From: Stavrev, Krassimir Sent: Tuesday, June 03, 1997 10:03 AM To: 'hyperchem@hyper.com' Subject: ZINDO/1 parameters for K Dear Hyperchemists, New ZINDO/1 parameters for potassium have been recently reported by Mark Plummer who's been using HyperChem to model hydrocarbons absorption in molten potassium salts, NaAlCl4, KalCl4 and Zn(AlCl4)2 catalysts. The results are published in the proceedings of the International Symposium on Advances in Catalysis at the ACS meeting in San Francisco, California last April. The parameters given below, courtesy of Mark, are shown to yield very consistent results with those in HyperChem for sodium and zinc. A) 0.876 for the s- and p-orbital Slater exponents, columns 2 and 3 in zindo_1.abp; B) 1.67 for the ionization potential of the d-orbitals, column 7 in zindo_1.abp; C) 6.44 for the s- and p-orbital bond parameters, columns 8 and 9 in zindo_1.abp Copies of the original work can be obtained from the author at Marathon Oil Company, PO Box 269, Littleton, CO 80160. Please note that Mark Plummer and Marathon Oil Company cannot be held responsible for the accuracy of these parameters as stated in the author's disclaimer. We will appreciate your feedback on these new parameters. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 From: Sandy Yates [sandy@ideafix.cps.unizar.es] Sent: Tuesday, June 03, 1997 3:39 PM To: 'hyperchem@hyper.com' Subject: HyperChem and CPUs Hi, I will be purchasing a new Computer system soon. I think I have settled for WinNT as the operating system. My question to all HyperChem people is this: With all thes CPU releases going on (AMD-K6, Pentium II, Cyrix M2) has anybody had the privalidge to compare CPUs for running HyperChem? Although, this is difficult, it would be interesting. Computer systems are difficult to compare. Maybe we could use a couple of "Standard" calculations for HyperChem to have a go at. Them submit the timings and particular system configuration. The Mathematica 3.0 people do this for interest. All the specifications, SPECftp95, WinTach..., may not bare any relation to the performance of HyperChem. This is where a set of HyperChem "standard" tests might give us something to go on, ie give use a broad idea if things are significant, or NOT. What do you think? Sandy ------------------------------------------------------------------------ Sandy Yates / Area de Mecanica de Fluidos / Centro Politecnico Superior Maria de Luna, 3 / 50015 ZARAGOZA. SPAIN Phone: Spain +34 76 761000 ext 5045. Fax: Spain +34 76 761882 Email: sandy@ideafix.cps.unizar.es From: willsd@zardoz.chem.appstate.edu Sent: Wednesday, June 04, 1997 5:50 AM To: hyperchem@hyper.com Subject: animation package I recall seeing a message posted to this list a few months ago regarding a shareware program that might be helpful in generating animations from hyperchem (or other programs). I thought that I had saved this message for when I might need the program, but I can't seem to find it, and now of course I do have a use for it. As I recall the program captured screen images at some user determined rate and saved these as an animation file (avi? quicktime?). This would be useful for sharing dynamics results with people who do not have hyperchem. Can anyone remind me of the details on this program? or Is there an archive of hyperchem listserve messages similar to the ccl archive? Thanks, Steve Williams willsd@appstate.edu From: Stavrev, Krassimir Sent: Wednesday, June 04, 1997 9:27 AM To: 'willsd@zardoz.chem.appstate.edu' Cc: 'hyperchem@hyper.com' Subject: RE: animation package Hi Steve, I am reposting below the original message by Christine Pastorek about HyperCAM. Please note that HyperChem 5 can create AVI-type movies from MD, Monte Carlo or Langevin dynamics snapshots files using Playback, see the Reference manual, pp. 252-254. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 From owner-hyperchem@hyper.com Sat Feb 15 00:24:09 1997 Date: Fri, 14 Feb 1997 10:08:14 -0800 From: pastorec@ccmail.orst.edu (Christine Pastorek) Subject: AVI format movies from HyperChem A colleague found the following shareware on the WEB. It will allow you to capture whatever is on your monitor to AVI format movie files for playback anytime. I just tried it out and it seems to work okay. For example, Start HyperCAM, setup a vibrational animation with HyperChem, and select record on HyperCAM. The output goes to a default file that you can rewind, replay, etc. To download it, set your WEB browser to: http://www.hyperionics.com/www/hypercam.htm (Win 95 only). A registered copy is $30. +-+-+-+-+-+-+-+-+-+-+-+--+-+-+-+-+-+-+-+-+-+-+-+--+-+-+-+- Dr. Christine Pastorek voice:(541)737-6732 Gilbert Hall 153 Dept. of Chemistry FAX:(541)737-2062 Oregon State University Corvallis, OR 97331 internet:pastorec@ccmail.orst.edu +-+-+-+-+-+-+-+-+-+-+-+--+-+-+-+-+-+-+-+-+-+-+-+--+-+-+-+- -----Original Message----- From: willsd@zardoz.chem.appstate.edu [SMTP:willsd@zardoz.chem.appstate.edu] Sent: Wednesday, June 04, 1997 5:50 AM To: hyperchem@hyper.com Subject: animation package I recall seeing a message posted to this list a few months ago regarding a shareware program that might be helpful in generating animations from hyperchem (or other programs). I thought that I had saved this message for when I might need the program, but I can't seem to find it, and now of course I do have a use for it. As I recall the program captured screen images at some user determined rate and saved these as an animation file (avi? quicktime?). This would be useful for sharing dynamics results with people who do not have hyperchem. Can anyone remind me of the details on this program? or Is there an archive of hyperchem listserve messages similar to the ccl archive? Thanks, Steve Williams willsd@appstate.edu From: Gerardo Burton [burton@qo.fcen.uba.ar] Sent: Wednesday, June 04, 1997 9:14 AM To: 'hyperchem@hyper.com' Subject: RE: animation package Hi the program you are looking for is HYPERCAM and makes AVI files. You can get it from Hyperionics WEB site at: http://www.hyperionics.com/ or email the author, Greg Kochaniak, at gregko@hyperionics.com They also have a screen capture utility called HYPERSNAP which is very good. Regards G. Burton Prof. Gerardo Burton e-mail: burton@qo.fcen.uba.ar Departamento de Quimica Organica Facultad de Ciencias Exactas y Naturales Universidad de Buenos Aires Tel/FAX 54 1 788-6915 ---------- From: Stephen D. Williams[SMTP:willsd@zardoz.chem.appstate.edu] Sent: Miércoles 4 de Junio de 1997 02:50 To: hyperchem@hyper.com Subject: animation package I recall seeing a message posted to this list a few months ago regarding a shareware program that might be helpful in generating animations from hyperchem (or other programs). I thought that I had saved this message for when I might need the program, but I can't seem to find it, and now of course I do have a use for it. As I recall the program captured screen images at some user determined rate and saved these as an animation file (avi? quicktime?). This would be useful for sharing dynamics results with people who do not have hyperchem. Can anyone remind me of the details on this program? or Is there an archive of hyperchem listserve messages similar to the ccl archive? Thanks, Steve Williams willsd@appstate.edu From: Stavrev, Krassimir Sent: Wednesday, June 04, 1997 12:19 PM To: 'Sandy Yates' Subject: RE: HyperChem and CPUs Hi Sandy, That's a very good question. We don't make HyperChem CPU specific. We compile the executables on Pentium machines but they run just as good on any of the processors you mentioned. Unfortunately, we don't have benchmarks on various machines yet; generally, HyperChem's performance should go up with the increase of the CPU speed. We'll appreciate feedback from our users on this subject. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: Sandy Yates [SMTP:sandy@ideafix.cps.unizar.es] Sent: Tuesday, June 03, 1997 3:39 PM To: 'hyperchem@hyper.com' Subject: HyperChem and CPUs Hi, I will be purchasing a new Computer system soon. I think I have settled for WinNT as the operating system. My question to all HyperChem people is this: With all thes CPU releases going on (AMD-K6, Pentium II, Cyrix M2) has anybody had the privalidge to compare CPUs for running HyperChem? Although, this is difficult, it would be interesting. Computer systems are difficult to compare. Maybe we could use a couple of "Standard" calculations for HyperChem to have a go at. Them submit the timings and particular system configuration. The Mathematica 3.0 people do this for interest. All the specifications, SPECftp95, WinTach..., may not bare any relation to the performance of HyperChem. This is where a set of HyperChem "standard" tests might give us something to go on, ie give use a broad idea if things are significant, or NOT. What do you think? Sandy ------------------------------------------------------------------------ Sandy Yates / Area de Mecanica de Fluidos / Centro Politecnico Superior Maria de Luna, 3 / 50015 ZARAGOZA. SPAIN Phone: Spain +34 76 761000 ext 5045. Fax: Spain +34 76 761882 Email: sandy@ideafix.cps.unizar.es From: Palocsay, Frank A [palocsfa@jmu.edu] Sent: Wednesday, June 04, 1997 11:06 AM To: Stephen D. Williams Cc: hyperchem@hyper.com Subject: Re: animation package Steve, The program was HyperCam (Hyper Camera) My student tells me it works very well. Here's the information I took off their "about" screen. Greg Kochaniak Hyperionics 3146 Chestnut Street Murrysville, PA 15668, USA e-mail: gregko@hyperionics.com http://www.hyperionics.com Hope this helps. Frank Palcosay palocsfa@jmu.edu On Wed, 4 Jun 1997 05:50:24 -0400 (EDT) "Stephen D. Williams" wrote: > I recall seeing a message posted to this list a few months ago regarding > a shareware program that might be helpful in generating animations from > hyperchem (or other programs). I thought that I had saved this message > for when I might need the program, but I can't seem to find it, and now > of course I do have a use for it. > > As I recall the program captured screen images at some user determined > rate and saved these as an animation file (avi? quicktime?). This would > be useful for sharing dynamics results with people who do not have > hyperchem. > > Can anyone remind me of the details on this program? > or > Is there an archive of hyperchem listserve messages similar to the ccl > archive? > > Thanks, > Steve Williams > willsd@appstate.edu -- Frank A. Palocsay (540) 568 6634 palocsfa@jmu.edu From: Felipe Medrano Valenzuela [fmedrano@guaymas.uson.mx] Sent: Wednesday, June 04, 1997 12:38 PM To: Stephen D. Williams Cc: hyperchem@hyper.com Subject: Re: animation package On Wed, 4 Jun 1997, Stephen D. Williams wrote: > I recall seeing a message posted to this list a few months ago regarding > a shareware program that might be helpful in generating animations from > hyperchem (or other programs). I thought that I had saved this message > for when I might need the program, but I can't seem to find it, and now > of course I do have a use for it. > > As I recall the program captured screen images at some user determined > rate and saved these as an animation file (avi? quicktime?). This would > be useful for sharing dynamics results with people who do not have > hyperchem. > > Can anyone remind me of the details on this program? > or > Is there an archive of hyperchem listserve messages similar to the ccl > archive? > > Thanks, > Steve Williams > willsd@appstate.edu > I need the same information thanks in advance Felipe Medrano fmedrano@guaymas.uson.mx From: Sandy Yates [sandy@ideafix.cps.unizar.es] Sent: Wednesday, June 04, 1997 5:06 PM To: 'hyperchem@hyper.com' Subject: HyperChem and CPU performance Hi, I was motivated to this question by the fact that with all the "performance" test that were available it really comes down to how your particular application runs, ie HyperChem. I was suprised by the interest I personally recieved to my previous message. I got 10 messages from people who wanted to know if I got any replies. This means that we, as a group, and maybe with the Web support from the Hypercube site might consider some tests that we could put some timings to for our systems that could be available for all. I am happy to collate the information and create a simply html file for posting - so as not to burden the Hypercube Inc folks. We need to invent some tests that might reveal the strenghts and weaknesses of processors. I am most interested in PC-type machines, especially, Cyrix-M2, AMD-K6 and Intel-Pentiun-II. The tests might cover simple ones within 16 Mb of RAM, and others beyond 16Mb of RAM (the disk performance for the cache will play a much bigger roll here.).. I plan to put 64Mb of SDRAM in my new machine. If anybody has some suitable tests in mind then let us share them. I guess the best way to run these tests would be to use the same file. These tests might help with the purchance of new PC. One question for Hypercube here is Will HyperChem use OpenGL or Direct3D in future releases?" Joel suggested that Hypercube was unlikely to impose special graphics recquirements on HyperChem. It would seem a shame to ignore and not utilize the new fast rendering graphics engines. Regards, Sandy ------------------------------------------------------------------------ Sandy Yates / Area de Mecanica de Fluidos / Centro Politecnico Superior Maria de Luna, 3 / 50015 ZARAGOZA. SPAIN Phone: Spain +34 76 761000 ext 5045. Fax: Spain +34 76 761882 Email: sandy@ideafix.cps.unizar.es From: jsalb@ix.netcom.com Sent: Wednesday, June 04, 1997 8:35 PM To: Sandy Yates Cc: hyperchem@hyper.com Subject: Re: HyperChem and CPU performance Sandy, While your post contains some very good ideas, I would point out that Hyperchem 5.0 and ChemPlus 1.6 have been out for months and the bugs they contain have still not been fixed. Indeed, there seems to be no evidence that any significant development is going on at Hypercube. The GUI remains far behind those offered by other computational chemistry packages. OpenGL? Direct3D? Not likely. Regards, Jesse On Wed, 4 Jun 1997 23:05:43 +0200 (MET DST), you wrote: >One question for Hypercube here is Will HyperChem use OpenGL or >Direct3D in future releases?" Joel suggested that Hypercube was unlikely >to impose special graphics recquirements on HyperChem. It would seem a >shame to ignore and not utilize the new fast rendering graphics engines. > >Regards, Sandy > >------------------------------------------------------------------------ > Sandy Yates / Area de Mecanica de Fluidos / Centro Politecnico Superior > Maria de Luna, 3 / 50015 ZARAGOZA. SPAIN > Phone: Spain +34 76 761000 ext 5045. Fax: Spain +34 76 761882 > Email: sandy@ideafix.cps.unizar.es > From: Ulrich Harttig [harttigu@ava.bcc.orst.edu] Sent: Thursday, June 05, 1997 12:31 AM To: hyperchem@hyper.com Subject: Amber94 forcefield