# Difference between revisions of "Talk:EoM:This project"

This project is based on an electronic version of the "Encyclopaedia of Mathematics", published by Kluwer Academic Publishers until 2003, and by Springer after that. The encyclopaedia goes back to the Soviet Matematicheskaya entsiklopediya (1977), originally edited by Ivan Matveevich Vinogradov.

The electronic version had its formulae written in $\rm \TeX$, which were saved as png images. On its way through the various publishers the original $\rm \TeX$ source code was lost, therefore, to edit a formula in one of these original pages requires to retype the code for that formula from scratch.

For the project, it will be of big help to transcribe the old pages. To make this easy, it was decided to use MathJax, which allows to use Plain $\rm \TeX$ or $\rm \LaTeX$ for formulae encoding.

Some help on this is given on the help pages.

The installed software and extensions for this project can be seen here: Special:Version. This collection may be updated and extended.

This is a useful page, but it should be more visible. Maybe the navigation toolbar should contain it (among "main page", ..., "help"). --Boris Tsirelson 08:14, 14 December 2011 (CET)

The footer "How to Cite this Entry" is appropriate for articles, but probably not for user pages etc. --Boris Tsirelson 15:05, 14 December 2011 (CET)

I have transmitted your proposals to the admins. Ulf Rehmann 22:44, 14 December 2011 (CET)

I rewrote Measurable space; please look. In particular, in the bibliography I gave links to MathSciNet (following a suggestion by Ulf Rehmann); but for now they use my private template User:Boris Tsirelson/MR; if you like it, please move it into Template:MR (with needed changes, if any). --Boris Tsirelson 21:43, 20 December 2011 (CET)

Ok, I have made MR a global template and have added a ZBL template as well, see Measurable space. --Ulf Rehmann 22:53, 21 December 2011 (CET) Same for Tamagawa number --Ulf Rehmann 00:38, 23 December 2011 (CET)
Good news: even working with no subscriptions, clicking a Zbl link I still get a review, in contrast to an MR link (MR says: "For users without a MathSciNet license, Relay Station allows linking from MR numbers in online mathematical literature directly to electronic journals and original articles. Subscribers receive the added value of full MathSciNet reviews.)".
That corresponds to Zbl's policy: They do allow up to three requests even to non subscribers. --Ulf Rehmann 10:40, 23 December 2011 (CET)
Bad news: even working with an MSC subscription (via my university), clicking a Subject Classification link I get:
A username and password are being requested by http://ams.math.uni-bielefeld.de. The site says: "MathSciNet Authentication".
--Boris Tsirelson 10:13, 23 December 2011 (CET)
That is different here: Under a vpn connection I do get access -- even from my home notebook, and I can switch on and off that access by using or not using vpn (as it should happen). But: after switching on, I have to reload the EoM page before my browser realizes that access is given.--Ulf Rehmann 10:40, 23 December 2011 (CET)
And still, the MSC template is not good for primary-only cases, because the "ParserFunctions" extension is missing. --Boris Tsirelson 10:21, 23 December 2011 (CET)
Unfortunately that is true. An updated software version which should fix this hopefully will be installed in early January. --Ulf Rehmann 18:44, 23 December 2011 (CET)
Here is an MSC template which works correctly, but without #if clause: User:Rehmann/sandbox/MSC --Ulf Rehmann 00:33, 7 January 2012 (CET)
Wow! --Boris Tsirelson 08:16, 7 January 2012 (CET)
MSCwiki links added to help non MSN supported readers (not perfect for secondary MSC)--Ulf Rehmann 14:10, 7 January 2012 (CET)

It seems, most of EoM articles are targeted at graduate math students and professional mathematicians; but some are accessible to an interested layman. I try to collect these; for now, starting on "A", "B". --Boris Tsirelson 10:36, 14 December 2011 (CET)

Should we introduce (two or more) levels, and mark articles accordingly? --Boris Tsirelson 11:17, 15 December 2011 (CET)

I am not sure. Maybe it is good to just have a list (category) of articles which are suited for non professionals, in order to attract such readers. --Ulf Rehmann 13:21, 15 December 2011 (CET)

### Preferred style

Many articles have (near the end) the section "Comments" (and often also the second "References" section after "Comments"). What does it mean? And should we follow this pattern when editing and creating articles? --Boris Tsirelson 11:22, 15 December 2011 (CET)

This is due to the history of the encylopaedia. It underwent various (print and electronic) editions, which not always resulted in a rewrite of an article, but just in an update by amending a comment (sometimes there are several, each with its own reference list). There is no need to follow this pattern now since wikipedia does allow better ways of updating.--Ulf Rehmann 13:21, 15 December 2011 (CET)
For some pages, it may be advisable to split them, or to at least introduce significant section titles. E.g., the page Hilbert theorem has 8 'references' sections, similarly the page Duality. --Ulf Rehmann 21:01, 7 March 2012 (CET)
On Wikipedia, in such cases as "Hilbert theorem" and "Duality" one usually makes a disambiguation page and several articles. --Boris Tsirelson 22:10, 7 March 2012 (CET)
Certainly. --Peter Schmitt 23:09, 7 March 2012 (CET)
Yes, this is one possibility to resolve these things, however this concept should be applied with care. Not every section deserves a single page, some are closely related, and it may be more illustrative e.g., to leave all sections of Singular point describing some singularities on algebraic varieties in one page, but separating the initial section about singular points of functions in another, as well as the parts on differential equations. Nevertheless: Separating things possibly may obscure otherwise obvious similarities, which to observe could give a deeper understanding. -- A practical problem with these pages is that the reference descriptors are not unique, so '[2]' may refer to different reference sources, depending on where it is used in the article. --Ulf Rehmann 00:25, 8 March 2012 (CET)
In such cases the top page can be used to give a (brief) survey pointing to separate pages where the individual cases are treated in more detail (including the references). This will help readers to find the information they are looking for -- either it is the survey they need, or they go to the specific topic to learn more. Long pages are not so convenient to browse on the screen (at least for my taste). --Peter Schmitt 02:06, 8 March 2012 (CET)
Ok, maybe this one will become your favoured: Serial scheme. :) --Ulf Rehmann 13:30, 8 March 2012 (CET)

#### Style of title and definition

I guess, some day the use of our new templates will become a part of recommendations posted on our help pages. Here is another question of style.

I observe a typical pattern in the existing articles: the title, mentioning a notion, is immediately continued by the definition of the notion. Sometimes a slanted text in between gives some context or some synonims.

I propose to use a definition style using a complete sentence with predicate consisting of copula+predicative, as this is common in (written) English. Synonyms could be given, maybe after a prefix "also:" and with an extra page for that synonym referencing to the actual page.Context extensions of some notion should be given within the text, if there is no need to distinguish against other uses of that notion, in which case a disambiguation page should be used. See below my suggestions.--Ulf Rehmann 19:43, 23 December 2011 (CET)
Also: subtitles should go over MSC entry --Ulf Rehmann 00:56, 25 April 2012 (CEST)

Examples:

Binary relation
A two-place [[Predicate|predicate]] on a given set.
Preferable: A binary relation is a two-place [[Predicate|predicate]] on a given set.--Ulf Rehmann 19:43, 23 December 2011 (CET)
Spectral density
of a stationary stochastic process or of a homogeneous random field in n-dimensional space
The [[Fourier transform|Fourier transform]] of the covariance function of a stochastic process which is stationary in the wide sense...
Preferable: In this article, the spectral density of a stationary stochastic process or of a homogeneous random field in n-dimensional space is described.--Ulf Rehmann 19:43, 23 December 2011 (CET)
Inter-quantile width
inter-quantile distance, inter-quantile range
Preferable: Also: inter-quantile distance, inter-quantile range --Ulf Rehmann 19:43, 23 December 2011 (CET)
The difference between the lower and upper quantiles of the same level (cf. [[Quantile|Quantile]]).

But in rare cases the definition appears much later (as in Berwald connection).

Maybe here a link to the text part giving the formal definition could be placed under the title.--Ulf Rehmann 19:43, 23 December 2011 (CET)

On Wikipedia the pattern of the beginning is different: the title is not a part of the definition; an example:

Binary relation
In [[mathematics]], a '''binary relation on''' a [[set (mathematics)|set]] ''A'' is a collection of [[ordered pair]]s of elements of ''A''.

