Wikipedia:Articles for deletion/PDFO

{{AFD help}}

:{{la|PDFO}} – (View AfDView log{{int:dot-separator}} [https://tools.wmflabs.org/jackbot/snottywong/cgi-bin/votecounter.cgi?page=Wikipedia:Articles_for_deletion/PDFO Stats])

:({{Find sources AFD|PDFO}})

Article about a software recently released; the article does not show any notability but a link to a page listing it together with other analogous software. User created the same article on it.wiki, es.wiki, fr.wiki (in all three it has been deleted), zh.wiki and de.wiki.

The {{tl|proposed deletion}} template was removed first by an IP editor, then put again by me (for which I apologize, I am not used to en.wiki's customs) and then removed again by the authorDr Zimbu (talk) 16:39, 3 July 2020 (UTC)

Arguments of Zhangzk

I am the one who created the article on en.wiki, zh.wiki, and de.wiki. However, different from being claimed, I did not create the fr.wiki article. Whoever created it, un grand merci. Unfortunately, the fr.wiki page has been lost due to a speedy deletion that is unjustifiable. It is a pity as the page was well edited.

First of all, thank Dr Zimbu for the efforts to maintain the high quality of Wikipedia, which I do appreciate.

The page should be kept. Here are the reasons.

• This is not really a "software recently released". It does nothing but collecting Michael J. D. Powell's original code (COBYLA, UOBYQA, NEWUOA, BOBYQA, LINCOA). The code has existed for a long time with broad applications.

• The cited reference "Decision Tree for Optimization Software" {{cite web|url=http://plato.asu.edu/sub/nlores.html#general|title= Decision Tree for Optimization Software|last=Mittelmann|first=Hans D.|accessdate=2020-07-03}} is not just "a page listing it together with other analogous software" but an authoritative page maintained by a leading expert in computational optimization (Dr. Hans D. Mittelmann {{cite web|url=http://plato.asu.edu/|title=Homepage of Prof. Dr. Hans D. Mittelmann|accessdate=2020-07-04}}, Full Professor at School of Mathematical and Statistical Sciences, Arizona State University{{cite web|url=https://asu.pure.elsevier.com/en/persons/hans-mittelmann|title=Arizona State University expertise finder database|accessdate=2020-07-04}}) that makes recommendations about optimization software. See "Benchmarking Optimization Software - a (Hi)Story"{{cite news |last=Mittelmann|first=Hans D.|title=Benchmarking Optimization Software - a (Hi)Story|url=https://link.springer.com/article/10.1007/s43069-020-0002-0|date=2020|volume=1|journal=SN Oper. Res. Forum}} for more information. Anyone in related research areas should be clear about this. See also the list of Derivative-Free Optimization solvers maintained by Prof. S. Le Digabel {{cite web|url=https://www.gerad.ca/Sebastien.Le.Digabel/MTH8418/|title=Prof. S. Le Digabel's Derivative-Free Optimization (DFO) Course Website|accessdate=2020-07-04}}, a leading expert in derivative-free optimization.

• The software is cited in articles published in reputable journals,e.g., Journal of Global Optimization {{cite paper | last1=Xi|first1=Min|last2=Sun|first2=Wenyu|last3=Chen|first3=Yannan|last4=Sun|first4=Hailin|title=A derivative-free algorithm for spherically constrained optimization| journal=Journal of Global Optimization|year=2020|accessdate=2020-07-03|url=https://link.springer.com/article/10.1007/s10898-020-00875-2|volume=76|pages=841-861}}. If deemed necessary, emails can be provided upon request in order to prove that more researchers/engineers are citing/using this software, but it cannot be done here since emails contain personal information.

• The software has been proved useful to the research community, which can be inferred from the download statistics. See, e.g., PePy {{cite web|url=https://pepy.tech/project/pdfo|title=PyPI Download Statistics by PePy|accessdate=2020-07-04}} and the download count recorded on PDFO homepage {{cite web|url=https://www.pdfo.net/docs.html#download|title=PDFO Download Count|accessdate=2020-07-04}}. It is worth noting that PDFO is not a piece of software for daily activities like online shopping or distant meeting --- this software is for scientific research. There is a scientific research project behind each download, so even 100 downloads are indeed a lot --- this software has more than 1000. In addition, some online discussions about related algorithms actually link to this Wikipedia page {{cite web|url=https://news.ycombinator.com/item?id=23004026|title=More Descent, Less Gradient|accessdate=2020-07-04}}.

