Talk:Scientific computing

Scientific computing and scientific computation

The article Scientific Computation claims that there is a difference between scientific computing and scientific computation. This seems to be the author's personal view, so I listed Scientific Computation for deletion at Wikipedia:Votes for deletion/Scientific Computation. Please join the discussion, especially if you do think that such a difference exist. -- Jitse Niesen (talk) 13:06, 20 August 2005 (UTC)

If there is a difference, it is very subtle. There is now only one article! Stewart Adcock 21:40, 22 September 2005 (UTC)

Misleading statement about Wolfram and Schmidhuber

In my opinion the statement that computational science is a new third mode of science, and stating that is a thesis of both Wolfram and Schmidhuber is misleading. I have not read either of their work, only wikipedia articles describing their work. Based on the Wikipedia articles on 'A New Kind of Science' and 'Jürgen Schmidhuber'. Based on my rudimentary understanding of their work from these articles, their main idea is that very basic programs give rise to complicated phenomena which could be used to study or describe the universe. Computational science is not at all about using cute little programs which generate complicated phenomena.

For the purpose of being bold, I removed it

New person: I undid the undo removing names of the noted "founders" of "computational science." While I personally don't buy the "third" argument, Ken Wilson (who has a Wikipedia page) deserves credit for the term before Stephen did. So too Larry and Sid. And I just remembered to add Stephen's PhD advisor Geoff (who doesn't have a Wikipedia page and the one there will require disarmbiguation) who left a faculty position at Caltech because of this argument.

I see the Gold star by JJL's wikipedia page for resolving conflict, but the edit almost has me believing that he or she is working for Wolfram Associates (and I am familiar with some of their "strong arm" tactics) and doesn't know much about computational science and scientific computing.

I've known both Stephen and Geoff for over 2 decades, and they both have what some consider abrasive personalities. I can understand why. The problem with Stephen's 'A New Kind of Science' is that you don't hear from him the problems with the ideas of celluar automata applied to and generalized to other fields like computational fluid dynamics (harder to do shock wave physics (transition zones, boundary conditions, etc.). And Stephen has addressed some of these criticisms in the new book (which I still am not quite willing to buy, but I've heard his speel before, and he has improved over the years, but he has lots more skeptical people).

I don't know enough about Schmidhuber to comment.

--enm 23:30 16 Oct 2007 UTC —Preceding unsigned comment added by 143.232.210.38 (talk) 22:25, 16 October 2007 (UTC)

The list of names in that section is too long and does not cite WP:RS. Your comments about the personalities involved are irrelevant (see WP:NPOV). But good job discerning the massive yet uninformed corporate conspiracy behind that one revert. JJL 23:25, 16 October 2007 (UTC)

List of programs

I concur with the removal of the list of specific universities offering these programs. JJL 14:14, 11 December 2006 (UTC)

Scientific computing vs computers science

I don't see a difference. It sounds to me like computer science applied by scientists of other disciplines. Is it different algorithms, more powerful computers, more complex software, or multidisciplinary studies? What's the difference? Oicumayberight 05:03, 7 January 2007 (UTC)

It's the study of numerical algorithms in particular vs. algorithms in general. JJL 03:16, 27 April 2007 (UTC)

Oicumayberight's use of the word "applied" is the critical word. It's not all merely numeric: I've had this hassle for a couple of decades. Real-time process control of experimental apparatus is Scientific Computing, there is a glossy magazine with that title (I think a weekly). Many people would like direct solution to problems. In some circles the terms algorithms and heuristics are dirty words. It's not all numeric; e.g., Mathatica, Matlab, and Maple

What this Wikipedia page (SC) fails to note is the emotional vitriolic debate about why Computational Science came about. There are people, many physicists for instance but also chemists and mathematicians, who actively hate computer science as a field (to them the World Wide Web is a creation of physics [arguably] not computer science [just ask Tim Berners-Lee for what he was dumped upon]). Similar steam gets generated in Scientific Visualization, but the arts, artists, and the art community are more of a target [rightfully in that case]. This vitriol is an ugly little competitive bit of science of the type James Watson wrote about in The Double Helix. Rob Pike when he was at Bell Labs also wrote a very good paper titled "Computer Science vs. Physics" which he once gave.

