Talk:Computational chemistry

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Computer science (Rated B-class, Top-importance)
WikiProject icon This article is within the scope of WikiProject Computer science, a collaborative effort to improve the coverage of Computer science related articles on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
B-Class article B  This article has been rated as B-Class on the project's quality scale.
 Top  This article has been rated as Top-importance on the project's importance scale.
 
WikiProject Chemistry (Rated B-class, Top-importance)
WikiProject icon This article is within the scope of WikiProject Chemistry, a collaborative effort to improve the coverage of chemistry on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
B-Class article B  This article has been rated as B-Class on the project's quality scale.
 Top  This article has been rated as Top-importance on the project's importance scale.
 

Archives[edit]

Discussion to 31 dec, 2006 is archived to Talk:Computational chemistry/archive 1

Automatic Peer Review[edit]

The following suggestions were generated by a semi-automatic javascript program, and might not be applicable for the article in question.

  • Please expand the lead to conform with guidelines at Wikipedia:Lead. The article should have an appropriate number of paragraphs as is shown on WP:LEAD, and should adequately summarize the article.[?]
  • See if possible if there is a free use image that can go on the top right corner of this article.[?]
  • There may be an applicable infobox for this article. For example, see Template:Infobox Biography, Template:Infobox School, or Template:Infobox City.[?] (Note that there might not be an applicable infobox; remember that these suggestions are not generated manually)
  • Please make the spelling of English words consistent with either American or British spelling, depending upon the subject of the article. Examples include: behavior (A) (British: behaviour), aluminium (B) (American: aluminum), meter (A) (British: metre), ization (A) (British: isation), isation (B) (American: ization), modeling (A) (British: modelling), modelling (B) (American: modeling).
  • Watch for redundancies that make the article too wordy instead of being crisp and concise. (You may wish to try Tony1's redundancy exercises.)
    • Vague terms of size often are unnecessary and redundant - “some”, “a variety/number/majority of”, “several”, “a few”, “many”, “any”, and “all”. For example, “All pigs are pink, so we thought of a number of ways to turn them green.”
    • Avoid misplaced formality: “in order to/for” (-> to/for), “thereupon”, “notwithstanding”, etc.
  • As done in WP:FOOTNOTE, footnotes usually are located right after a punctuation mark (as recommended by the CMS, but not mandatory), such that there is no space in between. For example, the sun is larger than the moon [2]. is usually written as the sun is larger than the moon.[2][?]
  • Please ensure that the article has gone through a thorough copyediting so that it exemplifies some of Wikipedia's best work. See also User:Tony1/How to satisfy Criterion 1a.[?]

You may wish to browse through User:AndyZ/Suggestions for further ideas.

I carried out this automatic review to give me some ideas to improve the article. Maybe you have some ideas too. Thanks, Bduke 07:41, 11 March 2007 (UTC)

Discussions (moved here from: User talk:Bduke)[edit]

Computational chemistry (History?)[edit]

There is no history section for this article, could you write a short one or simply tell me what the start date is for this branch of chemistry? Thanks: --Sadi Carnot 02:23, 25 April 2007 (UTC)

