From owner-chemistry@ccl.net Sun Jan 15 02:01:01 2006 From: "Victor Anisimov victor]-[outerbanks.umaryland.edu" To: CCL Subject: CCL: Semiempirical examples for teaching: success stories Message-Id: <-30502-060115015617-19442-QTSK6M3f/1F8NgNs+FMNPg*|*server.ccl.net> X-Original-From: "Victor Anisimov" Sent to CCL by: "Victor Anisimov" [victor,+,outerbanks.umaryland.edu] Terry, The tone you use is disgusting. You must respect unmoderated status of the list. Prove that you can do better and it will help you to cool down. Victor -- Victor Anisimov, PhD Postdoctoral Fellow School of Pharmacy University of Maryland 20 Penn Street, Room 601 Baltimore, MD 21201 Sent to CCL by: Terry Frankcombe [T.Frankcombe[]chem.leidenuniv.nl] *cough* *splutter* Sorry James, but that's complete nonsense (to put it politely). Since when have semiempirical extrapolations* been taken over experimental data? Complete codswallop. A lot of the claims you make in that paper are questionable at best. What you have shown are best interpreted as FAILURES of semiempirical and DFT methods, not wins. Terry *I say extrapolations to highlight that semiempirical methods applied outside their training set are not much more than guesses. Informed guesses, but guesses nonetheless. -- Dr Terry Frankcombe LIC / Theoretical Chemistry University of Leiden Postbus 9502, 2300 RA Leiden, The Netherlands Ph. +31 71 527 4533 Fax +31 71 527 4397 From owner-chemistry@ccl.net Sun Jan 15 18:20:00 2006 From: "James J. P. Stewart mrmopac%worldnet.att.net" To: CCL Subject: CCL: Semiempirical examples for teaching: success stories Message-Id: <-30503-060115181546-1646-zLrP9BhtLs7d5nt59tzycA]_[server.ccl.net> X-Original-From: "James J. P. Stewart" Content-Type: multipart/alternative; boundary="=====================_363566140==.ALT" Date: Sun, 15 Jan 2006 16:15:16 -0700 Mime-Version: 1.0 Sent to CCL by: "James J. P. Stewart" [mrmopac_._worldnet.att.net] --=====================_363566140==.ALT Content-Type: text/plain; charset="us-ascii"; format=flowed At 02:06 PM 1/14/2006, you wrote: >Sent to CCL by: Terry Frankcombe [T.Frankcombe[]chem.leidenuniv.nl] > > >*cough* *splutter* > >Sorry James, but that's complete nonsense (to put it politely). Since when >have semiempirical extrapolations* been taken over experimental data? >Complete codswallop. A lot of the claims you make in that paper are >questionable at best. What you have shown are best interpreted as >FAILURES of >semiempirical and DFT methods, not wins. Hi, Terry: You raise an interesting point. I assume by "codswallop" you are doubtful of the validity of my assertions in the article cited. A healthy skepticism is good in the sciences, and, when expressed as concisely as you have done, warrants a reply. The paper cited, "Stewart, J.J.P., Comparison of the Accuracy of Semiempirical and some DFT Methods for predicting Heats of Formation. J. Mol. Modelling, 2004. 10: p. 6-12", contains a list of 17 compounds for which I suggest the reported heats of formation are in error. Several researchers have expressed similar misgivings, suggesting that the results of semiempirical methods are of highly questionable accuracy, and asserting that the claims in that article are extraordinary. Extraordinary claims require extraordinary evidence, so to address this point, I checked the original references for all 17 compounds. In over half of them, there were simple and obvious errors in the journal publications. In some, such as 2-tert-butyl-p-cresol, the error was simply a faulty sign (the arithmetic was correct, until the final heat of formation was reported, so what was reported as 49.5 kcal/mol should have been -49.5, quite close to the calculated -52.0.) In other cases, such as 2,6,6-trimethyl-2-cyclohexen-1-one, the formula used was incorrect (in this case, the formula used was C9H13O, instead of the correct C9H14O). These errors obviously went undetected by the authors and by the reviewers, but the reported experimental heats of formation were in error. These errors became obvious when the reported H.o.F. were compared with the results of semiempirical calculations. These examples allowed me to prepare a talk which was presented at an ACS conference. At that meeting, NIST invited me to give the same talk at Gaithersburg, which, with some trepidation, I accepted. Following that talk NIST re-defined the reference heats of formation of the compounds cited to the values reported in the J. Mol. Mod. Article. So, in answer to your question, "Since when have semiempirical extrapolations* been taken over experimental data?", the answer is: Since the National Institute of Standards and Technology updated the WebBook in 2004. But even with the evidence of errors in many of the original articles, a skeptical chemist might assert that the experimental data is intrinsically better than theoretical results and should always be preferred. To address this point, a comparison of the reported H.o.F. of 20 simple molecules with related species was done, and reported in "Stewart, J.J.P., Use of Semiempirical Methods for Detecting Anomalies in Reported Enthalpies of formation of Organic Compounds. J. Phys. Chem. Ref. Data, 2004. 33(3): p. 713-724." This paper shows that the reported H.o.F are not only inconsistent with theoretical prediction but that they are also inconsistent with other experimental results, results that are otherwise completely consistent with other thermochemical data. To summarize, there are three reasons for doubting the accuracy of some experimental reference thermochemical data: (a) Modern semiempirical methods are of good accuracy and these methods can easily and reliably be used for detecting anomalies in published thermochemical data. Many tens of such errors were found. (b) In most cases, errors can readily be seen in the original journal reporting the questioned datum. (c) When the datum in question was compared with other experimental reference data, inconsistencies were found, these inconsistencies support the suspicion that the values of the datum is indeed inaccurate. Jimmy ( () () ) .-----------------oOOo----(_)----oOOo--------------------------------------. | James J. P. Stewart | | | Stewart Computational Chemistry LLC | E-mail: mrmopac()worldnet.att.net | | 15210 Paddington Circle | 39/03/15 N, 104/49/29 W | | Colorado Springs CO 80921-2512 | | | USA .ooo0 | Phone: USA +(719) 488-9416 | | ( ) Oooo. | | .------------------------\ (----( )--------------------------------------. \_) ) / (_/ --=====================_363566140==.ALT Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable At 02:06 PM 1/14/2006, you wrote:
Sent to CCL by: Terry Frankco= mbe [T.Frankcombe[]chem.leidenuniv.nl]