• The software has been covered in reputable third-party public media. For example, those who know Chinese may read the last section of The Optimized Life of a "Treasure Mathematician" {{cite web|title=专访: 宝藏数学家的优化人生 --- 知乎专栏「运筹OR帷幄」大数据时代的运筹学|accessdate=2020-07-04|url=https://zhuanlan.zhihu.com/p/137646015}}. Surely, this is not an article published in a newspaper or broadcasted in a radio program, but it seems that we are not anymore in the era that public media means only newspapers and radios. The information on Zhihu {{cite web| url=https://zhuanlan.zhihu.com/c_1078650287417487360|title= 知乎专栏「运筹OR帷幄」大数据时代的运筹学|accessdate=2020-07-04}} and its homepage {{cite web|url=https://www.jiqizhixin.com/users/1228d2ca-0acf-4169-8674-5eafc66caf7b|title=运筹OR帷幄 机器之心|accessdate=2020-07-04}} show clearly that it is indeed a reputable and professional public media. These references are relevant since it is mentioned that PDFO has a zh.wiki page.

Thank you very much.

{{Reflist}}

• Delete:WP:CITESPAM. Add Powell, M. J. D pages almost in the article, please also check all edits of this user. Willy1018(talk) 01:39, 6 July 2020 (UTC)
• Keep: Thank you for the checking. Sorry if I understood something wrongly, but these algorithms are among the most well-known works of Michael J. D. Powell. Why is it inappropriate to link to Michael's page? Thanks again. Zhangzk (talk) 02:40, 6 July 2020 (UTC)
• Merge to Michael J. D. Powell or Derivative-free optimization - presumably the latter. There is nothing notable about this package; it's just a software implementation of methods discussed elsewhere. No reason why this software shouldn't be covered in context. The article tellingly spends 3/4 of the text and refs on background and generalities; there's no case for a separate article here. --Elmidae (talk · contribs) 01:53, 11 July 2020 (UTC)
• :Thank Elmidae for the comments. Even though I do not agree that "the article tellingly spends 3/4 of the text and refs on background and generalities", I made some revisions to make it sound less like that. The notability of Powell's software is quite clear from the references mentioned above and a google search. Talking about merging, I would say it is more reasonable to keep this page while merging pages like COBYLA, UOBYQA, etc. into it. Thanks. Zhangzk (talk) 15:27, 11 July 2020 (UTC)

:::Note you already "voted" keep above, no need to "vote" keep twice (you may use eg. "Comment" instead). Pavlor (talk) 15:38, 11 July 2020 (UTC)

::::Thanks. Pavlor (talk) 16:17, 11 July 2020 (UTC)

:::Some bundling of those articles might actually be a really good idea - not sure why they ended up with articles of their own. But I don't believe that should be under the name of that package, unless that is the ONLY implementation(which is probablynot the case?) Still looks as if the best place for all that would be the existing Derivative-free optimization. --Elmidae (talk · contribs) 15:49, 11 July 2020 (UTC)

:::: Thank Pavlor for the correction. Derivative-free optimization is a vast field. If one merges the mentioned pages into derivative-free optimization, then many many other pages would be merged as well and we would end up with a huge page of derivative-free optimization. It is like to merge all pages on various Indo-European languages into one page. This software is the only implementation that synthesizes all of Powell's solvers with a uniform interface. Maybe I should explain more about these solvers: a person that is not in this field may think that each of Powell's solvers takes at most hours of coding, so anyone can implement her/his own version easily. (Un)Fortunately, that's not the case. Each of them took Powell years to implement, leading to tens of thousands of lines of Fortran code. Each of them is a masterpiece of Powell (for the role of Powell in numerical optimization, see his en.wiki page; in short, he was a founding father). That's why the solvers are powerful and impactive, and that's why this software is not unnotable. Thanks. Zhangzk (talk) 16:04, 11 July 2020 (UTC)

::::: Hmm. Looking through these individual algorithms, they are sizable but could well be bundled. As a reader I would certainly prefer to have these in one place rather than spread among half a dozen articles. And that bundle would make an obvious place to discuss the software too. How about derivative-free optimization algorithms, with subsections for TOLMIN, COBYLA, etc., and a final one for PDFO? --Elmidae (talk · contribs) 16:34, 11 July 2020 (UTC)