Thus, what is our preferred style for the start of an article? --Boris Tsirelson 17:28, 23 December 2011 (CET)

Since our language is English, we should use full sentences for definitions, as said above.--Ulf Rehmann 19:43, 23 December 2011 (CET)

### Subject classification

What I find more important is to classify articles by MSC (Mathematical Subject Classification) and possibly categorize them according to that. This classification should be done so that it is easily recognizable by bots collecting bibliographic info. Of course MSC is subject to change over time, and there should be a tool to update such a classification easily over the whole collection. (We are lucky, since such a change did just happen via MSC 2000 --> MSC 2010.) Does anybody know if there exist wikimedia extensions which do support a suitable classification (and possibly its update)?--Ulf Rehmann 13:21, 15 December 2011 (CET)

For now I added MSC codes to our categories. It will be much easier to update categories than all articles. About bots collecting bibliographic info, I have no idea. But probably a template can help. Do you know which text should appear on articles to this end? Boris Tsirelson 20:51, 15 December 2011 (CET)
Some html meta tag would do the job like
<meta name="description" content="2010 Mathematics Subject Classifiction 20-XXX" />
and
<meta name="keywords" content="associative algebras, Galois theory, ..." />.
But I think that ought to be done on a pagewise basis. --Ulf Rehmann 12:47, 17 December 2011 (CET)
As an experiment, I added "<nowiki><meta name="keywords" content="probability" /></nowiki>" to the end of "Probability"; it is invisible, but appears on the generated html as "<p><span class="tex2jax_ignore"><nowiki><meta name="keywords" content="probability" /></nowiki></span></p>". Will it do the job? --Boris Tsirelson 16:39, 17 December 2011 (CET)
It seems one cannot put some <meta> tag outside the <body> section of some wiki page (unless there is some wikimedia extension wich helps??). So maybe you just omit the <meta ... /> tag. But apparently at the time being, Google's bots don't visit the EoM pages at all.--Ulf Rehmann 19:33, 25 December 2011 (CET)
Not quite so. I just took an old article (not "Measurable space", this is too recent), asked Google for "interest in abstract differential equations is that the so-called mixed problems" and got the first answer: "Differential equation, abstract - Encyclopedia of Mathematics". --Boris Tsirelson 21:13, 25 December 2011 (CET)
Too slow. On my preprint server, articles are recognized by google within a few days.--Ulf Rehmann 21:49, 25 December 2011 (CET)
It is rumored that Google, seeing a new page, returns to it after some days; if it is changed, Google treats it as volatile, and visits frequently; otherwise it does not. But anyway, (a) I doubt we can hope for a high Google rank, and (b) I doubt we can have many visitors. Just because mathematicians are a small population. Students are a larger population, but these are not our readers, I guess. --Boris Tsirelson 23:04, 25 December 2011 (CET)
But (a),(b) are true for my preprint server as well... --Ulf Rehmann 23:35, 25 December 2011 (CET)
I suddenly realized that MSC template allows only for one entry as the "Primary". Isn't it too restrictive? MathSciNet definitely allows for several "primary" codes for articles, and when we write about notions eventually pertinent to many areas, it becomes simply impossible to give only one key. -- Sergei Yakovenko 08:57, 5 May 2012 (CEST)
Is this really true? See this: They say "one primary and one or more optional secondary classifications". MathSciNet says here (go to "How to use MSC"):
Every item in the MRDB receives precisely one primary classification, which is simply the MSC code that describes its principal contribution. When an item contains several principal contributions to different areas, the primary classification should cover the most important among them.
Can you point to a page for which you'd have more primaries? --Ulf Rehmann 10:27, 5 May 2012 (CEST)

### Categories

Concerning categorization I propose to use the sectioning scheme used by the International Congresses of Mathematicians as shown below (I have added the respective two digit MSC codes - please amend/correct):

ICM sectioning scheme, cf. ICM 2010 Corresponding MSC 2010 two digit codes
1. Category:Logic and foundations 03
2. Category:Algebra 06,08,12,13,15,16,17,18,19,20
3. Category:Number theory 11
4. Category:Algebraic and complex geometry 14
5. Category:Geometry 51,52
6. Category:Topology 54,55,57
7. Category:Lie theory and generalizations 22
8. Category:Analysis 26,28,30,31,32,33,40,41,42,43,44
9. Category:Functional analysis and applications 46,47,48
10. Category:Dynamical systems and ordinary differential equations 34,37 (39)
11. Category:Partial differential equations 35
12. Category:Mathematical physics 70,74,76,78,80,82,83,85
13. Category:Probability and statistics 60,62
14. Category:Combinatorics 05
15. Category:Mathematical aspects of computer science 68
16. Category:Numerical analysis and scientific computing 65,68
17. Category:Control theory and optimization 49,93
18. Category:Mathematics in science and technology 81,86,90,91,92,94
19. Category:Mathematics education and popularization of mathematics 00,97
20. Category:History of mathematics 01
List of preliminary MR, Zbl, and MSC templates. --Ulf Rehmann 23:30, 12 January 2012 (CET)
This is obsolete, pleass see below. --Ulf Rehmann 00:16, 27 January 2012 (CET)
The tableau (or some variants) has been used by IMU for the recent ICM section setup, see ICM 2010. (Cf. also: more verbal, ICM 2006, ICM 1998 --Ulf Rehmann 13:21, 15 December 2011 (CET), modified: 19:33, 25 December 2011 (CET)
A good idea; I shall do. --Boris Tsirelson 15:54, 17 December 2011 (CET)
I've made a template here and used it in Tamagawa number. It would be better to use the template "[ 2010 Mathematics Subject Classification {{{1}}}{{#if:{{{2|}}}|,({{{2}}})|}} ]", since the call {{MSC|11F70}} should not generate "11F70,()". Unfortunately, the "ParserFunctions" extension is still missing.
An "MSC" template is worth using, since (1) changing it one can change the appearance of all MSC at once, and (2) in 2020  :-) one can instruct the template to replace some codes with new ones. --Boris Tsirelson 16:57, 18 December 2011 (CET)
--Boris Tsirelson
I have set up some algebra (etc.) subcategories with some new template User:Rehmann/sandbox/MSCtop‎. (Eventually all the MSC templates should be combined as soon as we will have the ParserFunctions extensions or so.) --Ulf Rehmann 23:29, 7 January 2012 (CET)
We need an msc macro which automatically produces not only the entry+link for the primary/secondary number, but also inserts the respective [[Category:<blah>]] line entries to categorize the pages. But this requires (at least) the ParserFunctions extensions, maybe more.--Ulf Rehmann 00:35, 8 January 2012 (CET)
The existing MSC template does not work correctly for such cases as [ 2010 Mathematics Subject Classification MSN: 60-01 | MSCwiki: 60-01  ]. --Boris Tsirelson 13:09, 12 January 2012 (CET)
Yes, I know. That cannot be resolved without "#if"-clause, so I drafted another template User:Rehmann/sandbox/MSC-X for this special purpose (middle entry an hyphen '-') (and nagged the sysadmins again for installing #if). --Ulf Rehmann 16:44, 12 January 2012 (CET)
Here is a list of all the (preliminary) MR. ZBL, and MSC templates. --Ulf Rehmann 23:26, 12 January 2012 (CET)
ParserFunctions works! Wow! --Boris Tsirelson 09:43, 19 January 2012 (CET)
Yes, that's good. It also offers some nice string handling functions. For further new extensions, see Special:version.--Ulf Rehmann 11:07, 19 January 2012 (CET)
##### New MSC & categorization templates

Here are new categorization templates. It suffices to inlude an MSC entry in a page, then the template will automatically add the respecitve categories to that file. Please use them from now and let me know any problems you might encounter. --Ulf Rehmann 23:41, 26 January 2012 (CET)

