User talk:Kiefer.Wolfowitz/Archive 6

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 1

←

Archive 4

→

Your comments on my suggestions for Applied mathematics article

As a courtesy I feel I should thank you. Also, I am not defensive, nor do I have a chip on my shoulder. I am 82 and I am exhausted. I am trying to edit articles on subjects about which I may be the only person left who knows about VERIFIABLE things which should be mentioned. This is not bragging, these are not matters of esoteric specialization. These are mundane facts, who worked with whom (verifiable, but as with Malvern / RRE / TRE not remembered by anyone else). I cannot compromise intellectual integrity by linking to articles that I consider unsatisfactory. Dealing with this, and coping with wiki legalisms seems to be taking more time than putting in substantive material. Michael P. Barnett (talk) —Preceding undated comment added 01:46, 12 February 2011 (UTC).

Hi Mr. Barrett!

I think that your suggestions were valuable and do merit consideration, but I think that most of the material would be better as part of the body of the article, rather than as part of the lead, which is supposed to summarize the body of the article (and not introduce new information). I wrote the "chip on your shoulder", about only one line in your comments (which made me feel defensive, I admit!): I and others recognize the sincerity and good will in your comments, and your courtesy in writing the drafts of the lead on the talk page, where you solicited comments in an exemplary fashion. In my experience, it was easier to contribute to specialty articles before modifying core articles, and I would imagine that you would find it more enjoyable to get more experience with articles on mathematical chemistry, where you seem to have expert knowledge. You certainly write well and carefully, and are able to reference your statements admirably.

I am delighted to read about your age and experiences, which are very welcome. I suspect that WP prose and content suffer from being written by a rather narrow range of editors, with less skill in writing.

Best regards, Kiefer.Wolfowitz (talk) 11:20, 12 February 2011 (UTC)

Ordinal optimization

Dear Kiefer.Wolfowitz,

How are you doing?

This is Samuel. I have been working on ordinal optimization for a while. Recently, I read your explanation of “ordinal optimization” (OO) in wikipedia (http://en.wikipedia.org/wiki/Ordinal_optimization). It seems that my understanding of OO is different from the description in that page. Thus I am writing to discuss this with you.

It seems that you used the term “ordinal optimization” to describe optimization on partially ordered set, which I agree is different from optimization on totally ordered set and thus deserves to be studied. However, in the area of discrete event dynamic system (DEDS), the term of OO has a different story.

DEDS’s refer to systems, the dynamics of which follow not only physical laws but also man-made rules. Power grid, transportation systems, and Internet are examples of DEDS. In a DEDS, the performance of a solution candidate can only be evaluated through simulation, which is usually time-consuming and noisy. And, the total number of solution candidates usually grows exponentially fast with respect to the system scale. Thus the simulation-based performance evaluation and the large (and discrete) search space make the optimization of a DEDS different from many traditional optimization problems, where the objective functions are shown in closed forms and the difficulty to evaluate a solution candidate is usually ignored.

Ordinal optimization was introduced in the area of DEDS in 1992 [1] to handle such simulation-based optimization problems. Since finding the optimal solution in a DEDS with practical scale is usually computationally infeasible, the basic idea of OO is “instead of finding the best for sure, OO finds a good enough solution with high probability”[2]. Because it is difficult to accurately evaluate the Cardinal values of solution candidates in such problems, the above basic idea emphasizes on clarifying the relative Orders among the solution candidates instead. So this method is called “Ordinal Optimization” to distinguish from previous cardinal optimization methods. The advantage of this idea has been mathematically justified [3,4,5] and demonstrated in many applications, which can be found in a most recent book on OO[2].

The above story is missing in page http://en.wikipedia.org/wiki/Ordinal_optimization.

It might be more appropriate to explain “Ordinal Optimization” using the above story. The current page http://en.wikipedia.org/wiki/Ordinal_optimization, instead, would be a wonderful explanation for “Optimization on Partially Ordered Sets”. Please let me know if you agree with my suggestion. Looking forward to hearing from you.

