Talk:Theoretical chemistry

Expansion:

I thought that maybe this should be more like a "portal" to the very diverse fields to be found. Everything that can be said about theoretical chemistry can be explained in detail in the sections it refers to (like DFT, hartree fock, computational and such)

But maybe this could be a spot for some history lessons as well! any suggestions on this?

edit by 213.84.48.6

I think the new edits are not really in line with the editorial line which ruled up to now. Whether this editorial line is a good thing can of course be discussed (see Talk:Quantum chemistry and Talk:Computational chemistry).

• Theoretical Chemistry: All theoretical approaches to chemistry
• Quantum chemistry: All quantum mechanical approachesto chemistry (since no quantum chemical methods developed after the 1970s without computational realization, this article is an historical introduction to the field.)
• Computational chemistry: All computational approaches to chemistry

As I understood : Computational chemistry is a branch of theoretical chemistry with strong overlap with quantum chemistry and quantum chemistry is also a branch of theoretical chemistry with strong overlap with computational chemistry. Which branches of theoretical chemistry are not within Computational chemistry are mathematical chemistry, drug design, design of chemical databases and other things (which I don't know about -- it even seems no wiki editor is ready to write much more than stubs about them, see Molecular modelling) which seem to be important and justify the existence of the Theoretical Chemistry article.

All what has been added to the article by 213.84.48.6 are quantum mechanical methods. The movement of the electrons is quantal (and cannot be approximated anyhow -- except in Rydberg atoms -- within classical mechanics) thus all semi-empirical and force-field methods are approximations to exact quantum mechanical treatments. They therefore pertain to quantum chemistry or computational chemistry. Your comments are welcome. --131.220.68.177 08:55, 16 August 2005 (UTC)

Re by 213.84.48.6

Well, I'have had some discussion here with a genuine theoretical chemist (I'm only a student so far). His conclusions so far: Theoretical chemistry == quantum chemistry (qm and semi-empirical) computational == everything that does not use the quantum electron formalism and focusses on anything larger than atoms So, I'm a bit stuck right now and I am going to do some more field research at the theoretical chemistry department in Amsterdam! I don't think that is right to place both comp and quant under theoretical, but I'm gonna think on it. User:WijzeWillem

The problem is that there aren't clear and unaminously agreed definitions for theoretical and computational chemistry. I don't necessarily agree with your friend's distinction.

I like the current opening sentence from the article (with my emphasis added): "Theoretical chemistry is the use of non-experimental reasoning to explain or predict chemical phenomena." By analogy, I would define computational chemistry with: "Computational chemistry is the use of computation to explain or predict chemical phenomena." The two mesh nicely and it is difficult to describe one without refering to the other. Much of computational chemistry is directly built on, or applies, theoretical chemistry. I disagree with the current introduction on the comp. chem article because I don't think it is right to say that comp. chem. is a branch of theo. chem. mainly because I consider computation (or simulation) to be an experimental approach. It is more like the practical application of theo. chem.

However, there will be plenty of people who disagree with my classification too. Stewart Adcock 12:52, 18 August 2005 (UTC)

I don't agree with your classification because I think computational chemistry is a branch of theoretical chemistry because you cannot consider a numeric simulation as an experiment. You are not watching reality but a model of it. This is a big difference with experiment. And, if you do so, you need to know which approximations, hypotheses or even axioms are behind this simulation. If you try to understand how far the simulation is a correct one, you have to do theory -- i.e. non experimental reasonning -- because you have to analyse how far the underlying hypotheses are relevant in the case you consider. This is by far not an experimental approach. I therefore think computational chemistry pertains to theoretical chemistry.--147.231.28.83 09:53, 23 August 2005 (UTC)

...because you cannot consider a numeric simulation as an experiment. Yes I can (and do!). We now have the danger of drifting into an argument about semantics, but I believe that a numerical computation can be an experiment. Even in physical studies, one could use a model system (e.g. a cell line instead of a real clinical subject, leaf samples instead of fields of crops) in tests. Are these studies not experiments? One still needs to know the approximations etc. behind the model. I don't make any distinction between a "real-world experiment" and a "computational experiment" in this sense. Just my two cents. Stewart Adcock 11:59, 23 August 2005 (UTC)

Well if I understand you well you mean : "I make experiments with a given code, i.e. I try some different input, and I observe different outputs." This is true : this is a experimental reasonning applied to the object which is the code. But the object you are analysing is not "chemistry" but the "code". So you are not doing computational chemistry but informatics. On the other hand if you study leaf samples instead of fields of crops you make the hypothesis you are not really making agriculture anymore but biology. You are still studying nature not a human made algorithm. --147.231.28.83 12:16, 23 August 2005 (UTC)

Not really. In computational chemistry, using your terms, the object you are analysing is still "chemistry" and the code is just a tool to facilitate that analysis. Crucially, it is the results generated by the algorithm that you are studying (and these will replicate nature to some extent, one would hope) and not the algorithm itself. (Of course, you might be studying the algorithm itself, but that would be deviating from the point of the original comment.)

Let's use a example. One of the most prevalent types of simulation in practical computational chemistry is ligand docking. One might hypothesise that a potential drug molecule binds to a specific protein and perform a computational experiment to predict whether this is true. Is this not an experiment?