Yes, I did it on Absorbing state, and strangely, it got three categories "Markov chains", "Probability and statistics", "Probability theory and stochastic processes". Before it had only the "Markov chains" category (this being a subcategory of "Markov processes" and so on, according to MSC). Is it the intended behavior? --Boris Tsirelson 09:14, 27 January 2012 (CET)
Yes: According to our table above the MSC 60Jxx requires an entry in "13. Category:Probability and statistics". The MSC yields two other categories 60-XX Category:Probability theory and stochastic processes and 60Jxx Category:Markov processes. --Ulf Rehmann 11:17, 27 January 2012 (CET)
Usually (I mean, in Wikipedia) members of a subcategory are not members of the category. See for example [1] and [2]. Otherwise high-level categories will be terribly overpopulated. --Boris Tsirelson 14:45, 27 January 2012 (CET)
That's right. However, right now there is no big danger of any overpopulated categories like that, and it is more cumbersome to click one's your way through all analysis or algebra related subcategories in order to find out what is there. If this changes, it is very easy to remove files of a big subcategory from some supercategory, since this requires only to remove one entry in Template:Category. This is the advantage of a list controlled categorization: Entries are generated "on the fly" every time a page is called.
OK, I see. --Boris Tsirelson 19:10, 27 January 2012 (CET)
Well, "Absorbing state" probably is now as it should be (please check).
And please look at "Category:Distribution theory"; any remarks?
As a consequence, now "Category:Probability and statistics" contains subcategories of different levels: both "Category:Probability theory and stochastic processes" and "Category:Limit theorems"; should it be so?
Maybe not. I had seen this last night as well for Category:Number theory and a few others, my solution was to use MSCwiki for indicating their MSC, which is not a good idea, as this will cause confusion. I now have modified Template:Category so that it does >not< categorize if called from a category page, so those pages have to be categorized by an explicit entry. --Ulf Rehmann 00:43, 28 January 2012 (CET)
(But do not bother about "Category:Probability", I'll delete it, but only after rerouting all links to it.) --Boris Tsirelson 21:16, 27 January 2012 (CET)
I wonder whether such a concept could be introduced for references as well. Generating the pagewise lists over a global list (or several such, the entries being already pre-formated) would ease to control the pagewise reference lists. Certainly many related pages share quite some reference entries. After all, TeX has a similar system (bibtex). --Ulf Rehmann 15:54, 27 January 2012 (CET)
Yes, I recall that such means are used on Citizendium in chemistry and biology articles, but not in math articles. I am not sure that the result is worthy of the effort. I also wonder, in which namespace the bibliographic entries will be. Mainspace? No, it is for articles only. Templates? No, it is editable by admins only. --Boris Tsirelson 19:09, 27 January 2012 (CET)

The page "Entropy theory of a dynamical system" has MSC Primary: 37A35 Secondary: 60G10; and Categories: Probability and statistics, Probability theory and stochastic processes, Dynamical systems and ordinary differential equations, Dynamical systems and ergodic theory, Ergodic theory (in this order). Thus, categories generated from the secondary MSC come first. It would be better to have them last. Or maybe not at all? Maybe only primary MSC should generate categories? --Boris Tsirelson 10:24, 15 February 2012 (CET)

I have changed Template:MSC. The category order for that article now is:
Dynamical systems and ordinary differential equations, Dynamical systems and ergodic theory, Probability and statistics, Probability theory and stochastic processes, Ergodic theory. Is it this what you have in mind?
Much better than before. Ideally, "Ergodic theory" should go before "Probability and statistics", but this can wait.
I think it is useful to also use the secondary msc for categorization: The category list should know all articles which are related to it. --Ulf Rehmann 14:41, 15 February 2012 (CET)
So, you are an inclusionist in this aspect... In Math Reviews (and arXiv emails) secondary articles are listed, but after primary articles, and marked as secondary. Otherwise they would make primary articles less visible. (And our categories will be quite populated in the previsible future.) --Boris Tsirelson 15:36, 15 February 2012 (CET)
There are a few possibilities to achieve this: One is to define, for each category, a "secondary" one with same title (amended by "sec" or so), and being a member of the first ("primary") one, and to list the secondary classified articles there. This could be done by a staightforward extension of the msc template. Concerning your request on "Ergodic theory", you either suggest where to put that Category in our table above, or another augmentation of the msc template might allow to input a further category explicitly. --Ulf Rehmann 17:35, 15 February 2012 (CET)
About secondary categories: yes, I like this idea.
About the place of "Ergodic theory": this is just one of the third-lever categories (MSC|37Axx), if we call the 20 categories of the table (the ICM categories) the first level, the second level being for example MSC|37, and the third level for example MSC|37Axx. In most cases I like to use three levels. --Boris Tsirelson 18:05, 15 February 2012 (CET)
Yes, I see what you mean. The first should be easy, the second would require some thought. The Template:Category is table driven, and an appropriate table to include what you call 3rd level categories has to be set up. I'd prefer to postpone this a bit right now. --Ulf Rehmann 21:14, 15 February 2012 (CET)
Some proposal: It might make sense to prepend the MSC to the names of MSC related categories, so that the name of [[Category:Ergodic theory]] would be "37Axx Ergodic theory". For the categories from our table that could be achieved easily, categorisation would be done automatically via the templates. At the end, a few of the old categories would have to be deleted. --Ulf Rehmann 00:37, 16 February 2012 (CET)
Could be useful (but I do not have a definitive opinion). In general, MSC is only useful for research level topics. But what to do with "basic" notions? Putting, e.g., injection in 03Exx or 26-XX is not really what is needed. --Peter Schmitt 01:34, 16 February 2012 (CET)
Yes, I also noted that MSC is only useful for research level topics, and I am often forced to use MSC|60-01, as in "Bernoulli trials". --Boris Tsirelson 07:01, 16 February 2012 (CET)
About "37Axx Ergodic theory": I am afraid, it is too insistent. If the reader wants to know the MSC of Category:Ergodic theory he/she can see it on the category page. --Boris Tsirelson 14:24, 16 February 2012 (CET)
Why not do it the other way? Take the MSC number as names for the categories, give the full classification on these category pages (avoiding external links), use plain language for non-MSC-induced categories, and use the MSC macro to display the full description. Wouldn't it be enough to put the classification at the end of the article? --Peter Schmitt 01:57, 18 February 2012 (CET)
##### Technical Categories

How about a few "technical categories"? I think of "Category:TeX encoded" and "Category:TeX encoding wanted".

People interested in editing certain pages but technically less skilled could just enter an entry [[Category:TeX encoding wanted]] and advertise these in order to get it converted by some $\rm \TeX$ freak. --Ulf Rehmann 11:52, 27 January 2012 (CET)

No objections. --Boris Tsirelson 14:48, 27 January 2012 (CET)

Here is this tex thing: Template:TEX. The state of the categories needs to to be updated.

One could think of more categories which reflect the state of pages, concerning their actualization state or the state of their references. --Ulf Rehmann 17:42, 29 January 2012 (CET)

##### The tableau again

But why is 37 in "Category:Partial differential equations"? It should be only in Category:Dynamical systems and ordinary differential equations. --Boris Tsirelson 16:55, 28 January 2012 (CET)

Probably because of mutual references between 35, 37, Hamiltonian and these KDF and NLS stuffs. But maybe you are right: if some article belongs into both that should be decided for that article individually. --Ulf Rehmann 19:28, 28 January 2012 (CET)
Yes; now "Approximation by periodic transformations" does not belong to "Category:Partial differential equations", which was rather unexpected (at least to me). --Boris Tsirelson 20:17, 28 January 2012 (CET)
##### What categories?

What categories should be created in addition to those generated by the MSC template? For example: Should there be separate categories for Packings, Tilings, or only a common category, or not even that? --Peter Schmitt 02:41, 9 February 2012 (CET)

"Packing" is 52C15, 52C17; "tiling" is 52C20, 52C22; thus I'd put both into category 52C "Discrete geometry". --Boris Tsirelson 08:09, 9 February 2012 (CET)
Yes, of course, This is the second level MSC classification (and could be deduced/inserted automatically from the classification number). But this is still a rather broad subject. Should there be some finer categorization, not "isomorphic" to the MSC numbers? --Peter Schmitt 11:16, 9 February 2012 (CET)
I propose to discuss this after the existing categories are really a bit more crowded, so that we have a basis of articles on which we can decide what is needed. --Ulf Rehmann 19:23, 9 February 2012 (CET)

Many pages have italic subheads, eg Subgroup, index of a, Ellipse. Should the MSC go above or below it? I would have thought below (since the subhead is part of the title, it elaborates on it, so an MSC above the subhead means splitting the title in two). But most pages seem to have the MSC above the subhead. --Jjg 00:24, 25 April 2012 (CEST)

### References

Now, having ParserFunctions, we can choose a style for refs (and support it by templates).