parameters Hi, I wonder if anyone has sucessfully implemented the forcefield parameters from Amber94 in Hyperchem ? I know that it is possible to do that by generating a new parameter set. However since some atom types have been changed from Amber89 (amber3 in hyperchem) to Amber94 and I am not too familiar with the atom type rules I'd rather not tinker at this stage with the typeset. Thanks for any help. Ulrich Harttig Marine/Freshwater Biomedical Sciences Center Oregon State University, Wiegand Hall 232, Corvallis, OR, 97331 Phone: 541-737-6514 Fax: 541-737-1877 e-mail: harttigu@bcc.orst.edu From: Augusto Cesar de Camargo Neto [pateta@sysnetway.com.br] Sent: Thursday, June 05, 1997 11:21 AM To: 'hyperchem@hyper.com' Subject: RE: HyperChem and CPU performance. Looking for Web Site ? Hello Folks I have been working with HyperChem 2.0-3.0 for Windows and HyperChem 2.0 Beta for SGI. Nowadays I am working much more with WWW than Chemistry, but I am really interested in HyperChem Benchmarks. If you need some facilities to setup a Web Site to post these benchmarks, please talk to me. Best Regards. Augusto Camargo -----Original Message----- From: Sandy Yates [SMTP:sandy@ideafix.cps.unizar.es] Sent: Quarta-feira, 4 de Junho de 1997 18:06 To: 'hyperchem@hyper.com' Subject: HyperChem and CPU performance Hi, I was motivated to this question by the fact that with all the "performance" test that were available it really comes down to how your particular application runs, ie HyperChem. I was suprised by the interest I personally recieved to my previous message. I got 10 messages from people who wanted to know if I got any replies. This means that we, as a group, and maybe with the Web support from the Hypercube site might consider some tests that we could put some timings to for our systems that could be available for all. I am happy to collate the information and create a simply html file for posting - so as not to burden the Hypercube Inc folks. We need to invent some tests that might reveal the strenghts and weaknesses of processors. I am most interested in PC-type machines, especially, Cyrix-M2, AMD-K6 and Intel-Pentiun-II. The tests might cover simple ones within 16 Mb of RAM, and others beyond 16Mb of RAM (the disk performance for the cache will play a much bigger roll here.).. I plan to put 64Mb of SDRAM in my new machine. If anybody has some suitable tests in mind then let us share them. I guess the best way to run these tests would be to use the same file. These tests might help with the purchance of new PC. One question for Hypercube here is Will HyperChem use OpenGL or Direct3D in future releases?" Joel suggested that Hypercube was unlikely to impose special graphics recquirements on HyperChem. It would seem a shame to ignore and not utilize the new fast rendering graphics engines. Regards, Sandy ------------------------------------------------------------------------ Sandy Yates / Area de Mecanica de Fluidos / Centro Politecnico Superior Maria de Luna, 3 / 50015 ZARAGOZA. SPAIN Phone: Spain +34 76 761000 ext 5045. Fax: Spain +34 76 761882 Email: sandy@ideafix.cps.unizar.es From: Ostlund, Neil S., PhD [ostlund@hyper.com] Sent: Thursday, June 05, 1997 12:44 PM To: hyperchem@hyper.com Recently, Jesse Salb (jsalb@ix.netcom.com) has made comments about HyperChem that I find very misleading and somewhat objectionable. I would like to comment on them. * While your post contains some very good ideas, I would like to point out that HyperChem 5.0 and * ChemPlus 1.6 have been out for months and the bugs they contain have still not been fixed. HyperChem 5, LIKE EVERY MAJOR SOFTWARE PACKAGE, contains a certain number of bugs. The bugs in our current release seem to be quite minor, as far as we can tell, and patches available on our www site have addressed the principal ones. Like other major commercial software developers, we don't issue informal and continuous bug fixing but address known bugs in future software releases unless a bug is really catastrophic. We are aware of no catastrophic bugs in HyperChem 5. We generally are quite proud of HyperChem's robustness compared to other molecular modeling software. * Indeed, there seems to be no evidence that any significant development is going on at Hypercube. We believe HyperChem 5, our relatively recent new release, adds a host of new capability to HyperChem and we are very proud of this product. Obviously, we don't describe the details of our ongoing development and new products to the public but you can be assured that much significant development is going on at Hypercube - none of which I am sure Jesse knows or should know about. * The GUI remains far behind those offered by other computational chemistry packages. I remain convinced that our GUI is far ahead of any other manufacturer. A good GUI is primarily concerned with making it easy to do chemistry and molecular modeling not just with generating artwork. No one has features like HyperChem for dealing with drawing 3D structures, selection, basis sets, transition states, etc. A host of other major GUI features of HyperChem seem to be beyond any competitor. * OpenGL? Direct 3D? Not likely. It is a mystery to me how Jesse knows anything about our development plans much less details around whether we are adopting specific libraries such as these. One of HyperChem's great strengths in the past has been the 3D rendering capability that was available from in-house code that required only 2D graphics cards and 2D primitives. As 3D hardware primitives become available in commodity PC graphics cards and 3D libraries like OpenGL and Direct3D become mainstream, you can be sure Hypercube will be there. Neil S. Ostlund President and CEO Hypercube, Inc. 1115 NW 4th Street Gainesville, Florida 32601 Tel: (352) 371-744 Fax: (352) 371-3662 E-Mail: ostlund@hyper.com WWW: http://www.hyper.com From: J.M.MAUD [J.M.MAUD@swansea.ac.uk] Sent: Thursday, June 05, 1997 1:59 PM To: hyperchem@hyper.com Subject: ZINDO\S-SCI Dear Colleague, I am trying to understand some literature results in which oligothiophene triplet-triplet (T1-Tn) transition energies were calculated using ZINDO/S-SCI. The energies showed an extreme dependence on the number of orbitals included in the CI. The calculations were not described in great detail, and upon attempting to repeat some of them I have experienced some difficulty. Depending upon the number of orbitals allowed in the CI one or both of the following problems can occur : 1. The "lowest "T1-Tn energies are negative. Displaying the spectrum does not reveal this problem - it is necessary to look at the log. 2. The log entry which gives the CI states as expansions of the different configurations (Slater determinants) becomes "garbled" for the last few (high energy) CI states. Formatting is lost, large numbers of asterisks (************) appear together with very large coefficients (10-15 figures before the decimal point). Can anyone respond to the following comments/questions ? 1. I assume that negative transition energies have no "physical" significance. 2. Is ZINDO/S-SCI in any way appropriate for the calculation of T1-Tn energies ? Is it inappropriate ? 3. Is there a bug in the software (Hyperchem 4) ? If not a bug then simply a failure of the diagonalisation algorithm ? I have previously had similar problems when attempting the calculation of transition energies of radical cations using ZINDO/S-SCI. Thanking you in advance for any help. John M Maud Department of Chemistry University of Wales Swansea Singleton Park Swansea UK From: Stavrev, Krassimir Sent: Thursday, June 05, 1997 5:31 PM To: 'J.M.MAUD' Subject: RE: ZINDO\S-SCI Dr. Maud, CI excitations may vary largely depending on the extent of the orbital basis used in the calculations, as the number of states included in the CI matrix changes with it. For organic molecules, the ground and excited states typically converge to a certain energy disposition at 10x10 CI, meaning 10 occupied, 10 virtual orbitals. If you increase monotonically the orbital range, you should be able to observe the differences in the excitations becoming smaller and smaller. The negative energy difference is something different which I would like to examine in more detail. I would appreciate if you e-mail me a copy of the hin-file of your molecule and the CI specifications you use. What I suspect is going on, is that your lowest CI triplet state drops below the reference state and this causes the first energy to become negative. This has nothing to do with the diagonalization or bugs in the code, but can be indicative that the RHF is a bad reference for the triplet-triplet excitations you are trying to examine. I'll probably be able to tell you more when I examine your system. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: J.M.MAUD [SMTP:J.M.MAUD@swansea.ac.uk] Sent: Thursday, June 05, 1997 1:59 PM To: hyperchem@hyper.com Subject: ZINDO\S-SCI Dear Colleague, I am trying to understand some literature results in which oligothiophene triplet-triplet (T1-Tn) transition energies were calculated using ZINDO/S-SCI. The energies showed an extreme dependence on the number of orbitals included in the CI. The calculations were not described in great detail, and upon attempting to repeat some of them I have experienced some difficulty. Depending upon the number of orbitals allowed in the CI one or both of the following problems can occur : 1. The "lowest "T1-Tn energies are negative. Displaying the spectrum does not reveal this problem - it is necessary to look at the log. 2. The log entry which gives the CI states as expansions of the different configurations (Slater determinants) becomes "garbled" for the last few (high energy) CI states. Formatting is lost, large numbers of asterisks (************) appear together with very large coefficients (10-15 figures before the decimal point). Can anyone respond to the following comments/questions ? 1. I assume that negative transition energies have no "physical" significance. 2. Is ZINDO/S-SCI in any way appropriate for the calculation of T1-Tn energies ? Is it inappropriate ? 3. Is there a bug in the software (Hyperchem 4) ? If not a bug then simply a failure of the diagonalisation algorithm ? I have previously had similar problems when attempting the calculation of transition energies of radical cations using ZINDO/S-SCI. Thanking you in advance for any help. John M Maud Department of Chemistry University of Wales Swansea Singleton Park Swansea UK From: Joel Polowin [JoelP@agiss.com] Sent: Friday, June 06, 1997 11:33 AM To: 'hyperchem@hyper.com' Subject: Re: ZINDO\S-SCI John M Maud writes: > 2. The log entry which gives the CI states as expansions of the > different configurations (Slater determinants) becomes "garbled" for > the last few (high energy) CI states. Formatting is lost, large > numbers of asterisks (************) appear together with very large > coefficients (10-15 figures before the decimal point). I can't comment on the other part of your message, but this was an obscure bug (trivial to fix -- merely unnoticed for a long time!) that was corrected for HyperChem 5. If I recall correctly, it's a formatting error that appears only when the number of transitions is evenly divisible by 5, or something like that; more details should be in the archive of old mailing-list messages, which you can find on the Hypercube website. Joel joelp@agiss.com From: Paul Maz [paul@scorpio.als.orst.edu] Sent: Friday, June 06, 1997 2:49 PM To: Ostlund, Neil S., PhD Cc: hyperchem@hyper.com Subject: Re: your mail I have to side with you, Niel, I think that Hyperchem is far and away the easiest molecular modeling interface I have ever seen. Version 5 also contains a wealth of new features and it appears to me that Hypercube is moving along at a very respectable pace as far as program development is concerned. I look forward to your future developments with anticipation! On Thu, 5 Jun 1997, Ostlund, Neil S., PhD wrote: > Recently, Jesse Salb (jsalb@ix.netcom.com) has made comments about > HyperChem that I find very misleading and somewhat objectionable. I > would like to comment on them. > > * While your post contains some very good ideas, I would like to > point out that HyperChem 5.0 and > * ChemPlus 1.6 have been out for months and the bugs they contain > have still not been fixed. > > HyperChem 5, LIKE EVERY MAJOR SOFTWARE PACKAGE, contains a certain > number of bugs. The bugs in our current release seem to be quite minor, > as far as we can tell, and patches available on our www site have > addressed the principal ones. Like other major commercial software > developers, we don't issue informal and continuous bug fixing but > address known bugs in future software releases unless a bug is really > catastrophic. We are aware of no catastrophic bugs in HyperChem 5. We > generally are quite proud of HyperChem's robustness compared to other > molecular modeling software. > > * Indeed, there seems to be no evidence that any significant > development is going on at Hypercube. > > We believe HyperChem 5, our relatively recent new release, adds a host > of new capability to HyperChem and we are very proud of this product. > Obviously, we don't describe the details of our ongoing development and > new products to the public but you can be assured that much significant > development is going on at Hypercube - none of which I am sure Jesse > knows or should know about. > > * The GUI remains far behind those offered by other computational > chemistry packages. > > I remain convinced that our GUI is far ahead of any other manufacturer. > A good GUI is primarily concerned with making it easy to do chemistry > and molecular modeling not just with generating artwork. No one has > features like HyperChem for dealing with drawing 3D structures, > selection, basis sets, transition states, etc. A host of other major > GUI features of HyperChem seem to be beyond any competitor. > > * OpenGL? Direct 3D? Not likely. > > It is a mystery to me how Jesse knows anything about our development > plans much less details around whether we are adopting specific > libraries such as these. One of HyperChem's great strengths in the past > has been the 3D rendering capability that was available from in-house > code that required only 2D graphics cards and 2D primitives. As 3D > hardware primitives become available in commodity PC graphics cards and > 3D libraries like OpenGL and Direct3D become mainstream, you can be sure > Hypercube will be there. > > > Neil S. Ostlund President and CEO > Hypercube, Inc. 1115 NW 4th Street > Gainesville, Florida 32601 > Tel: (352) 371-744 Fax: (352) 371-3662 > E-Mail: ostlund@hyper.com WWW: http://www.hyper.com > From: CYCLOLAB R&D Lab. [cyclolab@compuserve.com] Sent: Monday, June 09, 1997 7:26 AM To: Stavrev, Krassimir Subject: benchmarks Hi Krassimir, I'm