--enm 00:15 10 Oct 2007 (UTC)

Fractals and Chaos

These should not be external links here. Computers are used to iterate the chaotic maps and to draw the fractal pictures, but chaos and fractals have little if anything to do with sci. comp. Cf. the other fields listed, which apply numerical methods to practical problems and use nontrivial algorithms. JJL 03:16, 27 April 2007 (UTC)

Dear JJL. Methinks you are guarding the temple too narrowly. chaos theory does just what you say "apply numerical methods ...". I'll defer to others with more knowledge on fractals. Bellagio99 03:40, 27 April 2007 (UTC)

I think there is a difference between fields like computational chemistry, which are about the application of numerical methods, and chaos theory, where numerical methods are simply one of the many tools. So I removed chaos theory, and also (for the same reason) graph theory and social networks. -- Jitse Niesen (talk) 03:55, 27 April 2007 (UTC)

I am sorry, but I do social network analysis, and it is basically about the application of complex numerical methods. Do you have expertise in this area? You are defining the church too narrowly. I shall re-add. —The preceding unsigned comment was added by Bellagio99 (talk • contribs) 12:27, 27 April 2007 (UTC).

I strongly agree with Jitse Niesen. Social network analysis uses lots of tools like the SVD and graph-based clustering algorithms, but that's using the results of num. analysis/sci. comp., not doing sci. comp. Most everyone in a technical area uses it--the list would be endless if social network analysis set the standard. The Sandia folks are doing a good job of increasing the prominence of discrete methods in CS&E (see e.g. the Costa Mesa conference earlier this year) but that doesn't make every application of the subject matter an example of, or even strongly related to, sci. comp. JJL 12:36, 27 April 2007 (UTC)

Well, I strongly disagree. It's interesting that you don't cite social scientists. I suspect unconscious anti social science bias here. To obtain results, we "do" scientific computing. It's a bigger church than you realize. Please stop being sectarian. Bellagio99 14:37, 27 April 2007 (UTC)

I think you utilize the results of scientific computing. Psychologists developed LSA (e.g. Thomas Landauer), but the scientific computing angle--figuring out how to do that exceedingly large SVD--was done by computational scientists. Not every use of a computer to perform a scientific task is scientific computing, which creates the techniques. I think you're conflating the application of a subject with the subject itself. What's a scientific computing algorithm developed by and for social scientists? I've read minor modifications of clustering algorithms for, say, the karate school data set come from the social sciences community, but not a novel algorithm together with a proof of its correctness. When social sciences journals start publishing proofs of convergence w.r.t. an appropriate metric, I'll change my mind! JJL 18:04, 27 April 2007 (UTC)

(de-indenting) Bellagio, you are not citing social scientists either. These things have to be defined narrowly, otherwise we will end up with a huge list which is not going to be helping anybody. But I am of course willing to consider the possibility that social networks will fall in this narrow definition.

Let's try to be a bit more constructive than reverting each other. Bellagio is right, I'm not a social scientist, and as far as I know JJL is neither, so I propose that Bellagio writes a paragraph about scientific computing which mentions social networks. I think that if we see something more concrete, it will be easier to discuss whether social networks should be included in the article. -- Jitse Niesen (talk) 15:04, 27 April 2007 (UTC)

Good idea, except I am traveling internationally for the next 9 days and don't know what chance I will have. Frantically packing now. In the interim, why don't you post the same sort of info on the disciplines you think belong in the canon, and then we will have (a) a better article and (b) a better ability to see additional disciplines belong in the article. Cheers, Bellagio99 17:37, 27 April 2007 (UTC)

I think the meaning of science in scientific computing is fairly clear. Perhaps you should start a Social Scientific Computing article instead? I'm not trying to be facetious--it's just that the field is reasonably well-defined, and I don't think you appreciate the differences. Look at what's published in the leading journal in the field, the SIAM J. of Sci. Comp. Very little by social scientists appears there. JJL 18:04, 27 April 2007 (UTC)

As I said, I don't have time to deal with this now. And the fact that one journal guards its gates is most unpersuasive. Take a look at the literature in clique formation in scholarship. Good luck. Bellagio99 19:14, 27 April 2007 (UTC)

Computational Science/Scientific Computing

I switched the redirects. It seems to me that the name Computational Science is winning (e.g., http://www.siam.org/students/resources/report.php). JJL 23:55, 16 October 2007 (UTC)