User talk:Kiefer.Wolfowitz/Archive 8
![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 5 | Archive 6 | Archive 7 | Archive 8 | Archive 9 | Archive 10 | → | Archive 15 |
The Signpost: 28 March 2011
- News and notes: Berlin conference highlights relation between chapters and Foundation; annual report; brief news
- In the news: Sue Gardner interviewed; Imperial College student society launched; Indian languages; brief news
- WikiProject report: Linking with WikiProject Wikify
- Features and admins: Featured list milestone
- Arbitration report: New case opens; Monty Hall problem case closes – what does the decision tell us?
Speedy deletion declined: Martine Abdallah-Pretceille
Hello Kiefer.Wolfowitz. I am just letting you know that I declined the speedy deletion of Martine Abdallah-Pretceille, a page you tagged for speedy deletion, because of the following concern: Article in :fr says she is "professeure à l'Université de Paris VIII (« Vincennes à Saint-Denis »)". I think there is enough to pass A7, but maybe not WP:PROF - consider AfD. JohnCD (talk) 11:06, 1 April 2011 (UTC)
- Thanks for the courteous and helpful note, correcting my error. Cheers, Kiefer.Wolfowitz (Discussion) 05:53, 2 April 2011 (UTC)
Buggy editors since "upgrade" of JAVA
Netscape, Firefox (updated version 4), and Chrome have all frozen when I've used the editing templates in the last week, many times. Therefore, I've stopped worrying about the niceties of m-dashes and n-dashes, and now save changes with even slight edits. My apologies to Michael Hardy about my dashes. Kiefer.Wolfowitz (Discussion) 02:41, 5 April 2011 (UTC)
The next few items are motivated by the Monty Hall problem arbitration. Kiefer.Wolfowitz (Discussion) 16:33, 20 March 2011 (UTC)
Original mathematical research: Example
I argue that explaining mathematical results for the general public sometimes requires original exposition, glossing topics that are familiar to mathematicians. The current version of Shapley–Folkman lemma is simpler but uses some original simplifications of a research result (of Ekeland).
Current article: Sequential convergence simplification
For a separable problem, we consider an optimal solution (xmin, f(xmin) )
to the "convexified problem", where convex hulls are taken of the graphs of the summand functions. Such an optimal solution is the limit of a sequence of points in the convexified problem
- ^ (Ekeland 1999, pp. 357–359): Published in the first English edition of 1976, Ekeland's appendix proves the Shapley–Folkman lemma, also acknowledging Lemaréchal's experiments on page 373.
- ^
The limit of a sequence is a member of the closure of the original set, which is the smallest closed set that contains the original set. The Minkowski sum of two closed sets need not be closed, so the following inclusion can be strict
- Clos(P) + Clos(Q) ⊆ Clos( Clos(P) + Clos(Q) );
- Ekeland, Ivar (1999). "Appendix I: An a priori estimate in convex programming". In Ekeland, Ivar; Temam, Roger (eds.). Convex analysis and variational problems. Classics in applied mathematics. Vol. 28 (Corrected reprinting of the (1976) North-Holland ed.). Philadelphia, PA: Society for Industrial and Applied Mathematics (SIAM). pp. 357–373. ISBN 0-89871-450-8. MR 1727362.
{{cite book}}
: Invalid|ref=harv
(help) - Green, Jerry; Heller, Walter P. (1981). "1 Mathematical analysis and convexity with applications to economics". In Arrow, Kenneth Joseph; Intriligator, Michael D (eds.). Handbook of mathematical economics, Volume I. Handbooks in economics. Vol. 1. Amsterdam: North-Holland Publishing Co. pp. 15–52. doi:10.1016/S1573-4382(81)01005-9. ISBN 0-444-86126-2. MR 0634800.
{{cite book}}
: Invalid|ref=harv
(help) - Rockafellar, R. Tyrrell (1997). Convex analysis. Princeton landmarks in mathematics (Reprint of the 1979 Princeton mathematical series 28 ed.). Princeton, NJ: Princeton University Press. pp. xviii+451. ISBN 0-691-01586-4. MR 0274683.