Hi Sadi, I will have a go, but it will have to wait for about 12 days or so. I'm really busy in real life and that will force me to take a wikibreak next week. The question is a complex one. The term is later than the practice. When I started doing calculations in quantum chemistry, I thought of myself as a theoretical chemist or a quantum chemist. The term "computational chemistry" is much later, but it brought into the fold people who called themselves other things such as molecular modellers and so on. It needs to be handled carefully. --Bduke 06:30, 25 April 2007 (UTC)
Yes, thanks. Based on the history section of the computer article, I'm guessing that computational chemistry started developing in the 1970s (slowly) and then gained momentum in the 1980s? I plan to buy a basic computational chemistry textbook soon; maybe then I'll be able to contribute to the history section a bit. Later: --Sadi Carnot 14:40, 25 April 2007 (UTC)
As you know theoretical calculations in Chemistry go back to Heitler and London in 1927. The first calculations carried out on digital computers was much earlier than you think. It was in the early 1950s, when what we now call semiempirical calculations were carried out. The first ab initio calculations on diatomic molecules were in 1956 at MIT. The first polyatomic calculation using Gaussian orbitals was in the late 1950s at Cambridge. Huckel calculations, generated by computer in Berkeley and Oxford were published in 1964. The first version of GAUSSIAN (Pople's program) was 1970, but other codes - ATMOL, POLYAYTOM, IBMOL, and others - all predate GAUSSIAN. The first bibliography (200+ pages) on ab initio molecular calculations was published in 1971. The 1960s were extremely active. I myself first used a computer in 1961. Molecular mechanics started in the 60s or at least the 70s. However, none of this was called computational chemistry. That term was I think first used in the late 1980s. The series "Reviews in Computational Chemistry" going back to around 1990 have had a number of articles outlining the history of computational chemistry in different countries - USA, UK, France, Canada and Germany at least. They contain stuff about hardware and the organisation of computer centres to facilitate large numerical calculations that could well be used in some of the history of computer articles that are linked from the place you mention. Those articles are pretty poor. Geeks have no sense of history. --Bduke 21:49, 25 April 2007 (UTC)
Very nice information. It would be interesting to find out who first coined or used the term "computational chemistry" in its current sense? We'll have to keep our eyes open. Let me know when you start putting some of the above information into the article. Also, yes I agree writing computer code and writing articles are two different things. Talk soon: --Sadi Carnot 23:39, 25 April 2007 (UTC)
Hi guys. I just want to add that the Journal of Computational Chemistry was first published in 1980, so the term goes at least as early as that. I would guess that it came into use in the mid to late 1970s (but I'm too young to remember ;-). Another milestone worth mentioning is the paper of Metropolis et al in 1953 which introduced the Metropolis Monte Carlo method also widely used in computational chemistry for fluid simulations among other things. Of course, they didn't call it computational chemistry at the time. --Itub 06:14, 26 April 2007 (UTC)

Itub, good data facts. We'll have to add some of this to the history section whenever it gets started. Thanks for the input. --Sadi Carnot 07:18, 26 April 2007 (UTC)

New history stubby[edit]

I'm going to use some of the above information to start a history stubby for this article. Please feel free to clean and join in. --Sadi Carnot 07:35, 26 April 2007 (UTC)
Good start. I'll add references over the weekend. --Bduke 08:38, 26 April 2007 (UTC)
Good. I just looked in google books, and the earliest one I could find that uses the term computational chemistry is from 1970: Computers and Their Role in the Physical Sciences By Sidney Fernbach, Abraham Haskell Taub. Unfortunately google only shows tiny snippets for this book, but in one of the search results it used the term in quotation marks: 'It seems, therefore, that "computational chemistry" can finally be more and more of a reality'. [1] This suggests that the term must have been pretty new at the time. --Itub 07:57, 26 April 2007 (UTC)

I would want to have a good look at that book. I do not recall it and I am sure it had little influence on the field. One place to start is the preface of the first volume of "Reviews in Computational Chemistry", VCH, around 1990. Editors are Kenny Lipkowitz and Donald Boyd. It is not labelled as volume 1, but it started a series and the next was labelled volume 2. That preface discusses a definition of "computational chemistry" certainly and it may discuss the history of the term. I forget and do not have a copy. --Bduke 08:33, 26 April 2007 (UTC)

Yes, I agree it doesn't seem to be influential. Probably many people used the words "computational chemistry" in various random occasions but it didn't catch on. Reviews in Computational Chemistry is now available online, and I think the front matter of each volume is available for free. Check out http://www3.interscience.wiley.com/cgi-bin/bookhome/114034476 . They say that "it began emerging as a distinct discipline about 10 years ago" (this was written in 1989). --Itub 09:05, 26 April 2007 (UTC)
Good data, I’ll add this to the article. If someone finds older or better references please feel free to change them. I will also reason that Pauling and Wilson’s 1935 Introduction to Quantum Mechanics – with Applications to Chemistry and Heitler’s 1945 Elementary Wave Mechanics – with Applications to Quantum Chemistry were influential to the early development of computational chemistry? If I am wrong please feel free to change my edits. Talk later: --Sadi Carnot 17:31, 26 April 2007 (UTC)