Stewart Adcock 15:32, 23 August 2005 (UTC)

For me the problem is how far you can trust the results of your computation. For example if you optimize the geometry of NH3 in a small basis set you obtain a flat molecule. If you improve the basis set you obtain a pyramidal geometry. Deciding whether or not your results can be trusted is from my point of view a theoretical process. Some processes in chemistry can only be explained if you take electron correlation into account. You have to know which ones and to which extend it must be done. This decision step is also a theoretical reasonning. Comparing the results obtained within different approximation schemes is also a theoretical reasonning. If you use only one method, one code, and one level of approximation, then yes you are doing experiment but if you begin to compare or rationalize results then, in my opinion, you are doing theory.--147.231.28.83 11:43, 24 August 2005 (UTC)

How far you can trust the results of your practical study? For example, if you try to measure gravity simply by dropping a weight from high you obtain a very poor estimate. Deciding whether or not your results can be trusted (or to what extent they are applicable) is an essential analysis to be performed as part of any experimental analysis. One could argue that any statistical analysis or consideration of sources of error or comparison of alternative results are all theoretical processes. The fact remains that an experiment is still an experiment despite being performed in silico.

Let's use a hypothetical example: I have a potential drug target, but for any drug to be therapeutically useful it must bind to one particular protein but must not bind to a very closely related protein. Given a molecule that binds sufficiently strongly to both proteins, I might hypothesise that I can add a steric group to this molecule to prevent it binding effectively to the second protein. My experiment to test this would involve performing ligand docking to both proteins with some steric group appended to the molecule of interest. (The docking can be performed with a variety of different softwares, based on different search algorithms and descriptions of molecular interactions, and I'll probably use at least two. So, this is still applicable to your point about different theoretical models.) This type of calculation is bread-and-butter computational chemistry. Are you trying to tell me that this was not an experiment? Stewart Adcock 08:48, 8 September 2005 (UTC)

OK. I begin to understand. This is a way of thinking which is very far away from the one I am used to. I thank you very much for these explanations. This is a new world for me. Nevertheless, when you have made a computation let's say with different algorithms and selected I don't know how many candidate molecules for a particular drug, then you have a theoretical result, don't you? --at least it is not experimental because you have not tested your hypotheses directly on nature-- And you have to check whether some of them have really the property you were looking for, that is you check whether your theory was good by confronting it with experiment, don't you? If it is not the case you have to change of algorithm or change your strategy. This is exactly the same that we do in quantum chemistry. We stay hours in front of our computer preparing input and analyzing output and then go to the lab and compare our results with experiments. The only difference is maybe that our models are based on first quantum mechanical principles and your models are based on heuristic assumptions and molecular mechanics but I think basically the reasoning is in both case theoretical. You tell me what you do has more to do with experiment than with theory. Do you think that choosing a basis set has so much to do with theory. This is usually more a trying-and-error procedure (usually even strongly interfering with experiments) than a perfect ab initio method. 131.220.68.177 15:22, 9 September 2005 (UTC)

==

Very nice work on the most recent additions (sept 5 and 7 2005.) However, Mathematical chemistry contains more areas of research than graph theory. Graph theory is a *part* of mathematical chemistry, not *all* of mathematical chemistry!

==

OK Why not. That's a pity that no editor seems to be ready to expand the mathematical chemistry stub. --81.209.204.201 18:18, 7 September 2005 (UTC)

Reasons for caution

The opening sentence was completely wrong. I have replaced it by an opening paragraph that is reasonable, but have no time to go further. Michael P. Barnett (talk) 20:12, 30 January 2011 (UTC)

I put the opening sentence back, preceded by a section heading "Alternative approach". If this tactic is acceptable when I am working on an article that needs my attention, and it links to an article that is inadequate, with which I wish to remain as uninvolved as possible, and certainly avoid trying to edit material put together without awareness or understanding of standard textbook coverage, I will continue to do so. Michael P. Barnett (talk) 15:46, 2 February 2011 (UTC)

Sorry -- on reflection that last remark was inappropriately ungracious. I just took a look at discussion to date. Much of it valid. I am looking at topic from having started my undergraduate course work in Chemistry in 1945, writing survey articles and editing thematic issues of journals, that dealt with applications of software and math to chemistry across the board during past five years. This gives a wider than general perspective. But for editors without that background who brought article to present level, I really should commend their efforts. Back to question, trade off between benefit of articles being started by non-experts and hazards. I do not know where it is. But I will try to contribute, and to respect the efforts of other people. Michael P. Barnett (talk) 18:15, 2 February 2011 (UTC)

On further reflection, an online vehicle for people to learn about a subject by trying to put material about it together, for correction by an expert, can be an interesting way of teaching. But until an article has been stabilized accurately by dint of expert input, it is extremely dangerous for the article to be accessible as a source of information, particularly in an encyclopedic work that wants to become established as a source of accurate information rather than wildly incorrect misinformation. The problem with articles like this is that the typical reader will not think of turning to the Discussion, which can give some indication of the level of insight of contributors -- in this case particularly candid, in that one contributor reported asking a graduate student for an opinion. That is not the way to construct a dependable encyclopedia. A big question is how sets of articles such as the tree I have been led to from the Coulson biography came into existence. If someone has told a team who have no knowledge of a field to find terms that are used in it and write articles, that is counterproductive to a potentially disastrous extent. Michael P. Barnett (talk) 10:43, 3 February 2011 (UTC)

First, I note with sadness that Michael P. Barnett died earlier this year, so he can not continue the good work he started on this article. See the WP article on him at Michael P. Barnett. We do not need an alternative lede, so I have combinbed the two and done sone general cleaning up. I will try to do more. --Bduke (Discussion) 23:42, 29 December 2012 (UTC)