For now refs are coded like that:

<table><TR><TD valign="top">[1]</TD> <TD  valign="top">
H. Cartan,   S. Eilenberg,   "Homological algebra" ,  Princeton Univ. Press  (1956)
</TD></TR><TR><TD  valign="top">[2]</TD> <TD valign="top">
A. Grothendieck,    "Sur quelques points d'algèbre homologique"  ''Tohoku Math. J.'' ,  '''9'''  (1957)  pp. 119–221
</TD></TR></table>


and rendered like that:

 [1] H. Cartan, S. Eilenberg, "Homological algebra" , Princeton Univ. Press (1956) [2] A. Grothendieck, "Sur quelques points d'algèbre homologique" Tohoku Math. J. , 9 (1957) pp. 119–221

At least, we can create template(s) that save us the trouble of writing all these "<table><TR><TD..."

The best is probably the $\LaTeX$-like automatic numbering; but I doubt we can do it. If we cannot, then articles are not easily editable (not quite "wiki-wiki"), since a new source inserted in the middle of the list requires manual change of many numbers (both in the refs list and, worse, throughout the article). I did it several times working on "Measurable space"; not nice. No wonder that linking by numbers is not used on Wikipedia. On the other hand, our articles tend to be much less volatile than these of Wikipedia. Any comment?

--Boris Tsirelson 12:38, 19 January 2012 (CET)

The mediawiki equvalent of the table above is
{|
|-
|valign="top"|[1]||valign="top"| H. Cartan, S. Eilenberg, "Homological algebra" , Princeton Univ. Press (1956)
|-
|valign="top"|[2]||valign="top"| A. Grothendieck, "Sur quelques points d'algèbre homologique" ''Tohoku Math. J.'' , '''9''' (1957) pp. 119-221
|-
|}

which looks much easier and renders like this:
 [1] H. Cartan, S. Eilenberg, "Homological algebra" , Princeton Univ. Press (1956) [2] A. Grothendieck, "Sur quelques points d'algèbre homologique" Tohoku Math. J. , 9 (1957) pp. 119-221
In fact, the strings |valign="top" probably can be omitted:
 [1] H. Cartan, S. Eilenberg, "Homological algebra" , Princeton Univ. Press (1956) [2] A. Grothendieck, "Sur quelques points d'algèbre homologique" Tohoku Math. J. , 9 (1957) pp. 119-221
However, this concept conflicts with our earlier convention to use '|' as a separator between MR and Zbl numbers. Maybe that should be modified.
--Ulf Rehmann 20:55, 19 January 2012 (CET)
If you want automatic numbering, why not simply:
1. H. Cartan, S. Eilenberg, "Homolthogical algebra" , Princeton Univ. Press (1956)
2. A. Grothendieck, "Sur quelques points d'algèbre homologique" Tohoku Math. J. , 9 (1957) pp. 119-221
But why numbering at all? As has been said above: If there are references by number then automatic numbering (or manual renumbering) requires changing of the references, too, and if there are no references numbers are not used. In the text, references like (Cartan-Eilenberg, 1956) are more friendly to the reader. Personally, I prefer bibliographies in chronological order, possibly split in recommended reading, references, and historical sources.
--Peter Schmitt 23:35, 19 January 2012 (CET)
One may want numbered or labeled references, since this allows precise referencing, like "Thm. 1: <blah>. For proof cf. [42, p. 456f]" or the like. --Ulf Rehmann 00:35, 20 January 2012 (CET)
"Thm. 1: <blah>. For proof cf. [Cartan-Eilenberg, p. 456f]" is only slightly longer, carries the same information, and may often be understood without looking on the references. --Peter Schmitt 02:37, 20 January 2012 (CET)
One needs a reference list anyway, since most citations are from journals (and you don't want to quote the stuff say from ref 2 above inline). This could be achieved by using <ref "name"> ... </ref> tags and the {{reflist}} template, but then the automatically created list won't be sorted. So your proposal could be achieved if one uses just alphanumeric tags for each ref (as is often done in math literature) and sort the list by hand (unless a template is found to sort some list??) -- no renumbering needed in case of new refs.--Ulf Rehmann 13:08, 20 January 2012 (CET)

(unindent)
By a (not very extensive) search I found [3]and [4]. --Peter Schmitt 18:56, 20 January 2012 (CET)

Thanks. How would one make these programs work within a wikimedia template? --Ulf Rehmann 19:36, 20 January 2012 (CET)

#### Reference tables easily made sortable

This way they are easily extendable. (Please click on [sort] below. For details [see here])

[sort]
[Jac1] C.G.J. Jacobi, "Considerationes generales de transcendentibus abelianis" J. Reine Angew. Math., 9 (1832) pp. 349–403 Zbl 009.0357cj Zbl 14.0314.01
[Jac2] C.G.J. Jacobi, "De functionibus duarum variabilium quadrupliciter periodicis, quibus theoria transcendentium abelianarum innititur" J. Reine Angew. Math., 13 (1835) pp. 55–78 Zbl 013.0473cj Zbl 26.0506.01 Zbl 14.0314.01
[AM] A. Andreotti, A. Mayer, "On period relations for abelian integrals on algebraic curves" Ann. Scu. Norm. Sup. Pisa, 21 (1967) pp. 189–238 MR0220740 Zbl 0222.14024
[Griff2] P.A. Griffiths, "An introduction to the theory of special divisors on algebraic curves", Amer. Math. Soc. (1980) MR0572270 Zbl 0446.14010
[Mum] D. Mumford, "Curves and their Jacobians", Univ. Michigan Press (1978) MR0419430
[Griff1] P.A. Griffiths, J.E. Harris, "Principles of algebraic geometry", Wiley (Interscience) (1978) MR0507725 Zbl 0408.14001
[Se] J-P. Serre, "Groupes algébrique et corps des classes", Hermann (1959) MR0103191

Nice, but it does not solve the problem, which style to use: "[1]" or "[Jac1]". --Boris Tsirelson 07:50, 11 February 2012 (CET)
One has to use an alphabetic descriptor, if one does not want to rearrange the entries for every newly added item. --Ulf Rehmann 10:53, 11 February 2012 (CET)
OK, but probably we should have one way recommended to all. On Wikipedia, the style of refs is left to each author; should we? --Boris Tsirelson 12:19, 11 February 2012 (CET)
I think we should leave this to the authors. --Ulf Rehmann 22:01, 13 February 2012 (CET)

Here are two templates Template:Ref, Template:Cite which allow to define reference points {{Ref|Se}} which can be addressed by {{Cite|Re}}. An implementation with several sortable reference lists is given on the page Jacobi variety. --Ulf Rehmann 22:01, 13 February 2012 (CET)

A progress!
It would be still nicer, to have more (optional) parameters that allow (a) to make "See [M, Sect. 5]" rather than "See [M], Sect. 5" (on WP it is the "loc=" parameter), and (b) to put an anchor whose name is not the same as its visible text (like in [[aaa|bbb]]). --Boris Tsirelson 22:38, 13 February 2012 (CET)
Yes, it works nicely. --Boris Tsirelson 20:58, 14 February 2012 (CET)
The "Cite" template could generate "[H, p. 222]" rather than "[H,p. 222]" if called as "{{Cite|H|p. 222}}. --Boris Tsirelson 16:15, 16 March 2012 (CET)
Should be fixed, please check.--Ulf Rehmann 16:44, 16 March 2012 (CET)
Yes it works, thank you. --Boris Tsirelson 19:39, 16 March 2012 (CET)

Here is some comfortable toy to get the old reference tables transcribed into wikipedia tables, thereby ordering them and installing the appropriate links and anchors using Template:Ref and Template:Cite throughout the page.

The best way to use that script is via an external editor, which allows to employ the script as a "filter" for the editor's buffer content.