These two might help to get a little bit more of the computational stonage.--Stone 15:18, 2 May 2007 (UTC)

"Computational quantum chemistry" is not the same as "Computational chemistry" and may well predate the latter use for the wider field. I do not think that the Robinson book used the term with its modern meaning. I've been away. I will try to look at this over the weekend. --Bduke 09:20, 4 May 2007 (UTC)
I agree. It is hard for us to really decide when computational chemistry began by looking at the original literature, because it might border on original research. What we can say is what other people say about when it began (as in the intro of vol. 1 of Reviews in Computational Chemistry), or plain facts such as the publication dates of journals. However, the articles provided by Stone are certainly interesting and provide insight into the "stone age" or pre-history of computational chemistry. Given what we have, I believe that the best date is somewhere in the late 1970s. --Itub 09:33, 4 May 2007 (UTC)

Huckel calculations[edit]

The article says "In 1964, Hückel method calculations, which are a simple LCAO method for the determination of electron energies of molecular orbitals of π electrons in conjugated hydrocarbon systems, such as ethene, benzene and butadiene, were generated on computers at Berkeley and Oxford". I don't have access to the reference provided. Although I remember seeing one of those old books that consisted solely of Huckel calculation results (basically computer printouts), I don't think that ethene, benzene, and butadiene are good examples when talking about calculations done in the sixties. Those molecules can be solved with paper and pencil on the back of an envelope, so I imagine that they were done around 1930! For larger molecules the computer becomes more important. :-) --Itub 08:03, 5 May 2007 (UTC)

Yes, you are right. The references list a very large number of molecules. I have added what I think is the largest, ovalene. I have added some more references. However, this section on history is too narrow. It only deals with computational quantum chemistry. It needs more, such a mention of molecular mechanics and Allinger. --Bduke 09:13, 5 May 2007 (UTC)

Another interesting reference[edit]

There's a book called Theory and Applications of Computational Chemistry: The First Forty Years (ISBN 0444517197). I don't have it and it's insanely expensive, but you can read most of the first chapter on Amazon for free. They obviously consider a much earlier date for "computational chemistry" than Reviews in Computational Chemistry (ca 1965 vs ca 1979). But that's almost always the case with new fields: by the time the name becomes popular and journals appear, people have usually been working "in the field" for decades without noticing. ;-) --Itub 15:41, 22 May 2007 (UTC)

Interesting and as you say insanely expensive. I found I could read the first 3 pages of each chapter. Those from chapter 1 are full of errors. They talk about the first ab initio calculations in the mid 1960s. No, they were in the 1950s. They say that 40 years ago, the only high level language was FORTRAN IV. No, the early Boys work was not coded in FORTRAN. I used Mercury autocode, Elliott autocode and Algol 60 from 1960 onwards and did not learn Fortran until about 1966. It points to 1962 and the start of QCPE as the start of computational chemistry, but says nothing about whether the term was used then. I do not recall it being used by QCPE, which is after all, the Quantum Chemistry Program Exchange. Quantum Chemistry was the term used then. Then Molecular Mechanics was introduced and that term used. The collective term for both and other methods, Computational Chemistry, was much later. They are really just talking about when calculations that we now call computational chemistry were were carried out and are saying nothing about the use of the term. The first 3 pages of chapter 6 by Clementi are better history, but he does not use the term computational chemistry once (at a quick read). Looks like a useful book though. On a different matter, I have "Quantum Chemistry: the development of ab initio methods in molecular electronic structure theory" by Henry F, Schaefer III, published in 1984. It gives a brief summary of many key papers in the field. He now widely uses the term computational chemistry, but he does not use the term in that book. I think the term computational chemistry was not used, at least widely, until after that date, although the term computational quantum chemistry was used. --Bduke 00:49, 23 May 2007 (UTC)