sending to you also what I've written to Sandy about benchmarks. The reason is that I've found another email on this topic. I you are interesting in the molecules participated in my calculations, I can send you them also. Regards, Laszlo ***************** Dear Sandy, > will be purchasing a new Computer system soon. I think I have >settled for WinNT as the operating system. >My question to all HyperChem people is this: With all thes CPU releases >going on (AMD-K6, Pentium II, Cyrix M2) has anybody had the privalidge to >compare CPUs for running HyperChem? >Although, this is difficult, it would be interesting. Computer systems >are difficult to compare. Maybe we could use a couple of "Standard" >calculations for HyperChem to have a go at. Them submit the timings and >particular system configuration. I'm currently using a Pentium-100 (90)/48 MB RAM and 300 MB of permanent Virtual RAM (90 MHz processor is run with 100 MHZ) and my mainboard is a first (or early second) generation Pentium mainboard (bus speed is <60 MHz). HyperChem 4.5 under Win'95 I made some comparisons with a workstation (IBM 6000)-MOPAC and I used glucose (C6H12O6), alpha-, beta-, and gamma-cyclodextrins (aCD=C36H60O30, bCD=C42H70O36, gCD=C48H80O80) semi empirical calculations (AM1 but PM3 seems to be a liitle bit -max 5%- faster). In () the IBM MOPAC [SE] and Gaussian'94 [ab initio] results, and in [ ] the MOPAC-Win'95 results: Single Point (but averaged from optimizations): glucose: 5 sec (3sec) [1.9 sec] aCD: 11 min (4.5 min) [2.5 min] bCD: 30 min (15 min) [7.2 min] gCD: 50 min (30 min) [12.3 min] Optimization (requires ~400 Polak-Ribiere cycles, ~1000 calculations): glucose: 25 min (15 min) [10 min] aCD: 3 days (2 days) bCD: ~7 days (7 days) gCD: ~12-14 days (21 days) Ab initio: Single Point, direct SCF (STO-3G) glucose: 120 min (6 min) (STO-3G) aCD: ~12 days (2 days) 6-31 G* glucose: 120 min (20 min) 6-31 G** glucose: 3 days saved integrals, glucose: Single point, STO-3G: 5 min Optimization, STO-3G: ~3 hrs (2 hrs) A typical MM geometry optimization (~1000-1500 atoms, explicit water and 2 gCD + 1 organic molecule) takes about 5-6 hours with mm+ point charge, 1-2 hrs with AMBER or bio+ FF. I could test some calculations on a Pentium-133 with newer mainboard and i found tat it runs ~30% faster (with older mainboards the results were worse). So, I suppose that a 200 MHz CPU with 64 MB of RAM will run nicely under NT (it requires more "working" memory fro teh OS). Regards, Laszlo From: Morten Langgaard [ml@virgil.ruc.dk] Sent: Tuesday, June 10, 1997 9:29 AM To: hyperchem@hyper.com Subject: Why using HyperChem? Dear all, During the last few days there have been some contribution to this list regarding the quality and status of HyperChem. I have a few comments I would like to add to this discussion. The subject of this mail "Why using HyperChem?" is of cause provocative but it expresses very well how I feel about HyperChem. I have been a HyperChem user since November '92 (version 2) and have bought every upgrades up to now. In addition to this, I recently selected HyperChem among many competitors to be the teaching tool for molecular modeling in our PC-lab. Why? Is it because HyperChem offers some unique methods you can't get anywhere else? No, all methods in HyperChem are, as far as I can see, "just" another implementation of well-known codes. Is it because HyperChem offers fast and authentic codes? No, - I think that I just need to mention the "New bug in MM+" question from Gerardo Burton (the April top issue) in order to illustrate that we not always are dealing with authentic methods in HyperChem. This is not necessarily bad, but in this particular case I (still) think, that Hypercube's decision about not including a VESCF term in MM+ is wrong. Let me remind you of what Ostlund once (3, Oct, 1994) answered me on this topic: "I guess the short answer is that we are not enamoured of the MMP philosophy requiring quantum mechanical computation in molecular mechanics approaches and that we think MM+ may be a better pure mechanics approach. If we felt MMP was essential we would have implemented it or if we think users now are starting to demand it we will seriously look at putting it in". (the full answer can be found in the archive). Through these discussions both Polowin and Ostlund have called attention to the "default MM+ " force field as a "better" choice. Maybe, but where is the documentation? Don't you have any references? The 10 pages in the CC manual simply explain how is works. The Dreiding force field from which the "default MM+ " originates, doesn't have the best reputation (Gundertofte et. al. J. Comp. Chem. vol. 12, 4, 429-449, 1996). In the case that the "default MM+" force field is recommendable, I think it should have its own button in the force field selection box instead of being hidden in the selection of atom type. Regarding the speed: I find, that the overall performance of HyperChem is satisfactory, but when it comes to geometry optimizations, particular with ab initio methods, the inefficient optimization algorithm becomes a disastrous bottleneck. It is my experience that HyperChem is at least 10 times slower than G94w and at least 5 times slower than PC-Spartan. (this factor can be considerably worse, when using symmetry ). I cannot imagine anybody using HyperChem for ab initio calculation (providing that you have a choice) - for anything but the smallest molecules. On this topic Polowin once (27 Jul 1995) wrote: "HyperChem is faster for single point calculations than Gaussian92. However, Gaussian 92 is faster than HyperChem for geometry optimizations, principally because HyperChem doesn't calculate second derivatives analytically. Also, the optimization algorithms in Gaussian 92 (which use second derivatives) are better than those in HyperChem (which use first derivatives). HyperChem's speed could be improved (for semi-empirical calculations as well) by introducing better optimization algorithms and analytical second derivatives. These are goals for future releases of HyperChem. Remember that this is our first ab initio offering." Two years and one release later, I guess it is fair to say that things haven't changed much - correct me if I am wrong. Is it because HyperChem offers fast and high quality graphics? Well, I am pretty satisfied with the options and the capacity of the graphics offered by HyperChem. However, it is not hard to find alternatives with somewhat more power in the graphic engine. Is it because of the hardware lock? These devices simply make me furious. I acknowledge that it is possible to make trouble free installations of these locks, but I have spend many hours on problems especially in connection with our NT-network. I don't want to list all the troubles, maybe I will make a compilation one day. Normally, I would not complain about the shipment of new releases, but I believe that Jesse Slab's comments are understandable in the actual case of version 5. As far as I remember, the release was announced in August '96, I ordered it late in September and received version 5.0 in the middle of December. Unfortunately, this version didn't worked under NT - so I had to wait for 5.01. At that time I was informed that v. 5.01 was ready but not yet available on CD. I asked Victoria Barclay if it was possible to download the patch and got the following answer " No. As a matter of security we do not post core programs on places that are easily accessible". I am very pleased to find that this nonsense policy now has changed. Finally, I received the CD with version 5.01 the 14th of February and with the help of Ross Dickson at Hypercube, I had HyperChem up and running a little week later. In general, I think that the support offered by Hypercube is excellent and I will like to thank Graham Hurst, Joel Polowin and Victoria Barclay for the time they have spend on my questions. In spite of this, I do understand if customers, including myself, sometimes gets the impression that Hypercube is a little dogmatic. The bottom line of my discussion is, that if you want to use genuine methods and want peek performance, you should use the original programs. I anticipate that HyperChem could be just as good or maybe even better than some of the original programs - so let us see some of the "significant developments". Regardless of these drawbacks, some more serious than others, I still find, that HyperChem is the best molecular modeling packages on the market for Windows based PC's. HypeChem is extremely user friendly and holds a comprehensive amount of well-integrated methods. I also agree with Ostlund that HyperChem is very robust, though that I have seen some peculiarities with version 5. HyperChem is the perfect choice for molecular modeling in a classroom. Despite all my troubles with locks and the very late release of the NT version, I don't regret the choice. Finally, I have made a short list of features, I would like to see in the next release of HyperChem (I don't think that any of them will require "significant developments"). 1. Make an undo function in connection with the drawing tool. 2. Make an option for displaying distances, angels etc. on the screen(like the query button in PCModel). 3. Make a filter for importing results from Gaussian 94. 4. Make a filter for importing CSD files. Best regards Morten =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Morten Langgaard Department of Chemistry Phone: +45 46757781 * 2471 Roskilde University (RUC) Fax: +45 46757721 POB 260, DK-4000 Roskilde, Denmark E-Mail: ml@virgil.ruc.dk =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- From: Stavrev, Krassimir Sent: Tuesday, June 10, 1997 11:59 AM To: 'Sandy Yates' Subject: RE: system (CPU's) Hi Sandy, Sounds interesting! I'll think about a few network-accessible tests, that people can use as approximate benchmarks. Your data collection might be helpful too. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: Sandy Yates [SMTP:sandy@ideafix.cps.unizar.es] Sent: Tuesday, June 10, 1997 11:21 AM To: Stavrev, Krassimir Subject: system (CPU's) Hi, I posted a message to ask if anybody had compared the performance of HC on various CPUs (Cyrix, AMD and Intel). I recieved 30+ messages requesting any information that I get. People are interested in the results!! I am happy to compile the results and post them to the group. Though, as you know these type of results need to be interpreted broadly and with caution. I had interest in the results from PC users as well as SGI and Alpha users. Waht would be useful to do is to design some easy-to-implement tests, maybe 2 or 4, that people can run on there systems. This might be a scipt and HIN file do perform a task, e.g. geometry optimization. What do you think about this? Regards, Sandy ------------------------------------------------------------------------ Sandy Yates / Area de Mecanica de Fluidos / Centro Politecnico Superior Maria de Luna, 3 / 50015 ZARAGOZA. SPAIN Phone: Spain +34 76 761000 ext 5045. Fax: Spain +34 76 761882 Email: sandy@ideafix.cps.unizar.es From: Stavrev, Krassimir Sent: Tuesday, June 10, 1997 1:55 PM To: 'Morten Langgaard' Subject: RE: Why using HyperChem? Hi Morten, I would like to thank you for your thoughtful, in-depth publication on the listserver. Most of the suggestions you made are on our wish-list and I have every reason to believe that they will be implemented in HyperChem soon. I do appreciate your concern for bringing these issues to our attention again. I just want to add to your remark on the HyperChem referencing. The support desk maintains a current list of publications that cite HyperChem in international journals. I am attaching here a copy of this list in the case Joel or Victoria didn't send you an update already. The list quotes various applications of HyperChem which might be relevant to the topics you are interested in. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 From: Laurence Lavelle [lavelle@mbi.ucla.edu] Sent: Tuesday, June 10, 1997 4:41 PM To: hyperchem@hyper.com Subject: MOLPAK I am trying to obtain a copy of MOLPAK. I don't know much about this program except that it can generate molecular packing as it occurs in the crystal lattice. Any assistance would be appreciated. Thanks Laurence Lavelle """"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""" Laurence Lavelle, Ph.D. University of California Los Angeles Laboratory of Structural Biology & Molecular Medicine Molecular Biology Institute Los Angeles, CA90095-1570, USA Email:LAVELLE@MBI.UCLA.EDU Phone:(310)206-8270 Fax:(310)825-0982 http://www.mbi.ucla.edu/people/lavelle/lavelle.html """"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""" From: Alexandre Hocquet [ahocquet@tamarugo.cec.uchile.cl] Sent: Tuesday, June 10, 1997 12:04 PM To: hyperchem@hyper.com Subject: Re: Why using HyperChem (and MM+) ? At 14:29 10/06/97 +0100, Morten Langgaard wrote about MM+ and conjugation : >Let me remind you of what Ostlund once (3, Oct, 1994) answered me on >this topic: >"I guess the short answer is that we are not enamoured of the MMP >philosophy requiring quantum mechanical computation in molecular >mechanics approaches and that we think MM+ may be a better pure >mechanics approach. If we felt MMP was essential we would have >implemented it or if we think users now are starting to demand it we >will seriously look at putting it in". (the full answer can be found >in the archive). >Through these discussions both Polowin and Ostlund have called >attention to the "default MM+ " force field as a "better" choice. >Maybe, but where is the documentation? Don't you have any references? >The 10 pages in the CC manual simply explain how is works. The >Dreiding force field from which the "default MM+ " originates, >doesn't have the best reputation (Gundertofte et. al. J. Comp. >Chem. vol. 12, 4, 429-449, 1996). I actually have submitted an (independent) evaluation of MM+ force field to the Journal of Molecular Modelling. Regarding conjugation, the "default" MM+ sometimes perform better than the "normal" MM+ (that is to say with MM2 atom types assignments), sometimes not, sometimes (heteroaromatics) both fail providing satisfactory results.To be honnest, it is normal that such simple treatment yield poorer results than the MMP method, especially on molecules for which MMP was designed and tested. This treatment of conjugation has the merit to exist at least, which is not the case in some other packages that provide MM2 without the faintest treatment of conjugation and without the faintest warning. Regarding Dreiding, it may not have the best reputation but at least it has been one of the first force fields to provide a solution to a major problem of molecular mechanics : missing parameters. If newer force fields (MM3, MMFF94) possess this utility, this was not the case with MM2, so MM+ has the