{{cite book}}
: Cite has empty unknown parameter:|1=
(help); Invalid|ref=harv
(help); More than one of|MR=
and|mr=
specified (help)
Old version: Glossing closure of a set, at imho excessive length
For a separable problem, we consider an optimal solution
- xmin = (x1, ..., xN)min
with the minimum value f(xmin). For a separable problem, one could guess that an optimal point (xmin, f(xmin)) might be contained in the sum of the convex hulls of the summands' graphs
- (xmin, f(xmin) ) ∈ ∑ Conv (Graph( fn )),
but this guess would be wrong. In truth, a more complicated expression
- (xmin, f(xmin) ) ∈ Clos( ∑ Conv (Graph(fn ) ) )
using the closure of a set is needed.[1] (Topologically closed sets are defined in the following subsection.) An application of the Shapley–Folkman lemma represents the given optimal-point as a sum of points in the graphs of the original summands and of a small number of convexified summands.[1] ...
Closure of a set
The preceeding results use concepts of closed sets from mathematical analysis (the theory of calculus),[2] which we define. As suggested by the preceding subsection, closed sets are useful for proving that some minimum solution exists.[3]
- A set is closed if it contains its limit points.
An example of a non-closed set is the harmonic sequence
- ( 1⁄n : n is a positive natural number ) =( 1, 1⁄2, 1⁄3, ... ),
which converges to zero. The equation 0=1⁄n is unsolvable in natural numbers, which implies that zero is not a memberof { 1⁄n }. However, because zero is a limit point of the harmonic sequence, the set { 1⁄n } is not closed.
- The closure of a set Q, denoted Clos(Q), is the union of Q and all the limit points of Q. (This definition implies that the closure of a set is indeed a closed set.)
Thus, the closure of the harmonic sequence includes zero, which is its only limit point
- Clos( { 1⁄n } ) = {0, 1, 1⁄2, 1⁄3, ... }.
In terms of optimization theory, the set of the harmonic sequence lacks a minimum, but its closure has a minimum. This example shows the use of closed sets in optimization theory.
The Minkowski sum of two closed sets need not be closed, so the following inclusion can be strict
- Clos(P) + Clos(Q) ⊆ Clos( Clos(P) + Clos(Q) );
the inclusion can be strict even for two convex closed summand-sets.[4] In many problems, then, ensuring that the Minkowski sum of sets be closed requires the closure operation. In the optimization theory of the preceding subsection, the closure operation ensures that the set
- Clos( ∑ Conv (Graph( fn ) ) )
is indeed closed.
- ^ a b (Ekeland 1999, pp. 357–359): Published in the first English edition of 1976, Ekeland's appendix proves the Shapley–Folkman lemma, also acknowledging Lemaréchal's experiments on page 373.
- ^ Rockafellar (1997, pp. 43–44)
- ^ Rockafellar (1997, p. 72)
- ^ Rockafellar (1997, pp. 49 and 75)
Discussion: RE Monty Hall Problem arbitration
The second version just glosses Ekeland's analysis, using the closure of a set. Glossing this notion requires a lot of work.
The first version simplifies Ekeland's analysis, using the sequential closure of a set, which is the same thing here, because of the uniqueness of (reasonable) topologies in finite dimension. All mathematicians would recognize that the second version is essentially the same version, but this requires knowledge of undergraduate mathematics (or beginning USA graduate mathematics).
My concern is that Phil Knight's (original and imho even revised) finding in the Monty Hall Problem arbitration would seem to ban the simplification of the first version as WP:Original research. Sincerely, Kiefer.Wolfowitz (Discussion) 23:29, 13 March 2011 (UTC)
- On the talk page of the WikiProject Mathematics, editor JRSpriggs notes that mathematicians on WP most often invoke OR (especially by synthesis, I'll add) when dealing with tendentious editing. Otherwise, WP mathematicians use examples that are similar in spirit to published examples; simililarly with proofs. My concern is that such (trivial to a mathematician) examples or proofs may be barred by the proposed wording.