I have just realized that I have on my shelves a book entitled "Computational Chemistry" by A. C. Norris published in 1981 by John Wiley. The interesting thing is that none of the book is about what we now call computational chemistry. The sub-title is "An introduction to numerical methods". The blurb on the back starts "This book provides a practical introduction to numerical methods at a level suitable for undergraduate chemists". This, I think, supports the view that the modern usage of the term was not in use prior to 1980, when the author signed off on the preface. --Bduke 00:47, 24 May 2007 (UTC)

That is pretty good evidence. I think we could also add some of the history of textbooks in computational chemistry in the modern sense. I'm not sure which one was first, but it was possibly published after 1990. The real boom came after 2000. Two of the earlier books that I remember are Leach's Molecular Modelling (1996) and Jensen's Introduction to Computational Chemistry (1999). Leach doesn't use "computational chemistry" in the title, but the scope is basically the same as other computational chemistry books. But maybe I'm too young and don't know of much earlier books. ;-) --Itub 05:34, 24 May 2007 (UTC)

accuracy[edit]

This is merely a qualitative discipline which can guess orders of magnitude at best. Errors of 1% are the most accurate they can go, and 40%-50% errors would not be considered bad.—Preceding unsigned comment added by 137.205.132.170 (talkcontribs)

Your statement is so vague as to be completely meaningless. Percent of what? Calculating what? For molecular energies, errors much smaller than 1% are the rule. For heats of formation, errors of about 1 kcal/mol are common for small molecules, but for this type of measure it makes no sense to talk about percentages. For equilibrium constants, errors are much larger than 1%, and are often better measured in terms of orders of magnitude. For yet other observables the errors will be different. It makes absolutely no sense to talk about percent errors without reference to the observable and system size in question. --Itub 13:14, 6 June 2007 (UTC)
A real revert war while I slept! I entirely agree with Itub. Coulson once said that we are trying to calculate the mass of the captain of the Queen Mary by measuring the displacement of the liner with the captain on the bridge and then with him on the dock. We can get 1% in the total, but it is'nt good enough. As chemists we need energy differences. I would also point out that 1% in a bond length is 1 - 2 pm. We can do that fairly easily for small molecules. The current wording is fine. --Bduke 01:05, 7 June 2007 (UTC)
its a school example of weasel words. i had in mind energies, and looking just what has been acomplished for a simplest helium 2-electron system (see computational chemistry wiki) where errors are atrocious 1-4%, I would say that its rather a qualitative science. If you have another data, please put links. "very high accuracy" is weasel word and is a misleading lie, as errors of 1-10% are far from high accuracy. The percentage error needs to be quoted to quantify the statement. For energy levels of simple molecules, for chemical potentials etc - quote of percetntages in necessary. And of course, saying that for heats of formation talk of percentages is impossible is nonsense, as there are absolute values to compared with, as with any measurable quantity. Experimental results always come with relative errors, and it is a first and foremost question how does theory compare - and in this case, theory is next to useless.

I think you need to look at sources other than the computational chemistry wiki. First, all energies of helium have been calculated to within experimental error and indeed are most certainly known to far more decimal digits (probably 10 or more digits) from theory than from experiment. Indeed those calculations go back to the 1930s. Secondly, various good methods based on high order coupled cluster methods can give energies of formation to within experimental error. Such methods are the W2 and W3 methods and those using CCSDTQ with accurate relativistic corrections. Indeed, I have heard it argued that as the number of experimentalists who measure thermodynamic properties is decreasing (the field seems to now longer attract people) we are going to have to rely on theoretical calculations for such data. The person who made that remark is one of the people who is meeting that demand for high accurate calculations. Thirdly for small molecules, there are now several examples where theoretical predictions of bond lengths and bond angles forced the experimentalists to look again at their data. The earliest case is perhaps methylene, but there have been many larger examples more recently than that work. The scope of accurate calculations is rapidly moving to larger systems. --Bduke 03:10, 7 June 2007 (UTC)

