. Of course the difficulties involed in working with even a 4000 PE SIMD/ASIMD platforms make writing vectorizable code look simple (trust me on this -- I'm working on an 8000 PE MasPar right now). It isn't impossible by any means, but it is certainly more difficult than cf77 -o aggress. Of course when one does make it work, the rewards are benchmark times that can match a Cray costing 20 times as much (which would seem to me to be "cost effective" in terms of execution speed, though perhaps not in development time). Super-scaler/Super-pipelined RISC IS a great advance, but don't expect it to expell SIMD/ASIMD/MIMD from the middle and high end (though the death of Wavetracer may spell the end of any parallel penetration into the low-end). Parallel is here to stay. -- /* Dale Southard Jr. -- Smith Research Group Sr. Rigger */ /* Department of Chemistry AFF/I SL/I */ /* University of Toledo D-11216 */ /* dsouth@uoft02.utoledo.edu -- "Just another skydiving grad student." */ From mcdonald@cuchmc.chem.columbia.edu Wed Oct 7 06:50:49 1992 Date: Wed, 7 Oct 92 10:50:49 -0400 From: Clinton McDonald To: chemistry@ccl.net Subject: GO Plots, summary of responses Dear Netters, Thanks to Adi Treasurywala for supplying the following references with respect to "GO" plots for the analysis of dynamics trajectories. >>> Y.Seno and N.Go J Mol Biol (1990) v216 pp95-109. >>> >>> HOWEVER >>> >>> there's a little known fact that these are really Ooi plots!! >>> >>> See: >>> >>> T.Ooi and K Nishikawa Jerusalem Symp. on Quantum Chemistry >>> and Biochemistry v5 (1973) >>> Quentin McDonald mcdonald@cuchmc.chem.columbia.edu From shepard@dirac.tcg.anl.gov Wed Oct 7 05:28:42 1992 Date: Wed, 7 Oct 92 10:28:42 CDT From: shepard@dirac.tcg.anl.gov (Ron Shepard) To: chemistry@ccl.net Subject: more Barbie Doll comments This Barbie Doll discussion brings to mind a conversation I had with a 10th grade daughter of an acquaintence a few years ago. She was taking Geometry, enjoying the course, and doing well in it. I asked her if she had considered continuing in math or taking science courses later in high school or even later in college. Her reply was that she didn't do well in math and had always heard that math was required in science courses. I tried to make the point that Geometry was probably the FIRST REAL math course that she had ever taken. Everything before that is dominated by arithmetic, which although necessary is rather a shallow subject, and that her enjoyment of and performance in plane geometry was more representative of future work in that area. The reason this conversation sticks in my mind so much is that I felt "society" (or "culture" or whatever is the appropriate term) had already conditioned her to avoid "math", even with evidence to the contrary that she displayed some talent in the area. I guess I should conclude the story by saying that I believe she eventually went on to study commercial art. I can only assume (or hope optimistically) that our loss was someone else's gain. -Ron Shepard From WILLIAMS@flint.bucknell.edu Sun Oct 7 09:39:00 1992 Date: 07 Oct 1992 13:39:00 EDT From: Brian Williams Subject: DIPOLE MOMENTS FOR HETEROCYCLIC ORGANICS To: CHEMISTRY@ccl.net I HAVE RECENTLY BECOME INTERESTED IN DIPOLE MOMENT CHANGES BETWEEN GROUND AND EXCITED ELECTRONIC STATES AND HOW THESE AFFECT EMISSION/ABSORPTION PROPERTIES. I HAVE THE SENSE THAT COMPUTATIONAL CHEMISTS KNOW HOW TO ESTIMATE OR CALCULATE SUCH CHANGES, BUT I'M NOT TERRIBLY WELL VERSED IN THIS TYPE OF THEORY OR HOW MIGHT ONE GO ABOUT GETTING THIS KIND OF INFORMATION FROM THE TYPE OF CALCULATIONS OF ELECTRONIC DISTRIBUTION THAT I IMAGINE COMPUTATIONAL CHEMISTS DO. CAN ANYONE ON THE NET OFFER SOME ADVICE AS TO WHERE TO START OR AS TO WHETHER