Best regards,

Samuel Newsamuel (talk) 14:16, 20 February 2011 (UTC)

References

[1] Ho, Y.C., Sreenivas, R., Vakili, P.,"Ordinal Optimization of Discrete Event Dynamic Systems", J. of DEDS 2(2), 61-88, (1992).

[2] Ho, Y.C., Zhao, Q.C., and Jia, Q.S., Ordinal Optimization: Soft Optimization for Hard Problems, Springer, 2007.

[3] Dai, L., “Convergence properties of ordinal comparison in simulation of discrete event dynamic systems,” Journal of Optimization Theory and Applications, 91(2): 363-388, 1996.

[4] Xie, X.L., “Dynamics and convergence rate of ordinal comparison of stochastic discrete-event systems,” IEEE Transactions on Automatic Control, 42(4): 586-590, 1997.

[5] Lee, L.H., Lau, T.W.E., and Ho, Y.C., “Explanation of goal softening in ordinal optimization,” IEEE Transactions on Automatic Control, 44(1): 94-99, 1999.

— Preceding unsigned comment added by Newsamuel (talk • contribs) 14:13, 20 February 2011 (UTC)

Dear Samuel,

Thanks for your thoughtful and polite comments, and the opportunity for discussion. You know a lot more than I do, it seems!

:-)

The phrase "ordinal optimization" is natural for optimization on posets, and I am glad that Professor Ho coined the phrase (as I remember from my quick Google search months ago).

WP articles must place their topics in context. The natural context for your stub was an article on optimization on posets, which I pieced together in a few hours. I know that anti-matroids (convex geometries), greedoids, sub-/super-modular functions are very important in network stochastics and optimization, and that such results should be mentioned along with Ho's work. Finally, there needed to be mentioned the huge literature on selection problems in statistics, which are formally similar to Ho's problems, although his website claimed that they are limited to small problems, if my memory is correct.

It might be that the existing article on "ordinal optimization" would be better moved to an article entitled "optimization on posets" or "lattice optimization" (but then I may have forgot Veinott, Rothblum, and Stanfordesque economists!).

Maybe Ho's methods deserve their own article, but then the article needs to be a proper article, mentioning related results and providing some context, even more than would a review article for other professionals---since WP tries to have a broad audience.

Please forgive me for offering some unsolicited advice, which need not be relevant now. At least for future reference, it may be a good idea for you to review the WP policy on conflicts of interest. This may be useful if you write about your own research in the future.

Best regards, Kiefer.Wolfowitz (Discussion) 14:34, 20 February 2011 (UTC)

I looked again at your description, which is too close to a promotional blurb for Ho's work, for inclusion as an encyclopedia article. Your feature of Ho's methods might be better in an article on "optimization in DEDS systems", which should mention related work, e.g. starting with the references I provided in my quick expansion of your stub. Kiefer.Wolfowitz (Discussion) 14:43, 20 February 2011 (UTC)

Try to find secondary references that refer to Ho's "ordinal optimization" and establish its notability. Has this been the subject of an invited review article in a leading journal? Have other leading researchers written that it is an important result? Look at the WP policies about references. It is not sufficient to refer to Ho and his students' work. Secondary sources are needed, at least for establishing notability. Kiefer.Wolfowitz (Discussion) 14:48, 20 February 2011 (UTC)

A refreshing suggestion

Hope you enjoy your break, big guy. You're entitled (/ ). --Thomasmeeks (talk) 16:35, 25 January 2011 (UTC)

\

OOPS! Now I feel guilty .... Maybe I can get an administrator to delete my message before you read it! (What are you referring to? I've been writing to get the SF lemma done.) Sorry, Kiefer.Wolfowitz (talk) 17:07, 25 January 2011 (UTC)