It would be nice if the script were exercised by others as well to get some feedback about it's operation. Detection of authors' names depends on some heuristic (see the script's annotations), which so far worked pretty well.

See my recent output on Special:RecentChanges with summary "refs" etc to get an idea how the ref tables will look like. --Ulf Rehmann 14:32, 18 February 2012 (CET)

Now unicode letters are appropriately handled as well, see the reference list on Abstract algebraic logic#References. --Ulf Rehmann 14:23, 20 February 2012 (CET)

#### Globally defined reference lists

It occurred to me that many pages related to the same area might share some common (larger) reference list which may also be annotated guide for further reading. It would spare us repeating the same Refs/Cites ad nauseam. One might argue that this would lead to extra browsing, but in fact the current practice in any case tends to mimic that of mathematical papers with all references at the end, rather than "humanitarian" papers with references in the footnotes. I see no large difference between pointing to the bottom of the same (lengthy) page and pointing to a well designed list on a separate page.

Can we implement such style by templates? should we encourage it? Sergei Yakovenko 10:46, 23 April 2012 (CEST)

I think it is worth keeping a small number of references (those directly relevant) on each article, especially the papers relating to results mentioned. I think it would very useful to also having a larger subject reference list with annotations, however. Many references in the articles are not directly relevant, but supply 'background reading', which would be better served by more global lists. TBloom 12:53, 23 April 2012 (CEST)
In principle, we'd better wait for Ulf's opinion (he is in a trip now). But here is a technical remark. A text created once on an "auxiliary" page can appear on many "usual" pages (no extra browsing needed); this is called transclusion. --Boris Tsirelson 16:58, 23 April 2012 (CEST)

Is it written somewhere, which usernames are acceptable here?

No, its not (yet), finger glitch, corrected.--Ulf Rehmann 11:51, 26 January 2012 (CET)

As far as I see, "TrialAndError" is rejected, while "AgelessMale", "Abrar.p496" etc. are accepted. On Wikipedia, almost every name is acceptable. On Citizendium, only real names are (thus, "Rafael.greenblatt" would be " "Rafael Greenblatt" on CZ). We should publish a rule before punishing for its violation. --Boris Tsirelson 10:57, 26 January 2012 (CET)

Sure.--Ulf Rehmann 11:51, 26 January 2012 (CET)

## Time zones

I just noticed that the history pages display UTC/GMT, while the time stamps produced by signatures are CET. In order to avoid confusion, they should give UTC, too. --Peter Schmitt 01:18, 28 January 2012 (CET)

Also, "Recent changes" is cheating about the order of changes. For example: my edit of 30 March 2012, 11:38 (according to "Recent changes") is in fact made after Ulf Rehmann's edit of 30 March 2012, 18:45 (according to "Recent changes"). --Boris Tsirelson 13:44, 30 March 2012 (CEST)
Apparently something is wrong with the clock on this server. As you still can see, on "recent changes" two of my changes on "Schubert cylces" got a timestamp which is still in the future, though I edited these this morning.
Also, some links point to historical versions of pages, not to the most recent. I did already this morning email the sys admins about that, but as they are in the US they aren't yet awake, as it seems. Well, we'll see... --Ulf Rehmann 15:11, 30 March 2012 (CEST)

## Something is wrong

I try to edit Convergence of measures as usual (just adding MSC and category); but as a result the page becomes empty. (Twice.) --Boris Tsirelson 16:26, 7 February 2012 (CET)

I cannot confirm this: I edited the page twice, the first time under my admin account, and by adding MSC, the second time under my account TestUser with just ordinary user permission, and by adding the tex category "TeX wanted". Everything worked fine, as it seems. --Ulf Rehmann 19:00, 7 February 2012 (CET)
Rather mysterious. I got it again, this time on Cramér theorem. I added
{{MSC|60F10}}
[[Category:Limit theorems]]
did "Show preview" (nothing strange here) and then "Save", and got the empty page! --Boris Tsirelson 20:08, 7 February 2012 (CET)
More observations. This time I was not asked to answer an arithmetic question; but the previous time I was, with the explanation "since you are adding an external link". More: I use exactly the same browser and system as all the time before. --Boris Tsirelson 20:12, 7 February 2012 (CET)
And now, on User:Boris Tsirelson/sandbox1 I got again "Your edit includes new external links" but nevertheless the page did not get blank. --Boris Tsirelson 20:17, 7 February 2012 (CET)
And now it went smoothly on Cramér theorem with the same procedure as I did before.
But I also note that the behavior of the edit window changes (another reaction to right-click). --Boris Tsirelson 20:27, 7 February 2012 (CET)
I had no problem with adding a template (with question). A subsequent edit worked without question. (Yesterday I had to answer a question everytime I saved -- I don't remember if external links were mentioned (I probably did not read the text), but I certainly did not add external links (at most wiki links). My impression was: Having to confirm every edit is too much ... --Peter Schmitt 21:36, 7 February 2012 (CET)
MSC template generates external links. --Boris Tsirelson 22:21, 7 February 2012 (CET)
Yes, I have noticed this. But yesterday I did not use a template.
As for the external links: I'm not sure that this is useful, in particular to MathSciNet: This will either not work (without access rights), or produce a much too long list of results. And why link to the MSCwiki? A copy of the classification can easily be hosted here at EoM (and be integrated with the content). --Peter Schmitt 22:55, 7 February 2012 (CET)

I have notified the admins at Springer about this. It may be that they are still trying experimental software. --Ulf Rehmann 22:59, 7 February 2012 (CET)

It seems, now I understand it better. When I hit "Edit" I get the edit window with the wikitext; first it is in one font, but after about a second it switches to another font. Later, being asked "To help protect against automated spam, please solve the simple sum below and enter the answer in the box", I see the wikitext again in the former (larger) font, and this time the switch to the "right" (smaller) front takes several seconds. IF I solve the sum more quickly :-) not waiting for the font switch THEN I get the infamous empty page effect. --Boris Tsirelson 13:00, 13 February 2012 (CET)
I cannot confirm this either. I do see the switch of fonts, but I never see an empty page nor do I get an offer to answer an arithmetic question after login. To me, your problem description seems to indicate a timing problem concerning your network connection with the server. Anyway, such things should not happen, and I will notify the admins again. --Ulf Rehmann 17:08, 13 February 2012 (CET)
About arithmetic question after login: this time I get it only when I insert an external link (which happens via MSC template). About my network connection with the server: yes, sometimes it is bad. --Boris Tsirelson 18:24, 13 February 2012 (CET)

I decided to put this caveat here (where "something is wrong"), just in case it helps to save an innocent soul. Despite the recommendation to typeset in the sandboxes, all too often I am tempted to edit in the window provided by my browser (well, I work under Windows using Mozilla, that's already bad in itself). The accidentally typet Ctrl-Z to undo the "latest edit" will undo everything you ever typed in in the last half an hour or so. There is a Savior, Ctrl-Shift-Z for the "Redo". For the unexperienced wikings like me it may be an ultimate solution... -- Sergei Yakovenko 16:55, 30 April 2012 (CEST)

## Another useful template: "anchor"

The template User:Boris Tsirelson/Anchor is used in "Measurable space". An example: the code "[[Measurable space#subspace|subspace]]" generated a link to an "anchor" (entry point): subspace (try it); the anchor is set by the code "{{User:Boris Tsirelson/Anchor|subspace}}" in that article.

If you like this template, move it to the template namespace for wide use. --Boris Tsirelson 17:12, 9 February 2012 (CET)

Sure, that could be helpful. It should be equipped with some documentation.
How about this one this one? But I don't know whether a multiple named anchor is needed. An #if:{{{1|}}} clause may be useful, homever.--Ulf Rehmann 18:56, 9 February 2012 (CET)
For me, any working version will be good enough. --Boris Tsirelson 20:20, 9 February 2012 (CET)
Done: Template:Anchor --Ulf Rehmann 21:28, 14 February 2012 (CET)
Yes, it works nicely. --Boris Tsirelson 20:59, 14 February 2012 (CET)

## Originator

I noticed the following: Imported pages (first revision) show an "originator" on "How to cite this entry". This information disappears after the first edit -- is it completely lost or is it only hidden somewhere? (The first revision does no longer show it.) --Peter Schmitt 01:25, 16 February 2012 (CET)

I don't know, will ask. --Ulf Rehmann 13:16, 16 February 2012 (CET)

## Typography

Perhaps it is only my browser, but for my feeling the font used for inline TeX is slightly too large. --Peter Schmitt 00:42, 7 March 2012 (CET)

Here is how I see it.