- In practice, the proposed wording (if adopted more widely) may make it impossible to have mathematical articles achieve featured or good status, which would be a loss to WP and mathematics. Kiefer.Wolfowitz (Discussion) 09:48, 14 March 2011 (UTC)
- If someone provides a new example for a well established proof or method, that is analogous to someone using a different example to distinguish a simile from a metaphor, to avoid copyvio, and no-one would challenge it. The issue with your gloss is that without there being a published version from which you are working, you would have a hard time with verifiability. In this case, it would be analogous to someone explaining how some new kind of jet engine worked on the basis that they are an aircraft engineer, and other aircraft engineers would all know it was right. He probably is right, but we wouldn't accept it on Wikipedia until someone published it. Elen of the Roads (talk) 13:17, 14 March 2011 (UTC)
- It would be easy to find reliable references explaining the use of sequential convergence in general. However, this explanation has not appeared for this problem, because it is trivial and not needed for the specialists reading Ekeland's original work. (Ekeland is also interested in functional-analytic problems, so he needs the generality). I believe that such exposition is not regarded as "OR by synthesis" in the WP Mathematics Project. We are worried that villagers with pitchforks and torches may stop our articles on the way to GA or FA status! Kiefer.Wolfowitz (Discussion) 13:39, 14 March 2011 (UTC)
- Geometry Guy asked a question about another example in this article. I'll make sure that the Shapley-Folkman lemma is vetted by the math project first, before asking for A-status or FA status. (It may be that the WP community could refine the OR by synthesis language for mathematics, now that the MHP has gotten attention.) Kiefer.Wolfowitz (Discussion) 13:22, 14 March 2011 (UTC)
- I am troubled by the line, if any, between original research and original pedagogy or original illustration. I doubt this line can be drawn clearly but that seems necessary here at wikipedia; otherwise the research ban curtails development of original pedagogy that is crucial for many articles (not only mathematical ones). It's comforting that some people are paying close attention to this potential problem. Best wishes to all participants.
- Re the illustration under discussion, I think the Old version gloss --from subheading "Closure" to "This is an example use ..."-- would be instructive only for readers who would skip the entire section "Mathematical optimization". --P64 (talk) 18:11, 9 April 2011 (UTC)
Standing up for Gill
I understand you feel strongly about this, but keep in mind that WP:COI is a guideline with no bright line, so "violation" is always going to be a matter of degree in the eye of the beholder. I can easily think of far worst cases, which make anything Gill may have done negligible in comparison, but it's up to ArbCom to decide how to phrase it; they have probably seen a fair share of that e.g. Insisting on a proclamation of absolute non-culpability from them is probably counterproductive at this point. As they used to say "Don't shoot the piano player; he's doing the best he can". Tijfo098 (talk) 18:19, 21 March 2011 (UTC)
- Thanks for your commentary. I tried to give them a short statement, without argument, to avoid taxing their eyeballs. I have been encouraged that previously a few ArbCom members have indicated discomfort with the wording regarding Gill. Thanks again for your good judgment and advice. Kiefer.Wolfowitz (Discussion) 18:54, 21 March 2011 (UTC)
Thank you
The Barnstar of Diplomacy | ||
for your tireless liaison work at Wikipedia:Arbitration/Requests/Case/Monty Hall problem. Elen of the Roads (talk) 13:59, 25 March 2011 (UTC) |
Dear Elen,
That's very kind of you and I greatly appreciate the barnstar, especially coming from you.
I was gladdened also by your spelling out that one of the editors could try cooperatively editing for a spell at the simple English Wikipedia, and then simply ask for a reconsideration.
Warm regards, Kiefer.Wolfowitz (Discussion) 15:51, 25 March 2011 (UTC)
Optimering's message
A regrettable dispute
Hello,
This is a friendly message.