For what? I didn't see it, don't care, and won't look (& if I did, I'd probably smile, but I won't). I happened to look at your contribs log & found it hard to believe that one person could edit that long. So, I thought you were doing something un-WP-else (for a while). I felt perky enough to drop a note. Good luck on the article. Unseriously, I trust that you'll follow my commands on that exactly. As air pilots say (at least in the movies), over and out, Thomasmeeks (talk) 17:33, 25 January 2011 (UTC)

The women in my family have the strength of 10 healthy men. I unfortunatly am only like Steve Rogers before he took the super-soldier serum. Kiefer.Wolfowitz (talk) 17:38, 25 January 2011 (UTC)

Actually, the writing marathon has resulted in my first and hopefully my last incidence of "mouse arm". I shall need to reduce editing for the next week. Kiefer.Wolfowitz (talk) 09:31, 27 January 2011 (UTC)

Well, no wonder (per above) — & of which our exchanges might be the smallest contributor.

The best possible outcome IMO would be settling the matter between us in a principled fashion, b/c it would reveal an example of resolution that otherwise might seem nearly impossible, like the (conflatedly) proverbial lion and lamb.[1] That would be a far greater WP contribution than resolution of the immediate issue. I regret that I'm not smart enough to suggest principles that would allow you to do that. (OK, you might say the same to me.) Oh, well.

It is possible that a panel of Econ people would agree with one of your choices. I just believe otherwise. (I could be wrong on that of course.) In the meanwhile I'll plug away and try to make my case as best I can, not by proving that I'm right (a mission impossible), but by pointing to what I believe are strengths and weaknesses of proposed alternatives and arguments for them.

If we were sitting across from each other with a pitcher of beer between us, I think we'd have settled this a long time ago. A pity. P.S. If you have had any 2nd thoughts on your last Talk page Edit, I might welcome your revision. Before (or after) such, I'd be glad to indicate things I (still) find objectionable, some of which you might even agree with. --Thomasmeeks (talk) 15:08, 27 January 2011 (UTC)

Thanks for kind words. I saw this today, which reminded me of an earlier conversation. I won't wikify it, because of mouse-arm. TY - JOUR JO - Journal of Economic Methodology PB - Routledge AU - Wible, James R. TI - Charles Sanders Peirce's economy of research SN - 1350-178X PY - 1994 VL - 1 IS - 1 SP - 135 EP - 160 UR - http://www.informaworld.com/10.1080/13501789400000009

Hi Thomas,

I made some changes to my "barbaric yawp" entry, striking through comments made in especially bad taste. I'm sorry for not censoring myself better the evening when I wrote it very quickly and without sufficient inhibitions. I would be happy to remove stupidities, of course. Again, I'm sorry for misjudgments of tact, which may seem disrespectful, which was not my intention.

Also, I made a suggestion: Could we just call the sidebar-section "Methods" without "technical" or "mathematical" or "scientific", etc.?

Sincerely, Kiefer.Wolfowitz 18:41, 31 January 2011 (UTC)

Hi, KW. OK. Acknowledgement: From Jan. 29 on I was in engaged in formulating a point-by-point response to that entry, which I found in tone and substance nothing like what had gone before or came afterward. (I was so taken by your recent Edits, that I neglected to even Save my last Edit.) The most recent Edits have made moot some (possibly all -- I can't recall) of your points there.

I have another suggestion, which I think would save us both troubling about that further. It is this: simply withdraw what remains of the entire section (by "mutual agreement" or however either of us might like to put it). You could re-introduce any substantive points there elsewhere if you would wish, but in the meanwhile, its removal would allow both of us to move on.§

I appreciate your candor above. Thank you.