--Boris Tsirelson 08:01, 7 March 2012 (CET)
The screenshot shows it, too. The letters have not the same pointsize. For me, the following example shows the difference quite clearly: AA$A$A$\cal A$A$\mathbb A$A. --Peter Schmitt 12:11, 7 March 2012 (CET)

## EoM and WP

Hi. Could you please compare and contrast the EoM and the Wikipedia. For example, what would be the pros and cons for authors in deciding where to contribute. Thank you. Fgnievinski 21:52, 13 April 2012 (CEST)

Well, on one hand, on WP your article will be much more visited. On the other hand, most probably it will become somewhat incorrect, mostly because people will make local changes not worrying about inconsistencies thus created. Sometimes they will do quite stupid changes; sometimes they will insist in spite of your protest. And another problem: on WP, an editor (be it really an expert, or not at all) is not permitted to write something just because it is true; rather, you must give a reference (preferably, to a textbook); otherwise, your text has a chance to persist, and a chance to be removed. Also, someone can find an erroneous statement in a bad book, include it to your article, and insist. And, after all, WP is not intended to professional mathematicians; there, your article can be attacked as "giving not enough context", which means: too hard and special for a general reader. (Again, there is a chance to persist, and a chance to be deleted.) --Boris Tsirelson 08:52, 14 April 2012 (CEST)
Despite the obvious drawbacks, very often the WP math pages are quite reasonable as a zero approximation (with some technical, style or reference corrections). We (the editors) should probably encourage people to borrow heavily from (good) WP articles, adding the missing things. Sergei Yakovenko 11:53, 17 April 2012 (CEST)
However, our material should be strictly based on references to original refereed literature. We should not quote WP for facts. Remember that EoM has been and should be considered as a refereed source, even by WP, see this. Otherwise we would not need an editorial board. We should be aware of our publishing policy. --Ulf Rehmann 12:52, 17 April 2012 (CEST)
In principle, WP is even more strictly based on references. For example, here I sometimes spot errors in books (like this). On WP I could not do it, unless the criticism is already published in a "reliable source", since on WP I am just a user (called "editor"), not an expert; I may add to the sources no more that "results of simple calculations". Here, hopefully, I am an expert, entitled to go beyond the simple calculations. But it is also true that in practice WP contains many non-sourced statements; some of these are wrong; but worse, some of the sourced statements are also wrong (maybe even in the same proportion), for the reasons I already mentioned above in this section. Let me also mention that the WP rules discourage references to encyclopedias ("tertial sources") as well as research papers ("primary sources") and encourage references to textbooks and monographs ("secondary sources").--Boris Tsirelson 13:29, 17 April 2012 (CEST)
So we are different: We do rely on and will quote what they call "primary".--Ulf Rehmann 14:31, 17 April 2012 (CEST)
Ulf, I am not sure the original research papers are the best source. I would rather believe that we need to point our readers to the best "verified"/"certified" complete exposition, which is almost always a monograph or a textbook. Sergei Yakovenko 08:52, 19 April 2012 (CEST)
Agreed, insomuch this is available. But recent developments often don't have this. --Ulf Rehmann 09:24, 19 April 2012 (CEST)
Thus, a statement from WP could be reproduced here (I think so) provided that it is reproduced by an expert responsible for its correctness (and refs). But another obstacle exists. If Google robots will observe here a content from WP, our Google rank will decrease dramatically. (This obstacle was taken very seriously on Citizendium.) Probably we could convince a human that our article is more reliable than the corresponding WP article even if they are textually identical. But, how to convince the Google robots? --Boris Tsirelson 13:29, 17 April 2012 (CEST)
This is serious and a considerable argument against using content from WP. So, what we need is an authentic formulation of our texts.--Ulf Rehmann 14:31, 17 April 2012 (CEST)

## MathJax suggestion

I find the MathJax default behaviour a little "untypographical" in that on initial page-load one sees a load of raw LaTeX. After some experimentation on my personal webpages I have found that that the following MathJax config

<script type="text/x-mathjax-config">
MathJax.Hub.Config(
{
tex2jax:
{
inlineMath: [ ['$','$'], ["\$$","\$$"] ],
processEscapes: true,
preview: [["span", { style: { color: "Gold" } }, ["\u22EF"]]]
}
}
);
</script>


gives a more attractive user experience. The "preview" configuration specifies that a gold-coloured centred ellipsis be used instead of the raw LaTeX. See for example this page. Perhaps something similar (obviously using a different colour, perhaps red) could be used for EoM?

--Jjg 15:24, 16 April 2012 (CEST)

Thanks for this interesting suggestion. I don't have access to the EoM server installation, so I can't give it a try directly. However, I have servers running privately with a mathjax installation under mediawiki. I tried your code and it did not work the same way as on your page (in fact the server did not render tex images at all, it just did show the source code). The mathjax script on the EoM server (and on the one I run here) is more involved than the one on your site (look at the source code of any page, search for 'tex2jax'). --Ulf Rehmann 14:18, 14 April 2012 (CEST)

## Typographic consistency

Hi

I wonder if there are any views on typographic consistency in EoM, and if there are any policy plans for it. I ask because there seems to be some typographical "drift" in the articles so far TeXed. As an example, the unTeXed page Local-global principles for large rings of algebraic integers uses the traditional bold-face $\mathbf{Z}$ for the integers, $\mathbf{C}$ for the complex field and so on, whereas the TeXed General linear group uses blackboard bold $\Z$ for the integers. Regardless of my personal opinions (that blackboard bold is an abomination that should only be seen on a blackboard!) I think that it is more important that typographic consistency be maintained.

Could make some suggestions in this direction?

• That there is a policy page on typography for TeXers
• That the default policy be to maintain the typographic conventions of the original page
• That deviations from default policy are discussed and then clearly enumerated on the policy page
• That TeXers are encouraged to used "standard" mathjax macros (like \C for the complex numbers)
• That the standard mathjax macros reflect the policy (so \C should produce $\mathbf{C}$ until such a time as blackboard bold becomes policy)
• That redefining standard mathjax macros is treated as a fault in the page and corrected (perhaps by a script)

Thoughts?

Jim (--Jjg 15:21, 16 April 2012 (CEST))

Actually I am hesitating to be too rigid here at this time. This is a project joining very different mathematical communities which might have different typographical cultures. Also, the typography of the old files is not consistent either (may be for the same reason: many authors from many fields) - I have seen pages where \mathbf{C} did *not* mean "complex numbers". It is essential, however, that the typography is consistent in pages whose content is close, but I hope this will be regulated by the authors from the various areas. The macros predefined by mathjax are these. --Ulf Rehmann 18:38, 16 April 2012 (CEST)

## Editing Policy

I have "commissioned" 10 or so articles now to get things going in my area (where there is practically nothing on EoM for historical reasons). I guess these will start appearing in a few months. But many of the young wikipedia-generation mathoverflow-types I have approached have a similar complaint -- that the barrier-to-entry to making edits is too high, and the whole project appears too hierarchical.

We surely really want all young mathematicians on this site, editing it. Initially we may have to select/invite these people, but once it gets going and is more comprehensive and useful, we want anyone to edit, not just the people that we think of. So we want a very open, democratic structure.

The best way to get them involved is if there is no obstruction to them making tiny edits to begin with (typos, references, etc); later they may contribute a whole article.
Insisting on user registration and email verification is a big obstruction, and against the open source spirit. Wikipedia does not insist on this.

Of course we all fear (highly motivated and energetic) cranks, or people with bias or malice. But the lesson of democratic projects is to always trust the community. As Arend Bayer pointed out to me, contributors usually police their own contributions assiduously, quickly removing any bad alterations.

I vote for opening this site up completely. If problems arise, I imagine they can be corrected quickly and painlessly. If not, we could introduce some milder controls. For instance, every page could be attached to an editor, and alterations could automatically go via the editor before appearing online.