I like Bduke's recent edit, and I'm all for adding more details for the article about the accuracy for predicting specific properties, as some are harder than others. To clarify my statement about heats of formation, it's really not useful to give the error in percentage for two reasons: 1) a heat of formation is a difference, which can be zero. If you calculate that the heat of formation of H2 is 0.0000001 kcal/mol, instead of the correct 0 kcal/mol, what is the percent error? Infinite? 2) What matters for heats of formation is having the absolute error within "chemical accuracy"; that is, within what has a noticeable effect on experimentally observable properties. In many cases an error of about 1 kcal/mol is as good or better than experiment, and is good enough for predicting other properties. --Itub 06:17, 7 June 2007 (UTC)

On another accuracy related matter: just a little query about the mention of accuracy in the introduction. It says that ab initio calculations are the most accurate. I was under the impression that semi-empirical calculations could be as good as ab initio for molecules similiar to the basis set, and that the advantage of ab initio came with its ability to deal with 'exotic' molecules. i'm not an expert, but is the intro a bit misleading?Brokencalculator 10:29, 15 November 2007 (UTC)

That is a good point. The question is that semi-empirical can give accurate results but you can not rely on them doing so. With ab initio you have a better view of whether it is going to be accurate. --Bduke 10:52, 15 November 2007 (UTC)

GA Re-review[edit]

As part of classifying GAs in WP:UCGA, I noticed that how this article came to get a GA tag is a bit weird (it was done by anon IP directly involved with editing of the article over a year ago, it seems). I know you've got a scientific review so the information seems ok, but the style and writing in the article may need to be fixed up some for purposes of a GA re-review. --Masem 04:11, 30 July 2007 (UTC)

I think it became classified as a good article before I started edited WP at the end of October, 2005. I certainly did not think it was that good at that time, but thanks to several people it has become a lot better. I have been intending to put it forward for a GA re-view, but have not got around to it. Could you, or User:Giggy who added cleanup tags to the article, give us any suggestions about what does need to be fixed up? Thanks. --Bduke 07:32, 30 July 2007 (UTC)

This comment from Masem and the cleanup tag on the article has prompted me to do some more work on the article. Could others join me and let us see whether we can can get it really to GA status? In particular we need more examples, I think, and the sections on "Interpreting molecular wave functions" and "Chemical dynamics" in particular appear to need a lot of attention. Other things that have come to my mind are a brief mention of QSPR, QSAR and chemical databases in the lead and more detail further down. The list in the introduction of five major areas is not reflected in the content either below or above in the lead. Please either be bold and just edit or make suggestions here. --Bduke 09:21, 30 July 2007 (UTC)

After all these years on Wikipedia, I still have no idea what a "good article" is, because the criteria seem to be shifting, subjective, and arbitrary. So I don't really care about "good articles". I care more about featured articles because the criteria and process seems more transparent to me. In order to become featured, an article has to be comprehensive and balanced, which is not the case with here. It is too focused on quantum mechanics, with barely a mention of the classical methods. Besides the sections and topics you mention, here are other important omissions:
  • Monte Carlo is just a "see also" link.
  • The section on software is only about QM software
  • No mention of QM/MM or free energy calculations
  • Very little on classical MD; no mention of protein simulations, solvation, solvent models
  • No link to the molecular orbital article!
  • Not much mention of conformational analysis or reaction coordinate calculations
  • No mention of computational thermochemistry
These are many topics, to be sure. To be balanced and of reasonable length, the article whould all have to be written in "summary style", which might require making some sections shorter and moving the acetylene example to a more specific article. I'm also tempted to remove the table with software packages. --Itub 10:01, 30 July 2007 (UTC)
Lots of good points there. I suspect that this article is always going to be too technical to get to FA but we could try. Just a few minor quibbles about your 7 dot points, I hope to help us all to learn how to address them.
  • Point 2 - the table is supposed to be about codes that match 2 criteria. None have only HF and post HF, and it includes MM, so the 2 has to at least include one of MM, semi-empirical or DFT. I agree we should move it to a separate article like the other software lists. What about Quantum chemistry computer programs for a title?
  • Point 3. The table of software has but it is not obvious - Tinker with Gamess(US) and ONIOM in Gaussian, It needs something in the text.
  • Point 5. The MO article is terrible. I prefer the links to be articles one or two down the tree such as Hartree-Fock, but they are largely missing too. I'll look at that.
  • Point 7. I wrote an article on the composite methods the other day. It needs expanding. It is linked from Post-HF but it could also be a link from a section here.
Most of your points are areas I do not know too much about. Can you help out on them? --Bduke 10:38, 30 July 2007 (UTC)

