Timeline for What are the most fundamental classes of mathematical algorithms?
Current License: CC BY-SA 2.5
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 4, 2014 at 17:12 | history | made wiki | Post Made Community Wiki by Todd Trimble | ||
| Oct 4, 2014 at 17:03 | comment | added | Włodzimierz Holsztyński | In the same breath when mentioning Metropolis algorithm I would add simulating annealing. Also, when simulation annealing is done it could be followed by, say, greedy algorithm to get the last drops of optimization. | |
| Aug 3, 2010 at 20:38 | comment | added | András Salamon | And the optimizing compiler -- that is a whole bag of algorithms in one! | |
| Aug 3, 2010 at 20:34 | comment | added | András Salamon | Regarding #7, I thought Quicksort is pretty much a curiosity for large databases? Large sets of data are typically sorted with mergesort, which can be made to work well in the presence of a memory hierarchy, unlike Quicksort. In addition, what about metastability for #1 -- I thought for some types of problems Monte Carlo simulations just plain don't work, because the state space has densely connected regions that are connected via a few bridges. I think I prefer David Eppstein's list. | |
| Jun 9, 2010 at 23:18 | comment | added | Wadim Zudilin | Victor, isn't it hidden somewhere? In any case, the original sourse (Jon's review) is some kind of justification. If we try to select Top 10 on MO, I would expect an unpredictable list (different from Roland's and Jon's): I can't understand people's voting logic. | |
| Jun 9, 2010 at 18:44 | comment | added | Victor Protsak | Wadim: Greedy algorithm! | |
| Jun 9, 2010 at 13:27 | comment | added | Roland Bacher | No idea but Monte-Carlo based simulations of brownians and similar is perhaps not too far from the truth. | |
| Jun 9, 2010 at 12:49 | comment | added | Wadim Zudilin | Roland, that's a good point! But what algorithms are used at Wall Street? (This could be a separate question. :-) ) | |
| Jun 9, 2010 at 12:13 | comment | added | Roland Bacher | One argument for excluding cryptography: it should not exist in a perfect world! More convincingly, it is not really general purpose but adresses a concrete practical spaect of life. Including it, one has also to accept algorithms used at Wall-Street, algorithms useful for polls etc. | |
| Jun 9, 2010 at 10:52 | comment | added | Wadim Zudilin | I don't wish to say that I share my opinion with Jon, but he is really an expert in this area which one can feel reading the very nice review. As for LLL, I did ask Jon why it isn't included. He says that it's a development of (9). Cryptography is indeed not mentioned by the reasons explained in the article. Immiediately after seeing Roland's question I remembered Jon's review; I hope that the link is a good answer to the OP, isn't it? (Please downvote then. :-( ) | |
| Jun 9, 2010 at 10:13 | comment | added | Robin Chapman | Integer Relation Detection (1977) precedes LLL (1982, not included). How queer! :-) | |
| Jun 9, 2010 at 10:05 | comment | added | Charles Matthews | Strange to ignore totally the area of cryptography. The emphasis is obviously on "scientific computing". | |
| Jun 9, 2010 at 9:37 | comment | added | Roland Bacher | Interesting. (1), (9) and (10) should probably be added to my list. (3), (4), (6) and (8) are contained in point (3) and (7) is contained in (8). I am not sure about (5), it should at least be an arbitrary compiler and not necessary Fortran. | |
| Jun 9, 2010 at 8:46 | history | answered | Wadim Zudilin | CC BY-SA 2.5 |