§ Because of its duration there -- & for some time after my suggestion above -- with possible prejudicial effects, that sub-sub-section invites revisiting, though I'd hope not soon, & not only the sanitized version. --Thomasmeeks (talk) 00:15, 2 February 2011 (UTC)

Hi Thomas! Thanks for trying to forgive my poor judgment, which was about my worst WP behavior (so far!). Yes, please do revise everything, and cut-out as much as possible! I'm sorry for my Mirowskian "more heat than light" tone, again. Sincerely, Kiefer.Wolfowitz (talk) 00:26, 2 February 2011 (UTC)

Hi Thomas! I had the pleasure of reading a couple of essays on convexity by Peter Newman, whose essays you had commended earlier. I added them to the stub-start articles on convexity in economics and non-convexity (economics). Cheers, Kiefer.Wolfowitz (Discussion) 16:05, 15 February 2011 (UTC)

Newman's essays were quite accessible and I don't see why the new editor (Bruce E. Lawrence) replaced them with his own. Maybe because they were criticized as too mathematical by Stigler's review in JEL---an unfair criticism imho. (Maybe Stigler confused Newman's essay with Stephen M. Robinson's brief but profound article on convex programming, which was too dense and abstract for civilians and even paraprofessionals!) Kiefer.Wolfowitz (Discussion) 20:08, 20 February 2011 (UTC)

Logarithms and likelihoods

Log likelihood

Hi, you added some stuff on the log likelihood to the logarithm article. I find this piece of the article difficult to penetrate, especially for the targeted audience. I'm preparing a FA candidacy for the article. If you have a spare moment, could you maybe brush over this paragraph with a special emphasis on using as little jargon as possible? (If not, no prob, I can also do it myself later). Thanks and greetings, Jakob.scholbach (talk) 22:21, 15 February 2011 (UTC)

Hi Jakob! I'll look at it in 2 hours. (There were some inaccuracies in the previous treatment.) Please copy the paragraph here, striking-through anything impenetrable. Thanks! Kiefer.Wolfowitz (Discussion) 22:32, 15 February 2011 (UTC)

Thanks, but do take your time. Not urgent at all. Jakob.scholbach (talk) 22:55, 15 February 2011 (UTC)

Only Handle It Once (OHIO)! I'm on it! Kiefer.Wolfowitz (Discussion) 23:06, 15 February 2011 (UTC)

Jakob, I tried to copy edit the surrounding material.

Regarding the MLE, I don't think that the account can be simplified, because each step is essential. Your audience doesn't understand that the log is isotone and so any maximum can be pulled back to the original scale, for example. Sorry for not simplifying the discussion.

It would be better to give a simple example, for example, the likelihood function of two i.i.d. Bernoulli random-variables, which is the likelihood of the binomial distribution. You just discussed toin-cossing, and people learn from examples better than from abstract discussions. It's late here and I must sleep.

Best regards, Kiefer.Wolfowitz (Discussion) 23:32, 15 February 2011 (UTC)

Let me quote from the article on the likelihood function:

Log-Likelihood

For many applications involving likelihood functions, it is more convenient to work in terms of the natural logarithm of the likelihood function, called the log-likelihood, than in terms of the likelihood function itself. Because the logarithm is a monotonically increasing function, the logarithm of a function achieves its maximum value at the same points as the function itself, and hence the log-likelihood can be used in place of the likelihood in maximum likelihood estimation and related techniques. Finding the maximum of a function often involves taking the derivative of a function and solving for the parameter being maximized, and this is often easier when the function being maximized is a log-likelihood rather than the original likelihood function.

For example, some likelihood functions are for the parameters that explain a collection of statistically independent observations. In such a situation, the likelihood function factors into a product of individual likelihood functions. The logarithm of this product is a sum of individual logarithms, and the derivative of a sum of terms is often easier to compute than the derivative of a product. In addition, several common distributions have likelihood functions that contain products of factors involving exponentiation. The logarithm of such a function is a sum of products, again easier to differentiate than the original function.

As an example, consider the gamma distribution, whose likelihood function is

L(\alpha ,\beta |x)={\frac {\beta ^{\alpha }}{\Gamma (\alpha )}}x^{\alpha -1}e^{-\beta x}

and suppose we wish to find the maximum likelihood estimate of β for a single observed value x. This function looks rather daunting. Its logarithm, however, is much simpler to work with:

\log L(\alpha ,\beta |x)=\alpha \log \beta -\log \Gamma (\alpha )+(\alpha -1)\log x-\beta x\,.