For some reason we had a falling out on Template_talk:Optimization_algorithms. My recollection of the events is that we had been discussing changes to that navbox for months, off and on. I wanted to either delete the navbox or split it into several specialized navboxes because I thought it was too confusing to lump everything into one. You wanted to keep it the way it was, stating it held the most important items, but encouraged me to make edits to it. I finally did, and you apparently disagreed with the edits and considered them an insult somehow. This is most regrettable and I can assure you that it was not intended. I have studied your contributions to Wikipedia and I can see that you are very knowledgeable in certain fields of optimization and advanced mathematics. I think it is wonderful that you are contributing to Wikipedia and I think you and I do it for the same purpose: to help others attain the knowledge we have.
Now, your behaviour towards me was quite aggressive. You also exposed a real-world identity which you think is me (I will neither confirm nor disconfirm), which is actually a very serious offence on Wikipedia and could get you blocked, see WP:OUTING. Furthermore, you have expressed strong dissent and made derogatory remarks about metaheuristics in several talk pages, which means that you have a WP:COI in editing those articles regardless of your personal relations to researchers, publications, etc., because you apparently wish to demote metaheuristics. I have also seen your post making derogatory remarks about certain universities, which is also not suitable for Wikipedia discussions.
However, I think your presence and edits are generally valuable to Wikipedia and this dispute appears to be the exception. I truly regret that we had this falling out and I am willing to bury the hatchet, so to speak, so we can both continue contributing to Wikipedia without these unpleasant disputes. By the way, I have seen your concerns about Local unimodal sampling and as I note on its talk-page I will address them as soon as I can find the time.
Cheers,
Optimering (talk) 16:37, 23 February 2011 (UTC)
- As I stated on the talk page, WP:AGF may compel us to conclude that you are simply a fan of the thesis in question. You are welcome to ask an administrator to block me.
- I am not the first person to comment on your tendentious editing, which has occurred at least since October. At least 4 administrators have cautioned you about WP:OR and WP:COI concerns, and I see little reason to believe that you have taken them seriously, or considered the consequences of a larger discussion of your editing.
- If you want to add the thesis to articles or write WP articles about its contents, few would notice. Your promoting
such topicsmetaheuristics as "the most general/popular" optimization algorithms and adding "ant colony algorithm on the optimization template, raises your profile and subjects your edits to greater scrutiny. Kiefer.Wolfowitz (Discussion) 17:10, 23 February 2011 (UTC) (10:14, 24 February 2011 (UTC)) - People who solve real problems in optimization and computer science, e.g. Papadimitriou or Karp, have earned the respect of the optimization community. They carry the union card. I don't have a COI against "heuristics", and I would recommend that you read good articles interpolating between something theoretical and something empirical, something old and something new:
- Stefankovic,Daniel; Vempala, Santosh; Vigoda, Eric. Adaptive simulated annealing: a near-optimal connection between sampling and counting. J. ACM 56 (2009), no. 3, Art. 18, 36 pp.
- Kirkpatrick, S.; Gelatt, C. D., Jr.; Vecchi, M. P. Optimization by simulated annealing. Science 220 (1983), no. 4598, 671–680
- I am not an expert in heuristics or metaheuristics, generally, but knowing a fair amount about stochastic methods of optimization was a prerequisite to signing my name Kiefer.Wolfowitz (Discussion) 17:32, 23 February 2011 (UTC).
- You misunderstand me again. I did not post metaheuristics in the navbox believing it was the most popular, important, or to promote it in any way. I posted it because it was how I understood the headline 'methods calling functions,' which sounded to me like programmer's lingo for someone really meaning to say 'metaheuristic.' Also, the list was made in a general-to-specific ordering in which metaheuristics are more general and gradient/hessian are more specific. Inside that category I ordered metaheuristics which I perceive - from experience and citation counts, etc. - to be the most popular first, i.e. genetic algorithm first, and so forth. However, as User:Ruud Koot has mentioned the original navbox was actually highly focused on certain types of optimization methodologies and domains so it really ought to be expanded and split up. We have now started separating the navbox into several independent ones and this should hopefully resolve the dispute. I have no wish to make further comments on this or your other (rather presumptuous and condescending) remarks. Optimering (talk) 18:49, 23 February 2011 (UTC)
- Your edit was problematic for a number of reasons. Your earlier edit summary ("tried to order from general to specific, and from most to least popular/relevant") was less confidence-building than your current explanation, which I'll accept, in the spirit of moving on, which you suggested first.