I have copied this from the talk page of User:Masem, where he responded to my plea for help in much the same words as my reply to him above:

"For some, take a look at what some of the other editors have said over at Good Article Review. You do need a lot more in-line references (doesn't have to be new references if the existing ones adequately state the necessary details, use the (ref name="") tags to repeat them, but you do need more); make sure to properly format references using the correct WP:CITET; the article structure is a bit in question - I think you can move History to be the first section, and then have "Concept" (your current "Introduction"), "Methods" (which covers most of the rest of the article) and "Computer software". I'd also look at other articles in the Physics and Astronomy section of WP:GA to compare against. You also could probably do easily with one or two more pictures to help enhance the understanding of the article - make sure that that tone is written for general understanding (which mostly seems to be there) which pictures can be used to help significantly. --Masem 13:23, 30 July 2007 (UTC)"

I think we can do much of what he suggests but I can not think of a single photpgraph that would help. Any thoughts? --Bduke 22:29, 30 July 2007 (UTC)

Perhaps not photographs, but figures (the only photographs I can think of are of people or computers for the history section). For example, a figure showing the ball-and-spring concept of MM; a figure showing a PES; a figure showing molecular orbitals or electron density distribution; a figure (maybe even an animation) showing the idea of MD; a figure showing the LCAO concept (or an MO energy diagram); a figure depicting the docking or ligand binding concept; a figure of the quantum chemical topology concept. I'll try to help at some point but I need time to really sit down and work on it. On a separate note, I personally hate using citation templates but I don't mind if others use them. But the guidelines clearly say that they are optional, so that's not a valid reason for questioning the "goodness" of an article. --Itub 08:28, 31 July 2007 (UTC)
By unanimous consensus, the article has been delisted. The archived discussion can be seen here. Once the article is brought up to standards, it may be renominated at WP:GAC. Regards, Lara♥Love 21:45, 2 August 2007 (UTC)
Well, none of us knew this was going on and I have told them so. They need to communicate better. It is however no problem, as it never was really a good article. We can nominate it when we are ready. --Bduke 22:23, 2 August 2007 (UTC)

missing balance concerning references on software suites[edit]

Gaussian is mentioned, Gamess is not. Orca is not present in the Wikipedia at all. The article should give an overview of available software instead of promoting software that has been a commercial project for about 20 years, this is advertisement, not information. JPBoyd (talk) —Preceding undated comment added 22:18, 12 August 2011 (UTC).

Both GAMESS(US) (and GAMESS(UK)) and Orca are mentioned at List of quantum chemistry and solid state physics software, as are many more. That is where individual programs are mentioned. Programs can have their own articles if they meet wikipedia notability guidelines. Gaussian is only mentioned in a historical context and I think that is appropriate. --Bduke (Discussion) 15:24, 14 August 2011 (UTC)

First paragraph to be introductory. Yes?[edit]

Does this sentence, if it means anything at all, really belong in the first paragraph? "Its necessity arises from the well-known fact that apart from relatively recent results concerning the hydrogen molecular ion (see references therein for more details), the quantum many-body problem cannot be solved analytically, much less in closed form." - Rhodesh (talk) 05:09, 10 October 2013 (UTC)