*cough* *splutter*

Sorry James, but that's complete nonsense (to put it politely).  Since when
have semiempirical extrapolations* been taken over experimental data? 
Complete codswallop.  A lot of the claims you make in that paper are
questionable at best.  What you have shown are best interpreted as FAILURES of
semiempirical and DFT methods, not wins.

Hi, Terry: You raise an interesting point. I assume by "codswallop" you are doubtful of the validity of my assertions in the article cited.  A healthy skepticism is good in the sciences, and, when expressed as concisely as you have done, warrants a reply.

The paper cited, =93Stewart, J.J.P., Comparison of the Accuracy of Semiempirical and some DFT Methods for predicting Heats of Formation. J. Mol. Modelling, 2004. 10: p. 6-12=94,  contains a list of 17 compounds for which I suggest the reported heats of formation are in error. Several researchers have expressed similar misgivings, suggesting that the results of semiempirical methods are of highly questionable accuracy, and asserting that the claims in that article are extraordinary.  Extraordinary claims require extraordinary evidence, so to address this point, I checked the original references for all 17 compounds.  In over half of them, there were simple and obvious errors in the journal publications.  In some, such as 2-tert-butyl-p-cresol, the error was simply a faulty sign (the arithmetic was correct, until the final heat of formation was reported, so what was reported as 49.5 kcal/mol should have been -49.5, quite close to the calculated -52.0.) In other cases, such as 2,6,6-trimethyl-2-cyclohexen-1-one, the formula used was incorrect (in this case, the formula used was C9H13O, instead of the correct C9H14O).  These errors obviously went undetected by the authors and by the reviewers, but the reported experimental heats of formation were in error.  These errors became obvious when the reported H.o.F. were compared with the results of semiempirical calculations.

These examples allowed me to prepare a talk which was presented at an ACS conference.  At that meeting, NIST invited me to give the same talk at Gaithersburg, which, with some trepidation, I accepted.  Following that talk NIST re-defined the reference heats of formation of the compounds cited to the values reported in the J. Mol. Mod. Article.  So, in answer to your question, "Since when have semiempirical extrapolations* been taken over experimental data?", the answer is: Since the National Institute of Standards and Technology updated the WebBook in 2004.

But even with the evidence of errors in many of the original articles,  a skeptical chemist might assert that the experimental data is intrinsically better than theoretical results and should always be preferred.  To address this point, a comparison of the reported H.o.F. of 20 simple molecules with related species was done, and reported in =93Stewart, J.J.P., Use of Semiempirical Methods for Detecting Anomalies in Reported Enthalpies of formation of Organic Compounds. J. Phys. Chem. Ref. Data, 2004. 33(3): p. 713-724.=94  This paper shows that the reported H.o.F are not only inconsistent with theoretical prediction but that they are also inconsistent with other experimental results, results that are otherwise completely consistent with other thermochemical data.

To summarize, there are three reasons for doubting the accuracy of some experimental reference thermochemical data:

(a) Modern semiempirical methods are of good accuracy and these methods can easily and reliably be used for detecting anomalies in published thermochemical data.  Many tens of such errors were found.
(b) In most cases, errors can readily be seen in the original journal reporting the questioned datum. 
(c) When the datum in question was compared with other experimental reference data, inconsistencies were found, these inconsistencies support the suspicion that the values of the datum is indeed inaccurate.