- I welcome your initiative with starting new nav boxes, as I stated on the page. In particular, a nav-box with problem classes is probably more useful to many persons than is a nav-box on optimization algorithms.
- The nav-box on optimization algorithms should continue to exist, and it should continue to appear on the bottom of the page of its algorithms. I would be pleased if the a line on graph/network algorithms would be added, as Ruud suggested. When I started a line on combinatorial algorithms, I just used the paradigmatic algorithms mentioned in the first edition of Cormen, Rivest, et alia, and the combinatorial algorithms mentioned in Shapiro and Minoux, for example. The weak state of many additional articles on combinatorial optimization and my own ignorance & time-limitations inhibited me from adding further algorithms. Again, I would encourage Ruud and others to add a graph/network algorithm line.
- I mentioned my ideas of building nav-boxes on optimization theory (foundations, mathematics) and on convex/variational analysis. These might be viable simply because the mathematics articles are in better shape. The optimization articles are generally weak, and many are inferior to their counterparts on French Wikipedia, for example. It would be good to prioritize expanding content, for example, with improving integer programming or starting Roger Fletcher's filter method, imho.
- I would encourage you to join the WikiProject groups on computer science and on systems and even on mathematics (which is large), and discuss your ideas there. I believe that such public discussions would help us both with redoing the optimization algorithms nav-box, etc.
- Sincerely, Kiefer.Wolfowitz (Discussion) 08:24, 24 February 2011 (UTC)
- P.S. I believe I was thinking of the Mockus family of stochastic optimization, rather than the Paulakus family, when I mentioned Lithuanians.
- Let us then accept your account of your editing with the local unimodal sampling article. You seem to have understood my concern with the use of sources and apparent OR ... trying ... to document notability of the thesis. I would suggest you develop an appropriate article (which you outlined above) in your sandbox, and then add it to the mainspace, replacing the lus article with a redirect. Kiefer.Wolfowitz (Discussion) 08:29, 24 February 2011 (UTC)
- You misunderstand me again. I did not post metaheuristics in the navbox believing it was the most popular, important, or to promote it in any way. I posted it because it was how I understood the headline 'methods calling functions,' which sounded to me like programmer's lingo for someone really meaning to say 'metaheuristic.' Also, the list was made in a general-to-specific ordering in which metaheuristics are more general and gradient/hessian are more specific. Inside that category I ordered metaheuristics which I perceive - from experience and citation counts, etc. - to be the most popular first, i.e. genetic algorithm first, and so forth. However, as User:Ruud Koot has mentioned the original navbox was actually highly focused on certain types of optimization methodologies and domains so it really ought to be expanded and split up. We have now started separating the navbox into several independent ones and this should hopefully resolve the dispute. I have no wish to make further comments on this or your other (rather presumptuous and condescending) remarks. Optimering (talk) 18:49, 23 February 2011 (UTC)
Others renewed the discussion of the edits of Optimering at the Conflict of Interest noticeboard. Kiefer.Wolfowitz (Discussion) 15:33, 6 March 2011 (UTC)
I'm just creating an archival copy of the nomination for Did You Know?. (The last nomination had its picture somehow corrupted.) Kiefer.Wolfowitz (Discussion) 17:37, 25 March 2011 (UTC)
- ... that, while the criss-cross algorithm visits all 8 corners of the Klee–Minty cube when started at a worst corner, it visits only 3 more corners on average when started at a random corner?