merit to provide it. However, i do agree with Morten that there should be a clearer distinction between MM+ and "default" MM+...and a &%$#@ Undo command ;-) Alexandre Hocquet Laboratorio de Cristalografia Facultad de Ciencias Fisicas Universidad de Chile Blanco Encalada, 2008 Santiago Centro Chile fono : 56 2 678 45 19 fax : 56 2 696 73 59 email : ahocquet@cec.uchile.cl From: Gerardo Burton [burton@qo.fcen.uba.ar] Sent: Saturday, June 14, 1997 11:26 AM To: 'hyperchem@hyper.com' Subject: Why using Hyperchem? Hi fellows I just want to add a few personal views to the subject. I have been using Hyperchem since version 2.0 with regular upgrades (3.0, 4.5 and now 5.0). When I first came in contact with HC, it opened to me and many of my colleagues a whole world of possibilities far beyond typical PC software that was available at the time like PCModel. Implementing semiempirical methods in Windows/PC with a user friendly interface was the major breakthrough I think. Graphics were quite good for that time also (1992) The further development of HC has been reasonable but not astounding, ab initio calculations incorporated in HC 4.5 are still impractical for serious work because they are very slow, version 5 runs faster and has a significant improvement over the 16 bit versions but still the calculation algorithms appear inefficient. Version 5 has several significant improvements, specially in its graphics, they are not superb as to quality an speed but the 3D representations are really helpful and much easier to visualize than the older contour plots. Further improvement here would be "cosmetic" just to make them look nicer, as the esential part is already there. The point here is that up to version 4.5 graphics were poor (almost no changes since 1992!) and the significant improvement was only enough to make them acceptable for today's hardware and software standards. However I think that the major flaw in HC is in the optimization algorithms. The lack of second derivative methods for the semiempirical and ab initio calculations is a problem (EF helps in version 5), and the implemented conjugate gradient methods are not "the best". So I usually run semiempirical calculations with AMPAC which may need 10% or less of optimization cycles than HC to reach a minimun! Even though my Sun Sparc 20 is slower than my pentium PC (up to 4x for the same Fortran source code) AMPAC will optimize structures 5-10 times faster than HC using BFGS. I have mentioned this problem in this list several times but nothing has changed. Maybe they are working on it? (I hope) Despite these drawbacks I still think that HC is an excellent software both for teaching and research. It cannot be THE only software to use but it is surely my choice for everyday work because it is friendly, easy to use and most of all its results are reliable. For tough problems (large molecules, repeated calculations, ab initio optimizations, etc) I still rely on AMPAC and Gaussian94 for running the calculations because of their speed, but I use HC to prepare the molecules and to view results. Maybe one day we'll get a software that will do everything, in the meantime HC is a great help and I hope it will continue to evolve and keep up with current developments. It might be a good idea if we users, would send lists of the improvements we feel that are needed in order that the HC people have a better idea of where they should focus their efforts to improve HC. I would propose the following (in order of priority): "Simple improvements" UNDO button Separate selection of MM+ and "Dreiding type" MM+ Context sensitive Help "Major improvements" Inclusion of more efficient optimization algorithms (e.g. BFGS, Powell) Inclusion of pi calculations in MM+ 2nd derivative optimization algorithms generally available: Diagonalized and Full Newton-Raphson Improvement in speed of the ab-initio calculations Conformational search methods (besides doing MD) which may be used with semiempirical methods in medium sized molecules (like those implemented in AMPAC) I hope some of these will make it into future releases of HC Regards Gerardo Prof. Gerardo Burton e-mail: burton@qo.fcen.uba.ar Departamento de Quimica Organica Facultad de Ciencias Exactas y Naturales Universidad de Buenos Aires Tel/FAX 54 1 788-6915 From: Gonzalo Hernandez [ghernand@bilbo.edu.uy] Sent: Saturday, June 14, 1997 3:23 PM To: 'hyperchem@hyper.com' Subject: J constraints -- [ From: Gonzalo Hernandez * EMC.Ver #2.5.02 ] -- Hi there, I am trying to implement the use of coupling constants constraints within HyperChem using a Visual Basic application that simply add this term to the total energy via DDE connection. I can make it if the optimization is not running but when i start a geometry optimization the HC doesn't respond to the DDE. It means that i can't change the value of the energy and i can't use the HC algorithm of optimization. - There is some way to add a term to the total energy when it is optimizing? - The new versions of HC (I'm using V3.0) have some differences in the DDE possibilities? Best regards Gonzalo Hernandez -- Catedra de Quimica Farmaceutica Facultad de Quimica. Universidad de la República Av. Gral. Flores 2124 Montevideo. Uruguay Tel:(598 2) 941805 Fax:(598 2) 941906 From: Brian Williams [williams@t2.chem.umn.edu] Sent: Saturday, June 14, 1997 6:30 PM To: hyperchem@hyper.com Subject: Interest in Educational Use Following the lead of one of the earlier comments, I too would be interested in knowing of individuals who are using or would like to use HyperChem in undergraduate education. From what I have seen, most people seem to be aware of SPARTAN, and the books of useful exercises developed by Warren Hehre are a good resource for individuals such as myself who are attempting to learn enough about computational chemistry to teach our students something about it. I think something similar ought to be encouraged for HyperChem. I have heard rumors that some work is being done at Northeastern University on educational applications of HyperChem; does anyone else know anything about this? Thanks Brian Williams Dept. Chemistry Bucknell University From: Sandy Yates [sandy@ideafix.cps.unizar.es] Sent: Sunday, June 15, 1997 10:21 AM To: Brian Williams Cc: hyperchem@hyper.com Subject: Re: Interest in Educational Use On Sat, 14 Jun 1997, Brian Williams wrote: > Following the lead of one of the earlier comments, I too would be > interested in knowing of individuals who are using or would like to use > HyperChem in undergraduate education. From what I have seen, most people > seem to be aware of SPARTAN, and the books of useful exercises developed > by Warren Hehre are a good resource for individuals such as myself who are > attempting to learn enough about computational chemistry to teach our > students something about it. There some resources to be found at http://www.hyper.com/ed_news.html There are a set of tutorials specifically oreintated to HyperChem. The first tutorial: 1. Semiempirical MO Calculations - Theory and Applications for HyperChem - by Henryette Roth (http://www.sciencelearn.com/ There has been a delay for the other Tutorials. I am interested in the spectroscopic data edition. I did expect it to be released at the beginning of the year, but nothing as yet. The first one is very nice. > I have heard rumors that some work is being done at Northeastern University > on educational applications of HyperChem; does anyone else know > anything about this? NO, I don't know about this. If any knows about this them I would be interested too. I am sure informations with respect to teaching with HyperChem could be used to fatten the web pages of Hypercube Eucational Section Regrds, Sandy ------------------------------------------------------------------------ Sandy Yates / Area de Mecanica de Fluidos / Centro Politecnico Superior Maria de Luna, 3 / 50015 ZARAGOZA. SPAIN Phone: Spain +34 76 761000 ext 5045. Fax: Spain +34 76 761882 Email: sandy@ideafix.cps.unizar.es From: Alexandre Hocquet [ahocquet@tamarugo.cec.uchile.cl] Sent: Sunday, June 15, 1997 5:50 AM To: hyperchem@hyper.com Subject: Re: Interest in Educational Use At 17:30 14/06/97 -0500, Brian Williams wrote: >Following the lead of one of the earlier comments, I too would be >interested in knowing of individuals who are using or would like to use >HyperChem in undergraduate education. From what I have seen, most people >seem to be aware of SPARTAN, and the books of useful exercises developed >by Warren Hehre are a good resource for individuals such as myself who are >attempting to learn enough about computational chemistry to teach our >students something about it. Dear Brian and hyperchemists, Last ACS meeting in San Francisco, i presented a comparative overview of Hyperchem, MacSpartan and Alchemy 2000, in teaching terms. I am also in the process of reediting it for submission to the J. Chem. Educ. Spartan did a huge amount of efforts on this subject, but i think Hyperchem has two unique features that the others do not have : - The number of methods provided at each theoretical level, that allows comparison of models, which i regard as an essential feature. - The ability to alter the parameters, that is to say to peer into the black box, which again is very important in a teaching perspective. Thus, the pedagogical applications that could be designed for Hyperchem specifically should take into account these features. Examples ? Could be ... - Testing the eficiency of different molecular mechanics force fields on different families of molecules. (Sauers, J. Chem. Educ., 73, p115 Sauers, J. Chem. Educ., 68, p817 and many more...) - Designing force field parameters for inorganic molecular mechanics (Canales et al., J. Chem. Educ., 69, p21) Alexandre Hocquet Facultad de Ciencias Fisicas Universidad de Chile Blanco Encalada, 2008 Santiago Centro CHILE fono : 56 2 678 45 19 fax : 56 2 696 73 59 From: Pastorek, Christine [PastorekC@quintessence.chem.orst.edu] Sent: Sunday, June 15, 1997 7:22 PM To: 'hyperchem@hyper.com' Subject: molecular modeling in the teaching lab At Oregon State Chemistry , we have added molecular modeling and computational chemistry exercises to three of our six Integrated Labs for undergraduates . In the sophomore year, HyperChem is used to model and calculated the heat of formation for two esters that the students synthesize in the lab. They compare the calculated values for the heat of combustion to the value they measure in lab by oxygen bomb calorimetry (Expt. 7 in Shoemaker, Garland, Nibler (SGN), "Experiments in Physical Chemistry", 6th Ed., 1996). In the junior year, students do a six week project that includes library searching, synthesis and workup of a di-substituted benzophenone, photophysical measurements and some photochemistry. HyperChem is used for molecular modeling and computation of the UV spectrum, IR spectrum and the output is compared to what is actually measured in the lab. This year several students pursued optimizing the geometry by following the change in the heat of formation and used the generated MO description of the orbitals to describe the UV transitions they observed in the lab. In the senior year, abinitio methods and theory are investigated using the Gaussian program -Joe Nibler's project . Students calculate various parameters for HCl and these are compared to the experimental values they determined in the lab (SGN Expt 37). The non-majors organic class this last term (170 students) used HyperChem in small groups to build molecules and calculate and animate the IR for spice oils that they collect and characterize in the lab. > We have been using HyperChem here and there in teaching for about > three years now-first as optional exercises, now as required exercises > in the Integrated Lab. Having fast computers was a key issue. Our > students pay into a University technology fund and some (only a > little) is made available as grants for development. Our department > recently received some money from this fund to upgrade our > undergraduate computer lab-Joe Nibler was the driving force for this. > We now have 6 fast pentium pro's, four with 64 and 2 with 128 RAM-this > makes a big difference in how much we can expect students to do with > these programs. Before we were using 486's and the benzophenone > calculations would take a couple of hours each; now they take about 10 > minutes. > > There was a HyperChem users conference last year-I got some good > ideas from some info that went on line after that and from Paul Dobosh > at Mount Holyoke College as part of the PEW project > http://aug3.augsburg.edu/pkal/pew/issue_9/chemistry/macromol.html. > Paul was very gracious and past along some exercises that he was > preparing for publication. One of the most useful was how to use > molecular dynamics to make reaction movies-I don't teach organic > chemistry perse-I haven't made much use of this myself-but it is neat > stuff! > > Does anyone know if the new ver. of HyperChem can communicate > directly with a separate program like Gaussian for Windows? I know > that the HyperChem architecture is open but I'm not sure if it is this > open. > > ----------------------- > Dr. Christine Pastorek > Department of Chemistry > Oregon State University > Gilbert Hall 153 > Corvallis, OR 97331-4003 > Voice: (541) 737-6732 > FAX: (541) 737-2062 > internet: pastorekc@chem.orst.edu From: Henrik Konschin [konschin@cc.helsinki.fi] Sent: Monday, June 16, 1997 1:40 AM To: hyperchem@hyper.com Subject: molecular modelling as a teaching tool I have been following the messages about using HyperChem as a teaching tool with interest. At our laboratory we have been involved in cooperation with two secondary schools in the Helsinki area, the students of which come to us for a ca 30 h lab course during the spring semester. These students come from the three last classes of the so called gymnasium where they prepare for the students' exam (after which they apply for entrance to a university). Most of the lab course is classic "wetware chemistry" with titrations, syntheses etc etc, but, we have also incorporated an exercise in molecular modelling using HyperChem. In this exercise the pupils usually are allowed to calculate various properties of some substance they have synthesized. For example, this spring they investigated dipole moments and structures of different conformers of benzoic acid and/or acetylsalicylic acid. They used both molecular mechanics and semiempirical methods in order to see the difference between various models. We also looked at the shape and the nature of different molecular orbitals and discussed chemical reactions in the context of frontier orbital models. Please note that this may be done without resorting to "fancy" quantum chemical terms. We have also looked at different protein structures which interests many