Jimmy

            &nbs= p;            ( () () )
 .-----------------oOOo----(_)----oOOo---------------------------------= -----.
 | James J. P. Stewart           &nb= sp;     |            &nb= sp;            &= nbsp;          |
 | Stewart Computational Chemistry LLC | E-mail:  mrmopac()worldnet.att.net  |
 | 15210 Paddington Circle           &nbs= p; |  39/03/15 N, 104/49/29 W           |
 | Colorado Springs CO 80921-2512      |            &nb= sp;            &= nbsp;          |
 | USA            &= nbsp;      .ooo0         | Phone: USA +(719) 488-9416         |
 |           &nb= sp;           (   )   Oooo. |            &nb= sp;            &= nbsp;          |
 .------------------------\ (----(   )--------------------------------------.
            &nbs= p;            &n= bsp; \_)    ) /
            &nbs= p;            &n= bsp;       (_/ --=====================_363566140==.ALT-- From owner-chemistry@ccl.net Sun Jan 15 19:24:02 2006 From: "Brian Salter-Duke b_duke===octa4.net.au" To: CCL Subject: CCL: Randell Mills Message-Id: <-30504-060115191115-22373-zRDhu+S/BYBhzYHTE+Ufnw.:.server.ccl.net> X-Original-From: Brian Salter-Duke Content-Disposition: inline Content-Type: text/plain; charset=us-ascii Date: Mon, 16 Jan 2006 08:39:49 +1100 Mime-Version: 1.0 Sent to CCL by: Brian Salter-Duke [b_duke:-:octa4.net.au] Randell Mills is famous for coming up with the theory that the H atom can be made smaller than in its ground state and that this can be used to generate cheap power. He has raised 25 million dollars in venture capital to create BlackLight Power Inc. He calls this H an hydrino. See the Wikipedia article on it:- http://en.wikipedia.org/wiki/Hydrino_theory This is of course very controversial and almost certainly nonsense. He has also put forward calculations on atoms and molecules that seem crazy yet give numbers in close agreement with experiment from a classical description of the electron. See:- http://www.blacklightpower.com/new.shtml Does anyone know whether this work has been properly assessed by a reputable scientist? His ideas are in a web book that has about 1000 pages. It is badly written and repetitive. It has not been published in the proper scientific literature. Yet it gives good answers. Is it fraud? Is he fooling himself? Why does it give good results? I ask this because I have been working on the Wikipedia page on Computational Chemistry and another editor has raised this and wants to know whether it should be mentioned on that page. The discussion is on the talk page. I am "Bduke" on that page as a wikipedia editor. Page: http://en.wikipedia.org/wiki/Computational_chemistry Talk Page: http://en.wikipedia.org/wiki/Talk:Computational_chemistry I really do not have time to wade through this sludge in Mills' online book. I really want to know whether anyone else has done so. Regards, Brian. -- Brian Salter-Duke (Brian Duke) b_duke###octa4.net.au Post: 626 Melbourne Rd, Spotswood, VIC, 3015, Australia Phone 03-93992847. http://members.iinet.net.au/~linden1/brian/ Honorary Researcher Fellow, Dept. of Medicinal Chemistry, Monash Univ. From owner-chemistry@ccl.net Sun Jan 15 20:31:00 2006 From: "Jonathan Essex jwe1,+,soton.ac.uk" To: CCL Subject: CCL: Meeting Announcement - Structure Guided Drug Design, April 2006 Message-Id: <-30505-060115193506-31558-kZSL/ekuTsPnuIaPx8pX7A ~~ server.ccl.net> X-Original-From: "Jonathan Essex" Sent to CCL by: "Jonathan Essex" [jwe1~~soton.ac.uk] ++++++++++++++++++++++++++++ DOCKING AND SCORING IN STRUCTURE GUIDED DRUG DESIGN April 5th - 7th 2006 Molecular Graphics and Modelling Society University of Southampton, United Kingdom Registration is now open for this meeting looking at docking, scoring and free energy calculations in Structure Guided Drug Design. The conference brings together experts in the field of docking and scoring, with a particular emphasis on solvation and free energy methods. Case studies are also included so as to highlight the crucial role of these methods in drug discovery today. Plenary speakers: Ruben Abagyan, Scripps Research Institute Chris Chipot, Universite Henri Poincare Andrew Good, Bristol-Myers Squibb Bill Jorgensen, Yale University Andrew Leach, GlaxoSmithKline Hans Matter, Sanofi-Aventis Didier Rognan, Universite Louis Pasteur Marcel Verdonk, Astex Therapeutics Abstracts for contributed oral presentations are invited, and a limited number of student bursaries are available. Please see the conference website for more information (www.soton.ac.uk/~jwe1/main.html) or contact Dr Christine Richardson (c.richardson() vernalis.com) or Dr Jonathan Essex (jwe1() soton.ac.uk). Closing date for registration is 5th March, 2006 -- Dr Jonathan W. Essex Phone: +44 (0)23 8059 2794 School of Chemistry Fax: +44 (0)23 8059 3781 University of Southampton Email: j.w.essex() soton.ac.uk Highfield http://www.soton.ac.uk/~chemphys/jessex Southampton SO17 1BJ United Kingdom