THERE ARE SOURCES FOR THIS KIND OF INFORMATION? IM AN EXPERIMENTAL P-CHEMIST BY TRAINING, AND NOT AFRAID OF LEARNING SOME MATH, BUT I NEED TO START AT AN ELEMENTARY LEVEL IN THESE KINDS OF QUESTIONS. THANX FOR ANY HELP! BRIAN WILLIAMS, WILLIAMS@BUCKNELL.EDU From jstewart@fai.com Tue Oct 6 17:11:22 1992 Date: Wed, 7 Oct 92 11:46:22-1795 From: jstewart@fai.com (Dr. James Stewart) To: chemistry@ccl.net Subject: Rotation Invariance, more thoughts Rotational Invariance Mark A. Thompson writes: >The concerns raised were not with possible rotational variations >in MOPAC, or any other program, but rather the inherant >rotational variance that may exist in the MNDO-family of methods >themselves (ie. MNDO, AM1, PM3). By far the most common reason that writers have claimed loss of rotational invariance in semiempirical methods stems from an old approximation that was corrected in 1982. In MNDO the variable Hpp = 1/2(Gpp-Gp2). All three quantities were stored in a BLOCK DATA statement, as N.NN. If, e.g., Gpp-Gp2 = 2.35, then Hpp = 1.175, which was rounded to 1.17 or 1.18. This small round-off resulted in loss of rotational invariance for 5 elements. When I wrote MOPAC in 1983, I `wrote out' Hpp, and replaced it everywhere by 0.5(Gpp-Gp2), see subroutine FOCK1. Since then, as I said earlier, writers have latched on to the `error' in the original formulation, and not noticed that it no longer exists. As far as the two-electron two-center integrals are concerned, there were 22 of these in the original program. Using trigonometry, it can be shown that one of these, the term is related to two others by = 1/2( - ) In the original program, the 22nd integral was calculated independently. This resulted in a small, but definite, loss of rotational invariance. By making this substitution, which can be seen in subroutine REPP, rotational invariance is restored. The source code of MOPAC-6 is in the public domain, and is readily available. So please feel free to check these points. Please do not think that I disapprove of questions being asked about the ideas and programming in MOPAC. Heaven knows, each new MOPAC has several HUNDRED small corrections made over the previous version, and MOPAC-7, when it's released, will correct about 250 small errors in MOPAC-6. What I'm saying is, there is always room for improvement. However, I'm sure there are no errors which involve loss of rotational invariance in the MNDO-family, and I'm sure that any loss due to programming errors cannot be more than about 0.00001 kcal/mol. In recognition of the research nature of the MOPAC software, only source code will be distributed. While this makes the program somewhat vulnerable to piracy, it is more important that everyone should be able to see inside the `black box'. In the past two years, most of the improvements to MOPAC have come from users who have had access to the code. Someday, these programs will be as reliable as some GUIs are today, at which point I suppose we will stop distributing source code. But that won't be tomorrow! >I have the highest regard for the positive impact these methods have had >on chemistry. Please keep in mind that I am merely searching for a >deeper understanding at this point. A good sentiment! The sceptical chymist is not dead! James J. P. Stewart FAX 719 488 9416 Email jstewart@fai.com From PEARLMAN@VAX.PHR.UTEXAS.EDU Wed Oct 7 12:45:08 1992 Date: Wed, 7 Oct 1992 17:45:08 -0500 (CDT) From: PEARLMAN@VAX.PHR.UTEXAS.EDU Subject: Talking Barbie is only 1/67th as bad as you might think To: CHEMISTRY@ccl.net I'm hesitant to post this since Jan suggested yesterday that we terminate the Barbie discussion. However, since I've seen no less than eight other Barbie postings since his "request," I thought one