The partial derivative with respect to β is simply

{\frac {\partial \log L(\alpha ,\beta |x)}{\partial \beta }}={\frac {\alpha }{\beta }}-x

If there are a number of samples x₁,…,x_n, then the joint log-likelihood will be the sum of individual log-likelihoods, and the derivative of this sum will be the sum of individual derivatives:

{\frac {n\alpha }{\beta }}-\sum _{i=1}^{n}x_{i}

Setting this equal to zero and solving for $\beta$ yields

\beta ^{*}={\frac {\alpha }{\bar {x}}}

where $\beta ^{*}\,$ represents the maximum-likelihood estimate and ${\bar {x}}={\frac {1}{n}}\sum _{i=1}^{n}x_{i}$ is the sample mean of the observations.

Cavalry

I alerted the WP project on Statistics, about your request for help.

Best regards, Kiefer.Wolfowitz (Discussion) 00:10, 16 February 2011 (UTC)

GA problem

Hi Jakob! I like your edit, but you may have trouble because you no longer define (parenthetically, for example) the likelihood function. Kiefer.Wolfowitz (Discussion) 01:14, 22 February 2011 (UTC)

Logarithm psychology

I undid your edits to Logarithm#Psychology, as the primary source didn't seem to support it and I could get easy access to the others. And I didn't understand the point or why it was worded so oddly (maybe you're not a native English speaker?). It looked to me like they said there is not a "difference threshold"; this is quite a different thing from saying that a threshold, however defined, is proportional to intensity and thereby defines a log scale. Anyway, if you can quote some relevant bits out of your sources, I'll be happy to help craft an edit that will reflect the point. Dicklyon (talk) 05:48, 17 February 2011 (UTC)

Hi Dycklyon.

Obviously, if the claimed "threshold" doesn't exist, then a claim of "its" proportionality to the intensity is false. (The experiment allows that a smaller thresh-hold exist, I note.) However, I reformulated the paragraph to avoid a distraction about Weber's mistakes.

Weber's difference equation needs to be reformulated as a differential equation, for the paragraph's close to make sense. I wrote a quick repair.

The articles are available on JSTOR, for which you can contact your local library or university.

You are the first WP editor to comment on my failings at English: Your substitution of "as" for "because" was an unhappy edit. However, you were right that "yield" is a mathematical cliché; I wrote better phrase.

Thanks! Best regards, Kiefer.Wolfowitz (Discussion) 11:44, 17 February 2011 (UTC)

The phrase that puzzled me was "that did support the power law". If you thought the power law was important enough to emphasize this way, why did you then take it out? And I don't agree with your logic. The conclusion about sensation being logarithmic in intensity, as Fechner claimed, really didn't hinge on the existence of a threshold, and your application of that paper seemed to be a case of WP:SYN. Dicklyon (talk) 16:34, 17 February 2011 (UTC)

I shudder reading the anemic phrase "that did support the power law", about which you rightly complained!

I thought that I had made the point about the distraction of the "least detectable thresh-hold" earlier. Peirce & Jastrow killed Fechner's estimate and Peirce disliked the idea of such a thresh-hold on his philosophic grounds, believing in continuity, I'll mention. Peirce and Jastrow don't criticize the logarithmic form as an empirical fact, but obviously supported it, there and in many other writings. My citation was not synthesis, but lack of ignorance! Peirce-Jastrow is famous as is Peirce's support of the logarithmic law, which is discussed by Jack Good.

For some reason, the WP article on the likelihood function attributes Peirce's ideas (and those of Allan Turing and Jack Good) to the arch-angel A. W. F. Edwards who ascends ever higher in the heavenly hierarchy of the Ronald A. Fisher cult. Smashing the idolatry of Fisher must be the work for iconoclastic youth, however, because my time grows dearer with the hour.