of the students because they have raed about them in both their biology and chemistry classes but seldom had the opportunity to "dive into" the structure and look at it from all angles, do mutations etc etc. The result in general? Well, these school kids have taken to molecular modelling like ducks to water. Of course many of them are quite proficient users of computers already. Some of them have done Web pages for the school and many of them know UNIX like their pocket. However, few of them have seen applications of molecular modelling as a support to the chemistry they learn. And since most of them are well versed in all sorts of computer games this is the nearest one perhaps can come - "playing" with molecules, looking for the active site of the protein instead of the lost treasure... The course evaluations afterwards have been extremely positive both with regard to the whole course and, particularly, with regard to the molecular modelling part. In fact, their school teachers have also become very interested in the possibilities of using molecular modelling. The problem for them usually is that the school budget won't allow the purchase of very expensive software, even if they have suitable computers available. I wonder if HyperChem Lite might be a solution here? Our own university students have been "indoctrinated" with molecular modelling during a special course in the subject (first lectured in 1992 by Dr. Leif Laaksonen and yours truly) where "hands-on" exercises have been done using HyperChem, but also many other modelling packages. These university level courses have been well attended and have led to a few Master's theses with a large part of molecular modelling in them. For our first year undergraduates much the same applies as for the school pupils. The first year undergraduates will use molecular modelling together with the more conventional lab exercises to acquaintance themselves with different ways (theoretical and experimental) to approach a given problem. There is an article describing our cooperation with the secondary schools in the Finnish periodical "Kemia-Kemi" (Stenman, Vauhkonen, Konschin, Kemia-Kemi, Vol.23 (1996) p.99). However, there is no special emphasis on the molecular modelling and, unfortunately (for all English-speakers) the article is written in Swedish... Hopefully this communication will help all of us realize that one may really go to the "grass roots" even with molecular modelling. Cheers HenrikK Dr. Henrik Konschin, CChem, MRSC, Department of Chemistry, Laboratory for Instruction in Swedish, University of Helsinki, P.O.Box 55, FIN-00014 University of Helsinki, Finland tel. 358-0-19140177 fax. 358-0-19140169 email:henrik.konschin@helsinki.fi From: KEVIN SUDLOW [KP-SUDLOW@wpg.uwe.ac.uk] Sent: Monday, June 16, 1997 5:45 AM To: hyperchem@hyper.com Subject: Problem I am currently investigating the compound CF3IF4 and have come up with a curious problem. I have done a geometry optimization using STO-3G. Then I started to rotate the IF4 group relative to the CF3 and do single point calculation but the strange thing is that the total energies hardly vary at all! The total energies all hover around -4755042.44 kcal/mol and when I say hover I mean within plus or minus 0.02 of a kcal. Am I missing something obvious? or is it a bug in the software? Any help would be appreciated. Kevin Sudlow. email: kp-sudlo@uwe.ac.uk Faculty of Applied Sciences University of the West of England, Bristol From: Stavrev, Krassimir Sent: Monday, June 16, 1997 8:30 AM To: 'Henrik Konschin' Subject: RE: molecular modelling as a teaching tool Hi Henrik, Thank you for your publication in the HyperChem newsgroup. The student version of HyperChem Lite we distribute with electronic manuals costs presently $69, and I believe that this price makes the software affordable even for schools on a limited budget. Note, however, that the Lite version is still available only as a standalone program, meaning that for a computer lab with, say, 30 PC's, one would still need 30 separate copies of HyperChem Lite. The mainstay (HyperChem 5.0) is presently available as a network license product. For further details and discount rates, please contact our dealers (http://www.hyper.com/dealers.html), or our sales offices at info@hyper.com. Thank you, again, for sharing your experience with us. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: Henrik Konschin [SMTP:konschin@cc.helsinki.fi] Sent: Monday, June 16, 1997 1:40 AM To: hyperchem@hyper.com Subject: molecular modelling as a teaching tool I have been following the messages about using HyperChem as a teaching tool with interest. At our laboratory we have been involved in cooperation with two secondary schools in the Helsinki area, the students of which come to us for a ca 30 h lab course during the spring semester. These students come from the three last classes of the so called gymnasium where they prepare for the students' exam (after which they apply for entrance to a university). Most of the lab course is classic "wetware chemistry" with titrations, syntheses etc etc, but, we have also incorporated an exercise in molecular modelling using HyperChem. In this exercise the pupils usually are allowed to calculate various properties of some substance they have synthesized. For example, this spring they investigated dipole moments and structures of different conformers of benzoic acid and/or acetylsalicylic acid. They used both molecular mechanics and semiempirical methods in order to see the difference between various models. We also looked at the shape and the nature of different molecular orbitals and discussed chemical reactions in the context of frontier orbital models. Please note that this may be done without resorting to "fancy" quantum chemical terms. We have also looked at different protein structures which interests many of the students because they have raed about them in both their biology and chemistry classes but seldom had the opportunity to "dive into" the structure and look at it from all angles, do mutations etc etc. The result in general? Well, these school kids have taken to molecular modelling like ducks to water. Of course many of them are quite proficient users of computers already. Some of them have done Web pages for the school and many of them know UNIX like their pocket. However, few of them have seen applications of molecular modelling as a support to the chemistry they learn. And since most of them are well versed in all sorts of computer games this is the nearest one perhaps can come - "playing" with molecules, looking for the active site of the protein instead of the lost treasure... The course evaluations afterwards have been extremely positive both with regard to the whole course and, particularly, with regard to the molecular modelling part. In fact, their school teachers have also become very interested in the possibilities of using molecular modelling. The problem for them usually is that the school budget won't allow the purchase of very expensive software, even if they have suitable computers available. I wonder if HyperChem Lite might be a solution here? Our own university students have been "indoctrinated" with molecular modelling during a special course in the subject (first lectured in 1992 by Dr. Leif Laaksonen and yours truly) where "hands-on" exercises have been done using HyperChem, but also many other modelling packages. These university level courses have been well attended and have led to a few Master's theses with a large part of molecular modelling in them. For our first year undergraduates much the same applies as for the school pupils. The first year undergraduates will use molecular modelling together with the more conventional lab exercises to acquaintance themselves with different ways (theoretical and experimental) to approach a given problem. There is an article describing our cooperation with the secondary schools in the Finnish periodical "Kemia-Kemi" (Stenman, Vauhkonen, Konschin, Kemia-Kemi, Vol.23 (1996) p.99). However, there is no special emphasis on the molecular modelling and, unfortunately (for all English-speakers) the article is written in Swedish... Hopefully this communication will help all of us realize that one may really go to the "grass roots" even with molecular modelling. Cheers HenrikK Dr. Henrik Konschin, CChem, MRSC, Department of Chemistry, Laboratory for Instruction in Swedish, University of Helsinki, P.O.Box 55, FIN-00014 University of Helsinki, Finland tel. 358-0-19140177 fax. 358-0-19140169 email:henrik.konschin@helsinki.fi From: Stavrev, Krassimir Sent: Monday, June 16, 1997 10:01 AM To: 'KEVIN SUDLOW' Cc: 'hyperchem@hyper.com' Subject: RE: Problem Hi Kevin, Rotational barriers are sometimes difficult to calculate. Typically they are of the order of few kcals/mol - close to the error limit of QM methods - show multiple minima and have characteristic constants orders of magnitude smaller than those for bending and stretching vibrations. Usually extensive basis sets and higher order quantum-mechanical methods are needed to correctly determine rotational barriers and related potential energy surfaces. A good overview on this subject is given by Uri Dinor and Arnold Hagler in "Reviews in Computational Chemistry", K. Lipkowitz and D. Boyd, Eds., VCH (1991), vol.2, pp. 129-147. For relatively small molecules, I would try ab-initio with a large basis set and would include MP2 corrections in the calculation. If RAM memory is not a problem, one may also choose direct SCF for the ab-initio calculation to possibly speed up the computation. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: KEVIN SUDLOW [SMTP:KP-SUDLOW@wpg.uwe.ac.uk] Sent: Monday, June 16, 1997 5:45 AM To: hyperchem@hyper.com Subject: Problem I am currently investigating the compound CF3IF4 and have come up with a curious problem. I have done a geometry optimization using STO-3G. Then I started to rotate the IF4 group relative to the CF3 and do single point calculation but the strange thing is that the total energies hardly vary at all! The total energies all hover around -4755042.44 kcal/mol and when I say hover I mean within plus or minus 0.02 of a kcal. Am I missing something obvious? or is it a bug in the software? Any help would be appreciated. Kevin Sudlow. email: kp-sudlo@uwe.ac.uk Faculty of Applied Sciences University of the West of England, Bristol From: David W. Craig [craig@hws.edu] Sent: Monday, June 16, 1997 8:30 AM To: hyperchem@hyper.com Subject: Re: Interest in Educational Use Here is an application to undergraduate education that we have been working on for the past to years. It will be presented at the fall ACS mtg in Las Vegas and we will be working on a JChemEd paper this summer. STRUCTURE DETERMINATION OF BIOLOGICAL MOLECULES IN THE UNDERGRADUATE CURRICULUM. D. W. Craig , M. Fountain, C. W. Larson, Department of Chemistry, Hobart and William Smith Colleges, Geneva, NY 14456 Students in Organic Structural Analysis and Biochemistry courses have collaborated in performing conformational analysis of peptides and nucleic acids using nuclear magnetic resonance spectroscopy and computational chemistry methods. The four week experiment includes (week 1) collecting 300 MHz NMR spectra and computing energy-minimized structures on assigned amino acids or nucleotides; (week 2) computing energy barriers to rotation about particular bonds in peptides and nucleotides to illustrate the global minimum problem; (week 3) Analyzing 1D proton, COSY, and NOESY spectra generating a list of distance and torsion angle restraints; and (week 4) performing restrained molecular dynamics simulations and energy minimization generating structures consistent with experimental observations. Students then compare these structures to those reported in the literature. A Computer laboratory with Xwindows connection to the NMR Sun Sparc Classic workstation made possible this laboratory series (NSF DUE-9452135 and #(?sorry, don't have this number right now) ). -- David W. Craig, Professor Department of Chemistry, Hobart and William Smith Colleges Geneva, NY 14456 Voice: 315-781-3611 FAX: 315-781-3587 http://www.hws.edu/~craig/home.htm From: CYCLOLAB R&D Lab. [cyclolab@compuserve.com] Sent: Monday, June 16, 1997 11:52 AM To: Stavrev, Krassimir Subject: re: RE: Problem Dear Krassimir, I'm sending you my answer to Kevin. I made a complete analysis of the rotational barrier of glucose with all available implementation of semiempirical and molecular mechanical methods. Are you interested in it? Some results has been appeared in the proceeding book of 8th Int. Symp. on Cyclodextrins but later I repeated my calulations with the constrained geometry optimizations and I got different results from teh "static" calculations. Have you found any solution to my script menu problem? regards, Laszlo ************letter to Kevin********** Dear Kevin, I suppose that your molecule is very small, so the energy differences will be relatively small. During the last two years I've been studying the rotational barriers of the CH2OH group of alpha-D-glucopyranose. The height of the barriers are max. about 5 Kcal/mole and I found some interesting results. E.g. molecular mechanics describe much better the rotational behaviors than semiempirical or ab initio (literature only) methods. Other interesting phenomenon was that I got (often) very different results, when I re-optimized my molecule with the Constrain option of the torsional angle. I can recommend you to optimalize your molecule using geometry constraint(s). If you are interesting about my results, I can send you the figures of my study by Zipping and UUEncoding the files. Regards, Laszlo From: Jeffrey J. Ayres [JeffAyres@worldnet.att.net] Sent: Sunday, June 15, 1997 1:52 PM To: hyperchem@hyper.com Subject: Re: Interest in Educational Use Brian Williams wrote: > > Following the lead of one of the earlier comments, I too would be > interested in knowing of individuals who are using or would like to use > HyperChem in undergraduate education. From what I have seen, most people > seem to be aware of SPARTAN, and the books of useful exercises developed > by Warren Hehre are a good resource for individuals such as myself who are > attempting to learn enough about computational chemistry to teach our > students something about it. I think something similar ought to be > encouraged for HyperChem. I have heard rumors that some work is being > done at Northeastern University on educational applications of HyperChem; > does anyone else know anything about this? Thanks > Brian Williams > Dept. Chemistry > Bucknell University Brian, At California State University we studied both Spartan and Hyperchem. The program was set up so that Spartan was introduced to the student first then Hyperchem, in different courses. Spartan was introduced