Eventually this site will (hopefully) be so heavily used as to be self-policing. In the beginning I can understand the desire for some form of control, but it risks setting the wrong tone and alienating precisely the most important people we need working on this site: young, open-source types who have a different mindset from us old fogeys. I can't imagine many of the most prodigious young contributors to mathoverflow putting up with our rules, yet many of them have written thorough and lengthy answers to mathoverflow questions that would make excellent EoM articles in themselves.

Richard (--Rpwt 18 April 2012)

"...and alterations could automatically go via the editor before appearing online" — such way is tried on Knowino; there, alterations appear online immediately, but the reader is warned that they are unchecked, and has the option to switch to the last checked version. But it is technically equally possible to do the opposite. --Boris Tsirelson 07:45, 19 April 2012 (CEST)

## Figures

I see that there are a few figures in the EoM, mostly they seem to be low-quality scans from the paper edition, so I guess that at some point they will need to be redrawn.

• If there a list of all figures somewhere? I couldn't find it.
• It might be useful to have some graphical guidelines for common figure types (classical line geometry, graphs of functions and so on), such as font type (serif, typewriter, sans-serif?), font size, line thickness (especially for axes), ...
• It would be nice if high-quality vector formats (pdf, svg) were available as well as PNGs
• On WP they have the "commons" for storing and developing such figures. Perhaps EoM could do the same, or could create an niche on the WP commons to avoid the admin overhead.

--Jjg 17:27, 19 April 2012 (CEST)

Hopefully, some day we'll have the "Asymptote" system (see User talk:Nbrothers#Requests for Software Extensions, example on test wiki, the site of that system) very useful when making pictures. --Boris Tsirelson 07:53, 20 April 2012 (CEST)
I'm not sure I'd see that as a great advantage -- of course everyone should be free to use the tool that they like to make a picture, but if those pictures are not in a standard format (pdf, svg, png) then they would not be so useful. Moreover, for any reasonably complex plot, running the asymptote interpreter on every page load seems a waste of resources. Are we to also add server-side generation for plots using MetaPost, xy-pic, POV-Ray,... ? --Jjg 14:15, 20 April 2012 (CEST)
First, the wiki software is rather cute, it cashes pages so that they are not generated again if nothing could change. Second, separate pictures are also quite a mess; you must upload each, specify its license... and then they are not "wiki-wiki": if someone (be it you or not) wants to make a small change to it, then... what to do?? An "asymptotic" picture can be changed readily by anyone acquainted with "asymptote", just as a TeX formula can be changed readily by anyone acquainted with TeX. Third, an "asymptotic" picture on a wiki is a collective work (just as the wikitext) and therefore is not licensed separately. Fourth, of course, a picture may be too complex for "asymptote" and so made in pdf, svg, png — with the corresponding advantages and disadvantages. --Boris Tsirelson 16:06, 20 April 2012 (CEST)
I'm paranoid about backups, and as a result I have never lost work to a hardware or OS failure. But I have lost work when I was depending on Open Source software whose maintainer decided to give up on the project and move on to other things (not that it matters, the projects were RLab, a matlab-like package, and emTeX). Consequently I am nervous about using small projects like asymptote (2 developers) for infrastructure. Don't get me wrong, I think asymptote is a nice bit of software (not my favourite for drawing, but let 1000 flowers bloom ...), but are you sure it will be there in 30 years? PDF will be. SVG will be. Sure, if the developers abandon asymptote and move on, you can take the source and maintain it yourself, but will you? Or will it be easier to take the source for the 150 EoM plots made with asymptote and convert those to another format, cursing your younger self while you do it.
--Jjg 11:50, 21 April 2012 (CEST)
I see; a problem indeed. --Boris Tsirelson 13:36, 21 April 2012 (CEST)
With reference to a list of figures, I find that searching for "Fig" seems to do the trick and returns 202 hits --Jjg 22:23, 22 April 2012 (CEST)
I have made a download of all pages to my local machine and did a count of the string "Figures:" which gave me a number of 731. It's good suggestion to set up a list of all such figures. Will try to do this when back home next week (travelling right now). --Ulf Rehmann 23:11, 22 April 2012 (CEST)
Ok, a first try: User:Ulf Rehmann/figures. One may want a better format of this list, let me know what you think. --Ulf Rehmann 00:10, 23 April 2012 (CEST)
Looks fine to me, there are many more than I expected (and some are going to be challenging to redraw). Perhaps this list could be used as a checkist (mark the link as "done" when the image is redrawn), or perhaps a {{FIG:done}} tag would be useful? --Jjg 00:20, 23 April 2012 (CEST)

#### Example page and suggested guidelines

I have created some example images using what I would think would be reasonable general guidelines for images

• For the raster images displayed on the pages, widths of 400px ("half-width" image) or 800px ("full-width" image)
• For geometric line drawings, examples: ellipse, tangent line
• At most 2 different line thicknesses, 0.35mm and 0.2mm
• No line colour except black if possible, greys by second preference
• Font is computer modern if possible, times if not
• Font size is 10pt in the image displayed on the page (can be bigger in the linked pdf)
• For 3d function plots [TO DO]
• For 3d topological images using POV-Ray, example: ellipsoid
• Monochrome unless justified by complexity
• For the object body use a small value of transmittance (transmit 0.3), and a small amount of Phong shading (finish {phong 0.1})
• I rather like the texture used in the "floor" of this image, it gives perspective while not overwhelming the central object of interest. The technique is described here.

The embedding of the images in these example uses

<span id="Fig1">
[[File:Ellipse-1.png| right| frame| Figure 1. The ellipse ([[Media:Ellipse-1.pdf|pdf]]) ]]
</span>


where the span allows one to link to the figure (bit of a kludge, is there a better way?), the png and pdf file are named according to the page-name and figure number. The right could be replaced by center for large and complicated plots, but frame should always be present (else one does not get a caption). The link to the pdf uses Media rather than File, so that the link is direct to the pdf rather than the pdf resource page.

Comments? --Jjg 18:26, 23 April 2012 (CEST)

Nice. About "named according to the page-name and figure number": it works as long as a figure is used on one page only; sometimes this condition may fail. --Boris Tsirelson 07:45, 24 April 2012 (CEST)
Good point, but I wasn't thinking of this as a hard-and-fast rule for all images, but just a convention to avoid name-clashes between new versions of existing images on different pages -- the MW namespace is global, there can only be one "graph.png" --Jjg 13:08, 24 April 2012 (CEST)
I propose to extend the "Done" entries in User:Ulf Rehmann/figures by saying something like "xxx replaced by yyy", in order to keep track. Have done so with the existing "Done's". We should aim towards an extended page for each of these graphics, which should present each "incarnation' of the objects. Forthcoming graphics techniques may allow other types of graphics (e.g. animated versions etc.), which should all be included in such a page. I will try to provide a convenient setup for this. --Ulf Rehmann 22:55, 24 April 2012 (CEST)

## Manifold Atlas

A similar project (at least in the subfield of manifolds) was pointed out to me by Dima Panov: the Manifold Atlas Project. I imagine there are other such projects in other fields. We could consider using their material (properly cited), or asking its authors to contribute to EoM, or even merging with them ?

Richard (--Rpwt 19 April 2012)

A good question... --Boris Tsirelson 07:56, 20 April 2012 (CEST)
Should be easy: I was in contact with the director of the Manifold Atlas Project before we started this EoM, and he was very open for possible cooperation. So if something concrete should be arranged and I should help, please let me know. --Ulf Rehmann 01:11, 26 April 2012 (CEST)

Hadamard theorem was quite a mess, so I fixed it up with TeX, referencing, and put the sections in more logical places. It seems like it would be better split up into different articles for each of the different theorems, however. If people agree, can this be done/can I do it myself? TBloom 10:05, 21 April 2012 (CEST)

As for me, yes, it would be better, and yes, you can do it yourself. :-) --Boris Tsirelson 11:09, 21 April 2012 (CEST)
On the other hand, having all results on the same page may spare the reader from unnecessary clicking. Two-three screenfuls of text is an affordable size, not necessitating further splitting. Sergei Yakovenko 07:58, 22 April 2012 (CEST)

I'm also not sure how to label equations - Wikipedia suggests NumBlk but the package seems to be missing. TBloom 10:09, 21 April 2012 (CEST)

But we have MathJax! See here for an example. --Boris Tsirelson 11:09, 21 April 2012 (CEST)

## Divisor

Carlo Madonna pointed out that a search for "divisor" brings up only the page Divisor, and not the page Divisor(2).