Best regards, Kiefer.Wolfowitz (Discussion) 16:53, 17 February 2011 (UTC)

Are there good secondary sources on this history that you care so much about? Dicklyon (talk) 21:56, 17 February 2011 (UTC)

Secondary, reliable sources on Peirce-Jastrow include the articles noted on randomized experiments and psycho-physics: Hacking, Stigler, and Dupue, for starters.

Jack Good's collection Good Thinking (republished recently, I believe) and other writings has a lot of references to Peirce and Turing and Good.

BTW, I am a fan of Edwards, who himself acknowledges predecessors, including the mighty Fisher (who was a great scientist and statistician, but also a over-rated and a jerk, to put it mildly).

Cheers, Kiefer.Wolfowitz (Discussion) 22:42, 17 February 2011 (UTC)

WP:WPM interview

Update: Thanks for participating in the interview. Just a heads up that section editor Mabeenot, has move the publication date to this coming Monday, 21 February. The final draft has now been posted. Please go through it to check for any inaccuracies, etc. Thanks again. – SMasters (talk) 23:48, 16 February 2011 (UTC)

Thanks for the alert and for your work.

I have only two comments, both about the two-paragraph quotation from my remarks:

My two-paragraph quotation gives me too much exposure, because the other editors have done far more work.
Following the quote, the statement "This is not necessarily a bad thing" is hard for me to understand. To what does "this" refer?

Thanks again for your work! Best regards, Kiefer.Wolfowitz (Discussion) 12:16, 17 February 2011 (UTC)

Thanks for your feedback. I have (hopefully) clarified the sentence. If you have any more comments, please put them on my talk page as it's difficult for me to monitor multiple talk pages, and I would like to keep all comments on this in the same place. Cheers. – SMasters (talk) 12:33, 17 February 2011 (UTC)

Replacing "this" by "the lack of FA articles" illuminated Charles's comments. Thanks. Kiefer.Wolfowitz (Discussion) 12:38, 17 February 2011 (UTC)

Signpost interview

The Signpost published its interview about WikiProject Mathematics.

As usual, the newest and least productive of the responding editors was not shy! Kiefer.Wolfowitz (Discussion) 19:39, 22 February 2011 (UTC)

Hi

Hi, I joined the mathematics and the statistics project groups. I'll also promote them to people like Olle Hagstrom. Do you know StatProb and Citizendium?

Richard Gill (talk) 10:25, 22 February 2011 (UTC) (Signed by K.W., to prevent a trout-slapping by sine.bot)

Hi Richard!

It's great that you joined the projects. You will enjoy some of the discussions.

Thanks for telling me about the StatProb, which looks very exciting: I shall have to see if it is related to Scholarpedia. StatProb should have better quality control than Wikipedia, and its consistent use of LaTeX will make it much easier to upload material. (On my FireFox browser, WP's LaTeX looks very weird.)

It would be great if Olle joined. With his energy and righteous zeal (said approvingly), he could be a one-man probability-and-statistics and science WikiProject! Kiefer.Wolfowitz (Discussion) 09:49, 24 February 2011 (UTC)

P.S. You were right about Calvinism being more important than Luther in terms of explaining obedience to authority. Helen Fein's macro-sociological Accounting for Genocide suggested that bureaucratic authority and efficiency accounted for (too many of) Holland's civil-servants' expediting the Holocaust (compared to neighboring countries). I have trouble explaining the wonderful Dutch football by Calvin, though!

From the Westminster confessions: Q. 127. What is the honour that inferiors owe to their superiors? A. The honour which inferiors owe to their superiors is, all due reverence in heart,658 word, 659 and behaviour;660 prayer and thanksgiving for them;661 imitation of their virtues and graces;662 willing obedience to their lawful commands and counsels;663 due submission to their corrections;664 fidelity to,665 defence,666 and maintenance of their persons and authority, according to their several ranks, and the nature of their places;667 bearing with their infirmities, and covering them in love,668 that so they may be an honour to them and to their government.669

I would suggest not quoting this at the MHP arbitration board! ;-) Kiefer.Wolfowitz (Discussion) 13:52, 24 February 2011 (UTC)