in a special chemistry class Advanced topics in Organic chemistry, HyperChem in special chemistry class Advanced topics in Organic Chemistry Biological Molecular Modeling. Spartan was introduced along with several other modeling packages, a Huckel calculating program, PC CHEM (similar to Molgen), and MOPAC. The tutorial at the end of Spartan was assigned as homework. The tutorial in Hyperchem was assigned as homework, to supplement the tutorial, a tripeptide was built in both ionic forms, in the gas phase and aqueous, and their energies compared. Assignments out a book are nice, however I appreciated the tripeptide problem because the professor designed the problem himself, instead of a third person with whom I had never meet. Spartan seemed easier to use, mainly because long calculations could be interupted and restarted without any major manipulations. Plus the Spartan builder was easier to use. Hyperchem impressed me because the user could choose optimization routines and molecular dynamics. Also Hyperchem shed some light on the black box calculations occurring in computational chemistry, the manuals accompanying Hyperchem were good in introducing the mathmatical formulas used in the optimization rountines (Fletcher-Reeves, conjugate gradient), and very good expanations in energy minimization routines (MM, AM, PM3, AMBER). Jeff Ayres MS Biochemistry seeking employment in computational chemistry or Biochemistry From: benk@arai.scw.che.tohoku.ac.jp Sent: Tuesday, June 17, 1997 8:40 AM To: hyperchem@hyper.com Subject: Hyperchem and Bond-Breaking Dear Fellow - Members, I confess that I am a novice with Hyperchemas well as molecular modelling and simulation in general. I know I will have to do alot of reading to be comfortable with the various aspects involved in the interpretation of the results obtained when one performs such a simulation. Before I get into the deep end, I would be ever grateful if someone could help me with the following queries: (i) I have come up with a mechanism for glucose decomposition (mainly bond breaking and isomerization) in sub and supercritical water based on the type of products obtained. However, to add some substance to my discussion, I would like to add some supporting "evidence", if you may, on how /why the preffered bonds break to form the products I have formed, as suggested by the mechanism. I was thinking about using Hyperchem - semi-empirical methods for the molecular dynamics simulations with a specified amount of water sorrounding the glucose molecule (to immitate the solvent effect as per density requirement) while I monitor the bond length. This is because a distance exceeding the theoretical bond length suggests bond breaking (BB) and so the preference to BB can be compared. Also at the end I will plot the atomic charges of the molecule, which might hint on the preffered areas for heterolytic cleavage (a large difference in atomic charges between bonded atoms). Does this make sense in the world of Hyperchem application?? (ii) If there are any papers which have used this approach to explain reaction mechanism, particularily bond breaking, could you please send me the appropriate reference ( titles/author etc) information. Sincerely, Ben K. **************************************** Bernard M. Kabyemela, Chemical Engineering Department, Faculty of Engineering, Tohoku University, Sendai, Japan Tel & Fax/Office: +81-22-217-7246 Tel & Fax/Home: +81-22-223-0149 Email: benk@arai.scw.che.tohoku.ac.jp **************************************** From: Stavrev, Krassimir Sent: Tuesday, June 17, 1997 10:58 AM To: 'benk@arai.scw.che.tohoku.ac.jp' Subject: RE: Hyperchem and Bond-Breaking Hi Bernard, I am sending you a copy of the list of papers exploring HyperChem which we maintain at the support desk. Perhaps you will find a topic you are interested in among the papers on the list. You can plot the calculated charges (Display/Labels), or save them in an image bmp-file (Edit/Copy) using HyperChem. I would appreciate receiving a preprint of your work using HyperChem. Thank you. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: benk@arai.scw.che.tohoku.ac.jp [SMTP:benk@arai.scw.che.tohoku.ac.jp] Sent: Tuesday, June 17, 1997 8:40 AM To: hyperchem@hyper.com Subject: Hyperchem and Bond-Breaking Dear Fellow - Members, I confess that I am a novice with Hyperchemas well as molecular modelling and simulation in general. I know I will have to do alot of reading to be comfortable with the various aspects involved in the interpretation of the results obtained when one performs such a simulation. Before I get into the deep end, I would be ever grateful if someone could help me with the following queries: (i) I have come up with a mechanism for glucose decomposition (mainly bond breaking and isomerization) in sub and supercritical water based on the type of products obtained. However, to add some substance to my discussion, I would like to add some supporting "evidence", if you may, on how /why the preffered bonds break to form the products I have formed, as suggested by the mechanism. I was thinking about using Hyperchem - semi-empirical methods for the molecular dynamics simulations with a specified amount of water sorrounding the glucose molecule (to immitate the solvent effect as per density requirement) while I monitor the bond length. This is because a distance exceeding the theoretical bond length suggests bond breaking (BB) and so the preference to BB can be compared. Also at the end I will plot the atomic charges of the molecule, which might hint on the preffered areas for heterolytic cleavage (a large difference in atomic charges between bonded atoms). Does this make sense in the world of Hyperchem application?? (ii) If there are any papers which have used this approach to explain reaction mechanism, particularily bond breaking, could you please send me the appropriate reference ( titles/author etc) information. Sincerely, Ben K. **************************************** Bernard M. Kabyemela, Chemical Engineering Department, Faculty of Engineering, Tohoku University, Sendai, Japan Tel & Fax/Office: +81-22-217-7246 Tel & Fax/Home: +81-22-223-0149 Email: benk@arai.scw.che.tohoku.ac.jp **************************************** From: Paul Maz [paul@scorpio.als.orst.edu] Sent: Tuesday, June 17, 1997 7:35 PM To: hyperchem@hyper.com Subject: g94 interfacing Hyperchem community, Has anyone come across a script, program or macro of any sort that will take gaussian 94 orbital population information and convert it into a format that Hyperchem 5 can read through its import molecular properties feature? It appears that all it would involve would be reshuffling of the atomic orbital coefficients as long as one used a basis set supported by Hyperchem... P. Mazurkiewicz From: Richard R. Doyle [doyle@cc.denison.edu] Sent: Wednesday, June 18, 1997 1:38 PM To: hyperchem@hyper.com Cc: bosma@cc.denison.edu Subject: Request for assistance I am forwarding the following request for a colleague of mine who is moving to Bradley University in the fall and will be purchasing Hyperchem to run on a PC. You may reply either directly to him or to the list. I will forward replies to the list to him. Thanks. -- Richard R. Doyle doyle@cc.denison.edu Department of Chemistry Denison University From: Stavrev, Krassimir Sent: Wednesday, June 18, 1997 1:01 PM To: 'doyle@cc.denison.edu' Subject: Re: request for assistance Richard: Tell your friend to email me the input file; I'll do a test run for him on a similar machine. We usually provide free support only to registered HyperChem users, but this seems to be an interesting topic which has recently drawn a lot of attention. Note, however, that these CPU comparisons are very rough estimates, as the computer performance depends on many factors. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 From: Wayne Bosma [bosma@cc.denison.edu] Sent: Wednesday, June 18, 1997 2:01 PM To: stavrev@hyper.com Subject: Hyperchem vs. Spartan Hi Krassimir, Dick Doyle sent me a copy of your e-mail to him re. my message. I appreciate your willingness to run the benchmark. I should point out that I have a copy of Hyperchem waiting for me at Bradley; thus, I WILL be a registered user once I fill in the card (which I will do once I get my computer set up there). My "conversion" is complete, but it's always nice to reinforce one's biases. I just built trans-stilbene using Spartan's molecule builder, clicked on the "minimize" button (which runs a Sybyl MM calcn), and then submitted the job either as an AM1 or STO-3G geometry optimization. It seems to me that the best comparison would come from your did something similar in Hyperchem. I've included the Spartan input files below, in case you'd like to see them (they're separated by rows of asterisks, which I added). I'm actually more interested in how fast in runs on a reasonably modern pentium (I'll be getting a 200 MHz machine with 64 MB RAM), rather than an SGI. I opted for Hyperchem rather than PC Spartan in my new job because Hyperchem started out as a PC program (I'll be getting a Sun for my primary computation engine -- neither package will run on that); had I been buying an SGI machine, it's likely I'd have gone with Spartan (which might have been a mistake, I suppose). Since PC's are so much cheaper and easier to maintain than SGI's, a performance that's even close would put the PC platform at a significant advantage. Thanks for running the benchmark for me. Wayne ********************* OPT AM1 0 1 6 -0.174879660 0.000000000 1.984136846 6 -0.458975098 0.000000000 0.490304662 6 0.458975098 0.000000000 -0.490304662 1 -1.513387546 0.000000000 0.233468944 1 1.513387546 0.000000000 -0.233468944 6 0.174879660 0.000000000 -1.984136846 6 -0.190973099 0.000000000 -4.786446052 6 -1.112080497 0.000000000 -2.543572470 6 1.271225885 0.000000000 -2.858722056 6 1.095952689 0.000000000 -4.245518436 6 -1.293723699 0.000000000 -3.930212039 1 -2.001919757 0.000000000 -1.929056764 1 2.284556391 0.000000000 -2.469465629 1 1.959813449 0.000000000 -4.903353306 1 -2.297020407 0.000000000 -4.345979415 1 -0.332755614 0.000000000 -5.862867631 6 0.190973099 0.000000000 4.786446052 6 1.112080497 0.000000000 2.543572470 6 -1.271225885 0.000000000 2.858722056 6 -1.095952689 0.000000000 4.245518436 6 1.293723699 0.000000000 3.930212039 1 2.001919757 0.000000000 1.929056764 1 -2.284556391 0.000000000 2.469465629 1 -1.959813449 0.000000000 4.903353306 1 2.297020407 0.000000000 4.345979415 1 0.332755614 0.000000000 5.862867631 ENDCART FROZEN ENDFROZEN PI ENDPI PAIRING ENDPAIR HESSIAN 3 2 2 13 13 3 3 3 3 3 3 13 13 13 13 13 3 3 3 3 3 13 13 13 13 13 1 2 1 2 3 2 2 4 1 5 3 1 3 6 1 6 9 5 9 10 5 7 10 5 7 11 5 8 11 5 6 8 5 10 14 1 11 15 1 7 16 1 8 12 1 9 13 1 1 19 5 19 20 5 17 20 5 17 21 5 18 21 5 1 18 5 20 24 1 21 25 1 17 26 1 18 22 1 19 23 1 ENDHESS *************************************************** OPT HF DIRECT STO-3G 0 1 6 -0.174879660 0.000000000 1.984136846 6 -0.458975098 0.000000000 0.490304662 6 0.458975098 0.000000000 -0.490304662 1 -1.513387546 0.000000000 0.233468944 1 1.513387546 0.000000000 -0.233468944 6 0.174879660 0.000000000 -1.984136846 6 -0.190973099 0.000000000 -4.786446052 6 -1.112080497 0.000000000 -2.543572470 6 1.271225885 0.000000000 -2.858722056 6 1.095952689 0.000000000 -4.245518436 6 -1.293723699 0.000000000 -3.930212039 1 -2.001919757 0.000000000 -1.929056764 1 2.284556391 0.000000000 -2.469465629 1 1.959813449 0.000000000 -4.903353306 1 -2.297020407 0.000000000 -4.345979415 1 -0.332755614 0.000000000 -5.862867631 6 0.190973099 0.000000000 4.786446052 6 1.112080497 0.000000000 2.543572470 6 -1.271225885 0.000000000 2.858722056 6 -1.095952689 0.000000000 4.245518436 6 1.293723699 0.000000000 3.930212039 1 2.001919757 0.000000000 1.929056764 1 -2.284556391 0.000000000 2.469465629 1 -1.959813449 0.000000000 4.903353306 1 2.297020407 0.000000000 4.345979415 1 0.332755614 0.000000000 5.862867631 ENDCART FROZEN ENDFROZEN PI ENDPI PAIRING ENDPAIR HESSIAN 3 2 2 13 13 3 3 3 3 3 3 13 13 13 13 13 3 3 3 3 3 13 13 13 13 13 1 2 1 2 3 2 2 4 1 5 3 1 3 6 1 6 9 5 9 10 5 7 10 5 7 11 5 8 11 5 6 8 5 10 14 1 11 15 1 7 16 1 8 12 1 9 13 1 1 19 5 19 20 5 17 20 5 17 21 5 18 21 5 1 18 5 20 24 1 21 25 1 17 26 1 18 22 1 19 23 1 ENDHESS **************************************************** ____________________________________________________ |Wayne B. Bosma | Dept. of Chemistry | |Visiting Asst. Professor | Denison University | | | Granville, OH 43023 | |bosma@denison.edu | (614) 587-6491 | |_____________________________|______________________| From: Stavrev, Krassimir Sent: Wednesday, June 18, 1997 2:25 PM To: 'Wayne Bosma' Subject: RE: Hyperchem vs. Spartan Hi Wayne, Here are some APPROXIMATE tests I was preparing to add to the newsgroup archive. I'll try your input soon; will let you know. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: Wayne Bosma [SMTP:bosma@cc.denison.edu] Sent: Wednesday, June 18, 1997 2:01 PM To: stavrev@hyper.com Subject: Hyperchem vs. Spartan Hi Krassimir, Dick Doyle sent me a copy of your e-mail to him re. my message. I appreciate your willingness to run the benchmark. I should point out that I have a copy of Hyperchem waiting for me at Bradley; thus, I WILL be a registered user once I fill in the card (which I will do once I get my computer set up there). My "conversion" is complete, but it's always nice to reinforce one's biases. I just built trans-stilbene using Spartan's molecule builder, clicked on the "minimize" button (which runs a Sybyl MM calcn), and then submitted the job either as an AM1 or STO-3G geometry optimization. It seems to me that the best comparison would come from your did something similar in Hyperchem. I've included the Spartan input files below, in case you'd like to see them (they're separated by rows of asterisks, which I added). I'm actually more interested in how fast in runs on a reasonably modern pentium (I'll be getting a 200 MHz machine with 64 MB RAM), rather than an SGI. I opted for Hyperchem rather than PC Spartan in my new job because Hyperchem started out as a PC program (I'll be getting a Sun for my primary computation engine -- neither package will run on that); had I been buying an SGI machine, it's likely I'd have gone with Spartan (which might have been a mistake, I suppose). Since PC's are so much cheaper and easier to maintain than SGI's, a performance that's even close would put the PC platform at a significant advantage. Thanks for running the benchmark for me. Wayne ********************* OPT AM1 0 1 6 -0.174879660 0.000000000 1.984136846 6 -0.458975098 0.000000000 0.490304662 6 0.458975098 0.000000000 -0.490304662 1 -1.513387546 0.000000000 0.233468944 1 1.513387546 0.000000000 -0.233468944 6 0.174879660 0.000000000 -1.984136846 6 -0.190973099 0.000000000 -4.786446052 6 -1.112080497 0.000000000 -2.543572470 6 1.271225885 0.000000000 -2.858722056 6 1.095952689 0.000000000 -4.245518436 6 -1.293723699 0.000000000 -3.930212039 1 -2.001919757 0.000000000 -1.929056764 1 2.284556391 0.000000000 -2.469465629 1 1.959813449 0.000000000 -4.903353306 1 -2.297020407 0.000000000 -4.345979415 1 -0.332755614 0.000000000 -5.862867631 6 0.190973099 0.000000000 4.786446052 6 1.112080497 0.000000000 2.543572470 6 -1.271225885 0.000000000 2.858722056 6 -1.095952689 0.000000000 4.245518436 6 1.293723699 0.000000000 3.930212039 1 2.001919757 0.000000000 1.929056764 1 -2.284556391 0.000000000 2.469465629 1 -1.959813449 0.000000000 4.903353306 1 2.297020407 0.000000000 4.345979415 1 0.332755614 0.000000000 5.862867631 ENDCART FROZEN ENDFROZEN PI ENDPI PAIRING ENDPAIR HESSIAN 3 2 2 13 13 3 3 3 3 3 3 13 13 13 13 13 3 3 3 3 3 13 13 13 13 13 1 2 1 2 3 2 2 4 1 5 3 1 3 6 1 6 9 5 9 10 5 7 10 5 7 11 5 8 11 5 6 8 5 10 14 1 11 15 1 7 16 1 8 12 1 9 13 1 1 19 5 19 20 5 17 20 5 17 21 5 18 21 5 1 18 5 20 24 1 21 25 1 17 26 1 18 22 1 19 23 1 ENDHESS *************************************************** OPT HF DIRECT STO-3G 0 1 6 -0.174879660 0.000000000 1.984136846 6 -0.458975098 0.000000000 0.490304662 6 0.458975098 0.000000000 -0.490304662 1 -1.513387546 0.000000000 0.233468944 1 1.513387546 0.000000000 -0.233468944 6 0.174879660 0.000000000 -1.984136846 6 -0.190973099 0.000000000 -4.786446052 6 -1.112080497 0.000000000 -2.543572470 6 1.271225885 0.000000000 -2.858722056 6 1.095952689 0.000000000 -4.245518436 6 -1.293723699 0.000000000 -3.930212039 1 -2.001919757 0.000000000 -1.929056764 1 2.284556391 0.000000000 -2.469465629 1 1.959813449 0.000000000 -4.903353306 1 -2.297020407 0.000000000 -4.345979415 1 -0.332755614 0.000000000 -5.862867631 6 0.190973099 0.000000000 4.786446052 6 1.112080497 0.000000000 2.543572470 6 -1.271225885 0.000000000 2.858722056 6 -1.095952689 0.000000000 4.245518436 6 1.293723699 0.000000000 3.930212039 1 2.001919757 0.000000000 1.929056764 1 -2.284556391 0.000000000 2.469465629 1 -1.959813449 0.000000000 4.903353306 1 2.297020407 0.000000000 4.345979415 1 0.332755614 0.000000000 5.862867631 ENDCART FROZEN ENDFROZEN PI ENDPI PAIRING ENDPAIR HESSIAN 3 2 2 13 13 3 3 3 3 3 3 13 13 13 13 13 3 3 3 3 3 13 13 13 13 13 1 2 1 2 3 2 2 4 1 5 3 1 3 6 1 6 9 5 9 10 5 7 10 5 7 11 5 8 11 5 6 8 5 10 14 1 11 15 1 7 16 1 8 12 1 9 13 1 1 19 5 19 20 5 17 20 5 17 21 5 18 21 5 1 18 5 20 24 1 21 25 1 17 26 1 18 22 1 19 23 1 ENDHESS **************************************************** ____________________________________________________ |Wayne B. Bosma | Dept. of Chemistry | |Visiting Asst. Professor | Denison University | | | Granville, OH 43023 | |bosma@denison.edu | (614) 587-6491 | |_____________________________|______________________| From: Laurence Lavelle [lavelle@mbi.ucla.edu] Sent: Wednesday, June 18, 1997 3:13 PM To: hyperchem@hyper.com; chemistry@infomeister.osc.edu Subject: Packing in the unit cell. Hi, I have spent some time tracking down MOLPAK only to find out that it does not do what I want. Apparently Chem3D is what I need. I'd appreciate any clarification on which program I should be looking for and where to find it. Currently I am in search of a program that will generate the contents of a unit cell (and possibly n+1 adjacent cells) given the coordinates of the asymmetric unit and cell parameters as input (CCD format). After obtaining the unit cell contents I intend to import the file into HyperChem Ver 5.01. I know ChemPlus 1.6 has the feature I need but it costs $400-$500 which is currently not feasible. Thanks Laurence """"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""" Laurence Lavelle, Ph.D. University of California Los Angeles Laboratory of Structural Biology & Molecular Medicine Molecular Biology Institute Los Angeles, CA90095-1570, USA Email:LAVELLE@MBI.UCLA.EDU Phone:(310)206-8270 Fax:(310)825-0982 http://www.mbi.ucla.edu/people/lavelle/lavelle.html """"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""" From: Joel Polowin [JoelP@agiss.com] Sent: Thursday, June 19, 1997 2:26 PM To: 'benk@arai.scw.che.tohoku.ac.jp'; 'hyperchem@hyper.com' Subject: RE: Hyperchem and Bond-Breaking Bernard M. Kabyemela writes: > (i) I have come up with a mechanism for glucose > decomposition (mainly bond breaking and isomerization) in sub and > supercritical water based on the type of products obtained. However, to add > some substance to my discussion, I would like to add some supporting > "evidence", if you may, on how /why the preffered bonds break to form the > products I have formed, as suggested by the mechanism. I was thinking about > using Hyperchem - semi-empirical methods for the molecular dynamics > simulations with a specified amount of water sorrounding the glucose > molecule (to immitate the solvent effect as per density requirement) while I > monitor the bond length. This is because a distance exceeding the > theoretical bond length suggests bond breaking (BB) and so the preference to > BB can be compared. Also at the end I will plot the atomic charges of the > molecule, which might hint on the preffered areas for heterolytic cleavage > (a large difference in atomic charges between bonded atoms). Does this make > sense in the world of Hyperchem application?? I think that your approach is reasonable in principle but probably not effective in practise. The chemical system that you describe is rather large for a semi-empirical calculation, but very tiny in experimental terms. I assume that you have some data describing experimental reaction rates for this reaction -- what you need to do is to estimate an approximate reaction time (or half-life, or whatever) for a *single* molecule under these conditions. And then compare that to the picosecond time scale which is typical of molecular dynamics calculations. I suspect that you will find that the expected real-life time to do this kind of simulation will be much too long, especially since you will probably need to repeat the work enough to get a reasonable statistical distribution of results. You can, of course, increase the reaction rate by increasing the temperature of the system, or by starting the simulation with a set of atomic velocities which tend to lead to the reactions you're looking for. Obviously, either change would bias the results. A further complication is that the periodic boundary conditions of HyperChem do not work with the quantum calculations. You would need to use other methods to try to constrain your system to a limited volume. You could, for example, surround the system you want to model with a shell of water molecules, with atomic charges calculated, and then select the molecules in your system and leave the shell molecules unselected. That way, only the molecules in your system will be fully included in the calculation; the atoms in the shell molecules would only affect the selected atoms by electrostatic interactions, and they wouldn't move. Thus, they would provide a kind of physical barrier. I offer the warning that setting up this kind of system is likely to be difficult at best -- I didn't have much luck with it myself. You might be able to find some way to get the same effect with distance restraints; that would lead to a bunch of other interesting problems. Joel joelp@agiss.com From: Scott Pavola [scotty42@hotmail.com] Sent: Thursday, June 19, 1997 5:44 PM To: hyperchem@hyper.com Subject: zindos_1.abp parameters To Whoever can help: I am a teacher/assistant-researcher working at Gonaga University with Hyperchem 3.0 and am trying to model a SnP2S6 crystal element and when I attempt a Single point and electronic spec calc using ZINDO/s it does not have data for Sn or S in the zindos_?.abp parameter files. Please tell me where or how I can obtain the data. Some of it I can determine but the Slater exponents are one of the pieces sending me on a goose chase. PLEASE HELP ME! Scott Jacobson Gonzaga University scotty42@hotmail.com or sjacobso@sd271.k12.id.us or www.lchs.ml.org/physics/ --------------------------------------------------------- Get Your *Web-Based* Free Email at http://www.hotmail.com --------------------------------------------------------- From: Ostlund, Neil S., PhD [ostlund@hyper.com] Sent: Thursday, June 19, 1997 8:30 PM To: hyperchem@hyper.com Subject: Educational materials. Folks, Brian Williams recently posted a question about resource materials - books, etc. for educational use of HyperChem. His interesting posting raises some questions that are dear to my heart and I would like to make some comments and ask some questions of our users out there (Note - I may start a whole new career posting listserve messages!). It will not surprise you that I think Hypercube's products are the best molecular modeling products available for use in teaching chemistry. Unfortunately, apart from our manuals, we have spent most of our efforts developing our software rather than developing specific education materials such as the Warren Hehre books from Wavefunction, Inc. The good thing about these books is that they are just as usable with HyperChem as with Spartan. The bad thing about these books (from Hypercube's perspective) is that since they come only from Wavefunction, Inc., they tend to help their sales rather than ours. I and others at HyperChem are very interested in developing specific teaching materials and we have a number of things under way. The purpose of this mail is primarily to try and stimulate some of you out there to prepare HyperChem related materials as well. I would be quite prepared to consider having Hypercube package and distribute to the general user community well-thought-out and well-executed teaching materials that you, our users, have prepared. They are much needed by the many people using HyperChem in teaching. Food for thought. Contact me if you have developed something that we should be interested in. Regards, Neil Ostlund ------------------------------------------------------------------- Neil S. Ostlund President and CEO Hypercube, Inc. 1115 NW 4th Street Gainesville, Florida 32601 Tel: (352) 371-744 Fax: (352) 371-3662 E-Mail: ostlund@hyper.com WWW: http://www.hyper.com From: Palocsay, Frank A [palocsfa@jmu.edu] Sent: Friday, June 20, 1997 8:03 AM To: Ostlund, Neil S., PhD Cc: hyperchem@hyper.com Subject: Re: Educational materials. Felow Chemists, I agree with Dr. Ostlund, we need more (and better) resources/materials if we are going to integrate molecular modeling into our curiculum. However, I feel that the BEST, most successful materials would be those that adress the field, rather than support a single product. Both Dr.'s Ostlund and Hehre are being very short-sighted. Perhaps they could cooperate and jointly support the development of really outstanding texts and experiments. That might give Dr. Hehre time to focus on his company's products and develop a decent version of PC-Spartan. In fact, some time ago, I suggested that the best product would be a synthesis of "real" Spartan and HyperChem. Maybe the two small companies could join forces in both publication and R&D. We faculty are being told our students "need to develop team-work skills" looks like these companies, and the gentlemen running them, should consider the same stategy. They both have fine products, together they'd be a killer! Frank Palcosay On Thu, 19 Jun 1997 20:30:01 -0400 "Ostlund, Neil S., PhD" wrote: > Folks, > > Brian Williams recently posted a question about resource materials - > books, etc. for educational use of HyperChem. His interesting posting > raises some questions that are dear to my heart and I would like to make > some comments and ask some questions of our users out there (Note - I > may start a whole new career posting listserve messages!). > > It will not surprise you that I think Hypercube's products are the best > molecular modeling products available for use in teaching chemistry. > Unfortunately, apart from our manuals, we have spent most of our efforts > developing our software rather than developing specific education > materials such as the Warren Hehre books from Wavefunction, Inc. The > good thing about these books is that they are just as usable with > HyperChem as with Spartan. The bad thing about these books (from > Hypercube's perspective) is that since they come only from Wavefunction, > Inc., they tend to help their sales rather than ours. > > I and others at HyperChem are very interested in developing specific > teaching materials and we have a number of things under way. The > purpose of this mail is primarily to try and stimulate some of you out > there to prepare HyperChem related materials as well. I would be quite > prepared to consider having Hypercube package and distribute to the > general user community well-thought-out and well-executed teaching > materials that you, our users, have prepared. They are much needed by > the many people using HyperChem in teaching. Food for thought. Contact > me if you have developed something that we should be interested in. > > Regards, > > Neil Ostlund > ------------------------------------------------------------------- > > Neil S. Ostlund President and CEO > Hypercube, Inc. 1115 NW 4th Street > Gainesville, Florida 32601 > Tel: (352) 371-744 Fax: (352) 371-3662 > E-Mail: ostlund@hyper.com WWW: http://www.hyper.com -- Frank A. Palocsay (540) 568 6634 palocsfa@jmu.edu From: James Lobue [jlobue@gsvms2.cc.gasou.edu] Sent: Friday, June 20, 1997 11:06 AM To: hyperchem@hyper.com Subject: Re: Educational materials. Recently Frank Palcosay wrote: -------------- Felow Chemists, I agree with Dr. Ostlund, we need more (and better) resources/materials if we are going to integrate molecular modeling into our curiculum. However, I feel that the BEST, most successful materials would be those that adress the field, rather than support a single product. Both Dr.'s Ostlund and Hehre are being very short-sighted. Perhaps they could cooperate and jointly support the development of really outstanding texts and experiments. That might give Dr. Hehre time to focus on his company's products and develop a decent version of PC-Spartan. In fact, some time ago, I suggested that the best product would be a synthesis of "real" Spartan and HyperChem. Maybe the two small companies could join forces in both publication and R&D. We faculty are being told our students "need to develop team-work skills" looks like these companies, and the gentlemen running them, should consider the same stategy. They both have fine products, together they'd be a killer! -------------- I cannot disagree with this sentiment more strongly. I speak from pure selfishness. I want to see these two companies compete for our attention. That is the way for them "together" to provide the best product(s). If two companies join forces there is no incentive to provide the best product unless you consider pride as an