- Reviewed: Susie Fishbein ([1])
- Comment: CRGreathouse has applauded this hook, which has 167 characters.
Created by Kiefer.Wolfowitz (talk). Self nom at 16:43, 25 March 2011 (UTC)
DYK for Criss-cross algorithm
On 5 April 2011, Did you know? was updated with a fact from the article Criss-cross algorithm, which you created or substantially expanded. The fact was ... that, while the criss-cross algorithm visits all eight corners of the Klee–Minty cube when started at a worst corner, it visits only three more corners on average when started at a random corner? You are welcome to check how many hits the article got while on the front page (here's how, quick check) and add it to DYKSTATS if it got over 5,000. If you know of another interesting fact from a recently created article, then please suggest it on the Did you know? talk page. |
The KM cube received more than 5K visits and the criss cross algorithm received over 4K visits. Kiefer.Wolfowitz (Discussion) 07:57, 7 April 2011 (UTC)
Ominous cubes: Prolegomena to a navbox ?
For sentimentality, I'll put a copy of this here. Kiefer.Wolfowitz (Discussion) 22:30, 22 April 2011 (UTC)
The old version had only two groups, "nice" and "ominous":
The ominous cubes were the Hellraiser cube, the Cosmic Cube, and (!) the Klee-Minty cube. Kiefer.Wolfowitz (Discussion) 19:28, 22 April 2011 (UTC)
Probably under the influence of April Fools' Day .... Kiefer.Wolfowitz (Discussion) 20:32, 21 April 2011 (UTC)
Template:Cubes has been nominated for deletion. You are invited to comment on the discussion at the template's entry on the Templates for discussion page. JaGatalk 07:31, 19 April 2011 (UTC)
Another optimization-related topic.... Kiefer.Wolfowitz (Discussion) 20:39, 21 April 2011 (UTC)
This is an automated message from CorenSearchBot. I have performed a web search with the contents of Variable splitting, and it appears to include material copied directly from http://reasoning.cs.ucla.edu/fetch.php?id=102&type=pdf.
It is possible that the bot is confused and found similarity where none actually exists. If that is the case, you can remove the tag from the article. The article will be reviewed to determine if there are any copyright issues.
If substantial content is duplicated and it is not public domain or available under a compatible license, it will be deleted. For legal reasons, we cannot accept copyrighted text or images borrowed from other web sites or printed material. You may use such publications as a source of information, but not as a source of sentences. See our copyright policy for further details. (If you own the copyright to the previously published content and wish to donate it, see Wikipedia:Donating copyrighted materials for the procedure.) CorenSearchBot (talk) 08:41, 12 April 2011 (UTC)
- The bot is wrong. I created this article from scratch, using material I remember from Michael Andersson's thesis under Ake Bjork at Linkoping University c. 2000.
- Adlers, Mikael; Björck, Åke. "Matrix stretching for sparse least squares problems". Numerical Linear Algebra with Applications. 7 (2). John Wiley & Sons, Ltd. ISSN 1099-1506.
{{cite journal}}
: Invalid|ref=harv
(help) - Alvarado, Fernando (1997). "Matrix enlarging methods and their application". BIT Numerical Mathematics. 37 (3): 473–505. doi:10.1007/BF02510237.
{{cite journal}}
: Invalid|ref=harv
(help) - Grcar, J. F. Matrix stretching for linear equations, Technical Report SAND90-8723, Sandia National Laboratories, November 1990.
- Vanderbei, Robert J. (1991). "Splitting dense columns in sparse linear systems". Linear Algebra and its Applications. 152: 107–117. doi:10.1016/0024-3795(91)90269-3. ISSN 0024-3795.
{{cite journal}}
: Invalid|ref=harv
(help); Unknown parameter|month=
ignored (help)
- The cited paper failed to claim to discover the Americas and Africa, strangely, although it fails to cite any literature from the established literature. Their Dean(s) must be so very impressed with their magnificent research accomplishments! Kiefer.Wolfowitz (Discussion)