more wouldn't hurt. The original Barbie posting suggested that Talking Barbie can only say four things --- one of which is something like 'Math is hard.' The actual case, however, is not quite that bad. Each Talking Barbie is "programmed" at the factory to say four sentences which are "randomly selected" from a total of 270 possible sentences. With 5,197,112,280 possible combinations, each child can be assured of having his/her *unique -- special -- PERSONAL* Talking Barbie which, no doubt, enhances the child-doll bonding which Mattel hopes will be established. The probability that the 'Math is hard' sentence will be selected is 1/270 + 1/269 + 1/268 + 1/267 (roughly, one out of 67). Please note that I agree that even a 1 in 67 chance of getting a math-phobic Barbie is inexcusable! (But it's 67 times better than a 100% chance.) Now, don't you wonder what some of the other 269 sentences might be? Time for me to get back to work. Cheers, --- Bob Pearlman From CHM001@MARSHALL.MU.WVNET.EDU Sun Oct 7 15:19:00 1992 Date: 07 Oct 1992 19:19 -0400 (EDT) From: "DR. LAWRENCE R. SCHMITZ" Subject: ORTEP To: chemistry@ccl.net I'm posting this for a friend. He would like to find a version of ORTEP that either: 1. runs on a Macintosh, or 2. runs on another platform but produces postscript output. Any responses can be sent directly to my friend, Gary Newton, at: newton%vaxchm.dnet@server.uga.edu Thanks! Larry From MCELWRE@cnsvax.uwec.edu Wed Oct 7 14:06:00 1992 Date: Wed, 7 Oct 1992 20:06 CST From: "NAME \"Robert E. McElwaine\"" Subject: BIOLOGICAL ALCHEMY To: chemistry@ccl.net BIOLOGICAL TRANSMUTATIONS A very simple experiment can demonstrate (PROVE) the FACT of "BIOLOGICAL TRANSMUTATIONS" (reactions like Mg + O --> Ca, Si + C --> Ca, K + H --> Ca, N2 --> CO, etc.), as described in the BOOK "Biological Transmutations" by Louis Kervran, [1972 Edition is BEST.], and in Chapter 17 of the book "THE SECRET LIFE OF PLANTS" [see Footnote] by Peter Tompkins and Christopher Bird, 1973: (1) Obtain a good sample of plant seeds, all of the same kind. [Some kinds might work better that others.] (2) Divide the sample into two groups of equal weight and number. (3) Sprout one group in distilled water on filter paper for three or four weeks. (4) Separately incinerate both groups. (5) Weigh the residue from each group. [The residue of the sprouted group will usually weigh at least SEVERAL PERCENT MORE than the other group.] (6) Analyze quantitatively the residue of each group for mineral content. [Some of the mineral atoms of the sprouted group have been TRANSMUTED into heavier mineral elements by FUSING with atoms of oxygen, hydrogen, carbon, nitrogen, etc..] BIOLOGICAL TRANSMUTATIONS occur ROUTINELY, even in our own bodies. Ingesting a source of organic silicon (silicon with carbon, such as "horsetail" extract, or radishes) can SPEED HEALING OF BROKEN BONES via the reaction Si + C --> Ca, (much faster than by merely ingesting the calcium directly). Some MINERAL DEPOSITS in the ground are formed by micro- organisms FUSING together atoms of silicon, carbon, nitrogen, oxygen, hydrogen, etc.. The two reactions Si + C <--> Ca, by micro-organisms, cause "STONE SICKNESS" in statues, building bricks, etc.. The reaction N2 --> CO, catalysed by very hot iron, creates a CARBON-MONOXIDE POISON HAZARD for welder operators and people near woodstoves (even properly sealed ones). Some bacteria can even NEUTRALIZE RADIOACTIVITY! ALL OF THESE THINGS AND MORE HAPPEN, IN SPITE OF the currently accepted "laws" of physics. Footnote: Chapters 19 and 20 are about "RADIONICS". ENTIRE BOOK is FASCINATING! UN-altered REPRODUCTION and DISSEMINATION of this IMPORTANT Information is ENCOURAGED. Robert E. McElwaine B.S., Physics and Astronomy, UW-EC