:::::: Thank Elmidae for the suggestion. However, as I mentioned, derivative-free optimization is a vast field, including numerous algorithms. I dare not create a page called derivative-free optimization algorithms covering only Powell's solvers. No ... that will definitely invoke a lot of enemies (lol). Thanks anyway. Zhangzk (talk) 16:39, 11 July 2020 (UTC)

:Note: This discussion has been included in the list of Software-related deletion discussions. Lightburst (talk) 21:32, 13 July 2020 (UTC)

:Note: This discussion has been included in the list of Computing-related deletion discussions. Lightburst (talk) 21:34, 13 July 2020 (UTC)

• Delete: none of the sources appear to contribute towards WP:GNG. In fact, most of them do not even mentioned the software suite PDFO. On the whole, the subject seems adjacent to notable topics: Powell himself is notable, as may be some of the algorithms he wrote. But the notability of this particular software is not evidenced by the sources (let me know if any of these assessments look incorrect).{{ source assess table |

{{SA

| src = [https://www.pdfo.net/ PDFO.net]

| i = n | ij = official website of the PDFO project

}}

{{SA

| src = [http://plato.asu.edu/sub/nlores.html#general Decision Tree for Optimization Software]

| i = y

| r = - | rj = personal website, but author is an expert in the field

| s = n | sj = long list of software with single sentence on PDFO

}}

{{SA

| src = [http://www.damtp.cam.ac.uk/user/na/NA_papers/NA2007_03.pdf "A view of algorithms for optimization without derivatives"]

| i = n | ij = Powell is the author of algorithms comprising PDFO

}}

{{SA

| src = Conn, Scheinber, Vicente Introduction to Derivative-Free Optimization

| i = y

| r = y

| s = n | sj = "PDFO" is [https://books.google.com/books?id=tGbUshriSyYC&q=PDFO#v=onepage&q=PDFO&f=false not mentioned] in the book

}}

{{SA

| src = Audet & Harre Derivative-Free and Blackbox Optimization

| i = y

| r = y

| s = n | sj = "PDFO" is [https://books.google.com/books?id=ejVBDwAAQBAJ&q=PDFO#v=onepage&q=PDFO&f=false not mentioned] in the book

}}

{{SA

| src= Larson, Menickelly, Wild [https://www.cambridge.org/core/journals/acta-numerica/article/derivativefree-optimization-methods/84479E2B03A9BFFE0F9CD46CF9FCD289 "Derivative-free optimization methods"]

| i = y

| r = y

| s = n | sj = "PDFO" is not mentioned in the journal article

}}

{{SA

| src=[https://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.28.1756 UOBYQA paper]

| i = n | ij = another publication of Powell

}}

{{SA

|src=[http://www.damtp.cam.ac.uk/user/na/NA_papers/NA2004_08.pdf NEWUOA paper]

| i = n |ij = another publication of Powell

}}

{{SA

|src=[http://www.damtp.cam.ac.uk/user/na/NA_papers/NA2009_06.pdf BOBYQA paper]

| i = n |ij=another publication of Powell

}}

{{SA

|src=[https://link.springer.com/chapter/10.1007/978-94-015-8330-5_4 A Direct Search Optimization Method...]

| i = n |ij=another publication of Powell

}}

{{SA

|src=[https://link.springer.com/article/10.1007/s10898-020-00875-2 "A derivative-free algorithm for spherically constrained optimization"]

|i = y

|r=y

|s=n |sj="PDFO" is not mentioned in journal article

}}

}}

I also object to all the mergers proposed above. This is a single software implementation of Powell's algorithms. Since he was not involved in the project at all, it would be inappropriate to merge it to his article. Merging to Derivative-free optimization is also too general: there are dozens of implementations of derivative-free optimization algorithms, this is just one of them. BenKuykendall (talk) 06:21, 14 July 2020 (UTC)

• Delete Ben raises many good points. The article fails to pass WP:GNG, and the author's arguments are irrelevant to notability. Note: the author of the article also appears to be the owner of the linked GitHub repo, so there may be a conflict of interest here. – FenixFeather ^{(talk)(Contribs)} 23:28, 14 July 2020 (UTC)

{{clear}}

:The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.