effective motivation. Unfortunately pride can always be postponed, but $'s will always be the over-arching motivation. As to the issue of educational resources/materials it is essential for us to continue to use these software packages and consider in detail how to incorporate them into the classroom at every level. This modeling capability provides a unique opporunity for us to examine exactly what we teach and to consider new curricula not previously considered because of the lack of appropriate technology. This will indeed mean that we come up with lessons that are specific to a particular product, but we are working in rather uncharted waters too with respect to curriculum. It would be better for us to contribute to the dialog than to bemoan the inconveniences of a competitive market. Cheers, Jim James M. Lo Bue Assistant Professor of Chemistry Georgia Southern University P. O. Box 8064 Statesboro, GA 30460 Ph: 912-681-0398 Fax: 912-681-0699 From: Stavrev, Krassimir Sent: Friday, June 20, 1997 2:25 PM To: 'hyperchem@hyper.com' Subject: benchmarks Hi, Sandy Yates has recently brought up an interesting question regarding HyperChem's performance on different computers. To my knowledge no such study has been done so far, and given the large number of personal responses Sandy has received, we felt it might be worth to present a few CPU time estimates that might be useful to roughly estimate individual speed and expected length of HyperChem calculations. Here are some "benchmarks" on different machines. Note that these are just approximate numbers which will strongly depend on the user's configuration, software which runs simultaneously with HyperChem, possible network connections, etc. Hyperchem v. 5.01 v. 4.5 |-----------------------||------| Sample/Machine P-100 P-133 PP-200 SGI -------------------------------------------------- MM+ 11 6 4 12 Zindo/S 50 32 9 9 Ab-initio 102 43 18 25 -------------------------------------------------- MM+ stands for the CPU time in seconds needed for the MM+ geometry optimization of anthracene, anthrace.hin from HyperChem \Samples; Zindo/S is a 2x2 orbitals CI-singles calculation on anthrace.hin; Ab-initio represents a single-point, STO-3G calculation on benzene.hin, again from \Samples; CI-none, + MP2 correction. HyperChem 5.01 was used on two Windows 95 PC's with 16 Mb of RAM annotated as P-100 (Pentium 100 MHz,), and P-133 , and on a Pentium Pro-200 NT 4.0 with 64 Mb of RAM. HyperChem 4.5 has been tested on a SGI O2 machine with 180 MHz IP32 processor and 96 Mb RAM. The CPU estimates should be viewed +-2 sec. due to the inaccurate way of recording the computational time. Sandy has promised to submit a summary of the responses she obtained; perhaps it will contain additional information in this regard. Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 From: Gerardo Burton [burton@qo.fcen.uba.ar] Sent: Friday, June 20, 1997 6:50 PM To: 'hyperchem@hyper.com' Subject: Benchmarks Hi I checked some of the tests mentioned by Krassimir, on my Pentium 133 with 32 Mb RAM and some numbers came out quite different (just for the extra 16 Mb RAM?) MM+ 2.5 sec Zindo/S 3 sec (!!!!) Ab initio 25 sec (MM+ optimization was made using Newton Raphson to a 0.01 gradient) Prof. Gerardo Burton e-mail: burton@qo.fcen.uba.ar Departamento de Quimica Organica Facultad de Ciencias Exactas y Naturales Universidad de Buenos Aires Tel/FAX 54 1 788-6915 From: Stavrev, Krassimir Sent: Saturday, June 21, 1997 8:30 AM To: 'Gerardo Burton' Cc: 'hyperchem@hyper.com' Subject: RE: Benchmarks Hi Gerardo, It's possible; as I said, my numbers are just approximate figures - have you stored the results in a log-file? I used print level 0 for mechanics and 2 for quantum calculations. This may have slowed me down a bit. The MM+ gradient criterion is the same; the optimization method was Polak-Ribiere. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: Gerardo Burton [SMTP:burton@qo.fcen.uba.ar] Sent: Friday, June 20, 1997 6:50 PM To: 'hyperchem@hyper.com' Subject: Benchmarks Hi I checked some of the tests mentioned by Krassimir, on my Pentium 133 with 32 Mb RAM and some numbers came out quite different (just for the extra 16 Mb RAM?) MM+ 2.5 sec Zindo/S 3 sec (!!!!) Ab initio 25 sec (MM+ optimization was made using Newton Raphson to a 0.01 gradient) Prof. Gerardo Burton e-mail: burton@qo.fcen.uba.ar Departamento de Quimica Organica Facultad de Ciencias Exactas y Naturales Universidad de Buenos Aires Tel/FAX 54 1 788-6915 From: Jeffrey J. Ayres [JeffAyres@worldnet.att.net] Sent: Saturday, June 21, 1997 1:44 PM To: hyperchem@hyper.com Subject: Re: Educational materials. James Lobue wrote: > -------------- > > I cannot disagree with this sentiment more strongly. I > speak from pure selfishness. I want to see these two > companies compete for our attention. That is the way for > them "together" to provide the best product(s). If two > companies join forces there is no incentive to provide > the best product unless you consider pride as an effective > motivation. Unfortunately pride can always be postponed, > but $'s will always be the over-arching motivation. In other words Hyperchem and Spartan will bend over backwards to please the educators. I don't recall reading in any chemistry or any scientific texts how money is a motivating factor. If the reason you are competing is for money the only thing your goals are set to acquire money, the only thing you can see is money. Personally I want to see more than just money, improving the human species for one. If you see three hunderd fat people and then see one person slim and trim, you want to see that slim and trim person again. As a matter of fact you might even become motivated to find out why that one person is slim and trim. If you see three hunderd old people then see one young person, you want to see that young person again. It has nothing to do with money. You would want to improve the species so that you would want to see only young fit healthy people. Staying up long nights in a lab to complete projects because the patient one floor below has one week to live has nothing to do with money, rather the continuation and betterment of the human species. I think you maybe getting to close to the computer technolgy business boom and forgeting about the academics high standards of morality and pursuit of knowledge. Jeff Ayres MS Biochemistry unemployed From: Pastorek, Christine [PastorekC@quintessence.chem.orst.edu] Sent: Saturday, June 21, 1997 4:22 PM To: 'hyperchem@hyper.com' Subject: HyperChem 5.01 two pentium systems/benchmark w/trans-stilbene For trans-stilbene (proposed by Wayne Bosma, see below) I have tried the following on two different PC's : > Starting from the geo minimized MM+ structure from HyperChem with > the termination condition that the RMS gradient be 1e-4, it took 1:52 > to optimize to the same RMS gradient using AM1. The time was the same > on a P200/32 Mb RAM and on a PP200/128 Mb RAM. I was surprised at > this. > > On the PP200/128 Mb RAM, starting with geo minimized MM+ structure > from HyperChem with the termination condition that the RMS gradient be > 1e-4, Polak-Ribiere algorithm, it has been 20 hours so far into the > optimization using STO-3G and the RMS gradient is 5 e-3 . The > limiting feature here can be the size of available space on the HD. > I'm running this from a 2 GB hard drive that probably only has about 1 > GB open. The run stopped once and prompted me to empty the Trash, and > once done, it continued on. This was a very nice function-I expected > the whole thing to crash at that point, but it didn't. Nice work > HyperCube! > > No screen savers on or log files were written in either case. > > Dr. Chris Pastorek > Department of Chemistry > Oregon State University > Gilbert Hall 153 > Corvallis, OR 97331-4003 > Voice: (541) 737-6732 > FAX: (541) 737-2062 > internet: pastorekc@quintessence.chem.orst.edu > > > Wayne Bosma posted the following to hyperchem@hyper.com > > I'm going to make a switch from Spartan (on the SGI) to Hyperchem (on > the > PC), and am curious about the relative calculation speeds of the 2 > packages > on those respective platforms. Currently, I have access to Spartan > and not > Hyperchem. Realizing that Hyperchem 5 has been recently optimized for > speed, and that the faster Pentium chips approach the speed of > recently-state-of-the-art UNIX machines, I wonder if somebody would be > willing to make a couple of Hyperchem runs for me. > > For a test molecule, I've chosen trans-stilbene. I did semi-empirical > AM1 > and ab initio STO-3G geometry optimizations, starting from the (Sybyl) > MM-minimized structure each time. I didn't do any "extra" > calculations > (dipole moment, vibrations, etc.). > > The results: > > Spartan, on an SGI O2, with a 180 MHz R5000 Processor, 64 MB RAM: > > AM1 calculation: 24 seconds > STO-3G calcn: 13 minutes, 40 seconds. > > Of course, this hardware is at least twice as expensive as a 200 MHz > Pentium system. Still, I wonder if the "newly optimized" Hyperchem 5 > would > give comparable performance on a cheaper machine. > > Wayne Bosma > > > > From: jsalb@ix.netcom.com Sent: Saturday, June 21, 1997 5:07 PM To: hyperchem@hyper.com Subject: (fwd) Re: Educational materials. Krassimir, Is 6.0 going to have a spell checker? Jesse Salb On Sat, 21 Jun 1997 10:43:33 -0700, "Jeffrey J. Ayres" wrote: >In other words Hyperchem and Spartan will bend over backwards to please >the educators. I don't recall reading in any chemistry or any >scientific texts how money is a motivating factor. If the reason you >are competing is for money the only thing your goals are set to acquire >money, the only thing you can see is money. Personally I want to see >more than just money, improving the human species for one. If you see >three hunderd fat people and then see one person slim and trim, you >want to see that slim and trim person again. As a matter of fact you >might even become motivated to find out why that one person is slim and >trim. If you see three hunderd old people then see one young person, >you want to see that young person again. It has nothing to do with >money. You would want to improve the species so that you would want to >see only young fit healthy people. Staying up long nights in a lab to >complete projects because the patient one floor below has one week to >live has nothing to do with money, rather the continuation and >betterment of the human species. I think you maybe getting to close to >the computer technolgy business boom and forgeting about the academics >high standards of morality and pursuit of knowledge. > > Jeff Ayres > MS Biochemistry > unemployed > From: Christian Buning [BUNING@akcomba.oci.uni-heidelberg.de] Sent: Monday, June 23, 1997 10:03 AM To: hyperchem@hyper.com Subject: Calculation of electronic transitions with ZINDO/S Dear Sirs, I would like to use the ZINDO/S-program distributed with HyperChem 4 to calculate charge-tranfer-spectra of transition metal compounds. The only problem I have is that I do not know how to interpret the information I get with the log-file. This means: What is " transition 1" and what means " 1 -> 3" f.i. in this context? I cannot assign these labels to the transitions I have in spectroscopy. Unfortunately I could not find any hint in the manual so I would be glad if you could help me. Thanks in advance. Best regards Christian Buning From: Birgit Grossmann [grossman@ruf.uni-freiburg.de] Sent: Monday, June 23, 1997 8:51 AM To: hyperchem@hyper.com Subject: COT and binding energies? Hi :-) I tried to calculate inner reorganization energies according to Nelsens method. This involves an optimization of the neutral species and then the single point calculation of the neutral molecule conformation with charge -1. With the anion and dianion of COT this works well, giving higher (less negative) binding energies for charges not optimized for. With neutral COT, I get -1720 kcal/mol for the neutral conf./neutral charge and -1724 for neutral conf./negative charge (method AM1, RHF (PM3, UHF doesn`t do much)) What do I do wrong? Thanks Birgit Grossmann, University of Freiburg/D From: Gerardo Burton [burton@qo.fcen.uba.ar] Sent: Monday, June 23, 1997 10:19 AM To: 'Stavrev, Krassimir' Subject: RE: Benchmarks Hi Krassimir yes you are right, with the log-file on, the ZINDO calculation dropped to 25 sec. Regards Gerardo Gerardo Burton e-mail: burton@qo.fcen.uba.ar Departamento de Quimica Organica Facultad de Ciencias Exactas y Naturales Universidad de Buenos Aires Tel/FAX 54 1 788-6915 ---------- From: Stavrev, Krassimir[SMTP:stavrev@hyper.com] Sent: Sábado 21 de Junio de 1997 09:30 To: 'Gerardo Burton' Cc: 'hyperchem@hyper.com' Subject: RE: Benchmarks Hi Gerardo, It's possible; as I said, my numbers are just approximate figures - have you stored the results in a log-file? I used print level 0 for mechanics and 2 for quantum calculations. This may have slowed me down a bit. The MM+ gradient criterion is the same; the optimization method was Polak-Ribiere. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 From: Stavrev, Krassimir Sent: Monday, June 23, 1997 10:32 AM To: 'Christian Buning' Subject: RE: Calculation of electronic transitions with ZINDO/S Hi Christian, Transitions are calculated from the diagonalized CI matrix and are labeled in increasing energy; for example, trasition 1 is usually the transition from CI state #1 to CI root #2, and this is annotated as 1 -> 2. You can see which MO excitations dominate this transition from the mini-table that follows each transition. Regards, Krassimir --- Krassimir Stavrev, PhD, support@hyper.com Hypercube, Inc. Florida Science and Technology Park 1115 N.W. 4th Street Gainesville, Florida 32601 Voice: 352-371-7744/1-800-960-1871 Fax: 352-371-3662 -----Original Message----- From: Christian Buning [SMTP:BUNING@akcomba.oci.uni-heidelberg.de] Sent: Monday, June 23, 1997 10:03 AM To: hyperchem@hyper.com Subject: Calculation of electronic transitions with ZINDO/S Dear Sirs, I would like to use the ZINDO/S-program distributed with HyperChem 4 to calculate charge-tranfer-spectra of transition metal compounds. The only problem I have is that I do not know how to interpret the information I get with the log-file. This means: What is " transition 1" and what means " 1 -> 3" f.i. in this context? I cannot assign these labels to the transitions I have in spectroscopy. Unfortunately I could not find any hint in the manual so I would be glad if you could help me. Thanks in advance. Best regards Christian Buning From: Stavrev, Krassimir Sent: Monday, June 23, 1997 10:42 AM To: 'Birgit Grossmann' Subject: RE: COT and binding energies? Hi Birgit, If you add an electron or 2, you probably change the spin multiplicity from 1 to 2, and again to 1, in the case of dianion. RHF is for giving typically higher energies than those obtain