Can this be fixed somehow ? An obvious fix would be to change the titles (to, say, "Divisor (of an integer)" and "Divisor (ring theory and algebraic geometry)") but I don't know how to do that. Richard (--Rpwt 22 April 2012)

Ho do you search? If searched using the page Special:Search, the page "Divisor(2)" does appear as well in the popup window (second entry) as within the search results as 9th page (out of 10), searched within namespace "Main" (scroll down search results!). If I put in the keyword "divisor" on the main page, "Divisor(2)" as well is shown in the popup window as second entry (tested with Chromium and Firefox).
However, such things should be handled by a ["disambiguation page"]. --Ulf Rehmann 08:40, 23 April 2012 (CEST)
Pages can be "moved", see the upper-right corner of the menu (triangle to the right of "View history"), but this creates a redirect. I do not know if it is possible (and then how) to delete existing pages... Sergei Yakovenko 07:03, 23 April 2012 (CEST)
We cannot delete pages, but Ulf (being admin) can; ask him here. --Boris Tsirelson 07:38, 23 April 2012 (CEST)
Then it could be a good idea to warn the new editors (especially novices) to be very careful when creating new pages, exploring instead the possibility of editing the old ones... Sergei Yakovenko 08:22, 23 April 2012 (CEST)
These seem to be the pages which have "divisor" within its title:
Additive divisor problem | Divisor | Divisor(2) | Divisor class group | Divisorial ideal | Divisor problems | Elementary divisors | Greatest common divisor | Number of divisors | Poincare divisor | Poincaré divisor | Unit divisor | Zero divisor
In case somebody sets up a disambiguation page for "divisor", please consider this list. --Ulf Rehmann 08:58, 23 April 2012 (CEST)
I just used the searcher in the main page. If you digit "diviso.." then you get few suggestions including the Divisor(2), but, when you finish to digit "divisor" then only the "divisor" link is suggested and you are redirected to the page on "divisor of an integer".
Once you are in this page nowhere appears a link for example to Divisor(2), or Divisor in a variety ... . I am very new to all this, even I did not succeed to find the link to the "special search" web-page.
I will check more. --Carlo Madonna 10:22, 23 April 2012 (CEST)

I would happily set up a disambiguation page for divisor, but I don't know how to do it. Can the technical people do it for us ? It could serve as a template for other disambiguations. We certainly also need to change that dreadful title "Divisor(2)", but again that would require someone with more privileges.

Richard (--Rpwt 23 April 2012)

I have no privileges, but I've made a disambiguation page: Measure algebra. The only action that needs admin rights is, deletion of unneeded pages (and not a problem: just ask Ulf, as shown above). --Boris Tsirelson 19:36, 23 April 2012 (CEST)
Ok, I did it by moving the old divisor page and replacing it by a disambiguation page here. But we still need to rename the page divisor(2) somehow. --Rpwt 23 April 2012

## Generated reference labels

The emacs macro which converts old-style references to new works quite well, but could I flag a couple of issues

• Where the transliteration of the original Russian author and the transliteration used in the translation of a Russian author's name differ, the latter is give in square brackets: eg (from Ill-posed problems) A.N. Tikhonov, V.I. [V.I. Arsenin] Arsenine, "Solution of ill-posed problems", Winston (1977). Then the emacs macro uses the name in the brackets to generate the label, in the above, [TiArAr], which seems a bit silly.
• There are several pages with references like [label] "Title" Editor (ed), Editor (ed) (i.e., there is no author preceeding the title) which fail to be processed by the script, examples are in Shannon sampling theorem and Travelling salesman problem

--Jjg 14:17, 23 April 2012 (CEST)

I have uploaded a version which addresses the first issue. Please observe the new warning on the page just mentioned. The second issue will be fixed later.
I have as well a similar perl script which automatically finds MR and ZBL numbers, which to some extent works quite well. It does make mistakes, in particular, if the bib references given are incorrect, e.g., give a wrong year number or misspell titles/authors.
Is there some interest in exercising this as well? Of course this requires you to have online MSN/ZBL access.
--Ulf Rehmann 19:13, 24 April 2012 (CEST)
Would be a pure wishful thinking to hope that one day a template similar to Amsrefs would be available for EoM? I understand that this is a lot of job to implement this fully, so may be just some "limited", baby version? The huge advantage of such template would be the possibility to copy directly from MathSciNet without manual editing. The [Wiki citation templates] do a similar job, but using the Wiki engine would in any case require manual edit. - Sergei Yakovenko 13:56, 28 April 2012 (CEST)
I don't have MR access sadly; but I did notice a minor error in the output on the 21:32, 24 April 2012‎ revision of Ellipse, the problem is the template content is M|R12345 when it should be MR|12345, maybe a minor bug in the script? --Jjg 00:06, 25 April 2012 (CEST)
That's a typo (corrected as I hope). The script does not find everything, some entries have to be made by hand. --Ulf Rehmann 15:20, 25 April 2012 (CEST)

## Arc simple arc Jordan arc

The name of Arc simple arc Jordan arc should be changed, but I'm not sure what to. Confusingly, there are already entries for Arc, Jordan arc and simple arc. There are four articles where there should be at most three, but I don't know enough about the area to say where to draw the line and how these should be merged. TBloom 10:47, 24 April 2012 (CEST)

On actually reading the definitions, it seems like Arc simple arc Jordan arc, Jordan arc and simple arc should all be merged under either the name of Jordan arc or simple arc. TBloom 10:48, 24 April 2012 (CEST)
Yes. And I wonder why "arc" is just a notion of discrete geometry (as I see in our "Arc" article). Is it the standard terminology? Or should it be something like "arc (projective geometry over a finite field)"? As for me, "arc" should be rather a disambiguation page with links to these (ultimately, two? the other being "arc (topology)"?) articles. --Boris Tsirelson 12:43, 24 April 2012 (CEST)
That sounds reasonable. Wikipedia, for example, has Arc (geometry) and Arc (projective geometry). Unless somebody weighs in otherwise, I'll make two arc articles and merge them accordingly, and make the existing Arc page a disambiguation page. Shall I go with Arc (topology) and Arc (projective geometry)? TBloom 12:55, 24 April 2012 (CEST)
I am just a probabilist, – not an expert in any of these two topics. Maybe experts can fine-tune the titles better? On the other hand, they will be able to do so afterwards (and move the articles). --Boris Tsirelson 14:47, 24 April 2012 (CEST)

Arc simple arc Jordan arc, Jordan arc and simple arc have now been merged into Arc (topology) and Arc has been moved to Arc (projective geometry). Arc is now a disambiguation page. If somebody knowledgeable could have a look at my disambiguation descriptions and my merging, that'd be great. Could somebody who can please delete Arc simple arc Jordan arc, Jordan arc and simple arc? TBloom 10:20, 25 April 2012 (CEST)

Only Ulf can delete; ask him there. --Boris Tsirelson 13:02, 25 April 2012 (CEST)

## Deficient pages

Insomuch pages are found to be erroneous or otherwise seriously defective, they should be put into the Category:Deficient. This can be achieved by putting the string {{DEF}} on top of everything else in that page, which calls the Template:DEF.

A reason for the page deficiency should be given in its talk page (the DEF template creates an appropriate link).

Also, deficient pages which are not already converted into TeX should be placed into the "TeX wanted" category in order to speed up revision. This can be achieved by putting the string {{TEX|want}} under the {{DEF}} entry. --Ulf Rehmann 15:14, 25 April 2012 (CEST)

Nice. And what about missing (wanted) pages?
About DEF and TEX|want: it seems, some DEF pages are so much DEF that it is better to rewrite them from scratch; and if so, then, why TEX|want? --Boris Tsirelson 18:56, 25 April 2012 (CEST)
missing ==> DEF (see Polytope as an example). If a complete rewrite is desirable please mention this and don't use TEX|want. --Ulf Rehmann 19:44, 25 April 2012 (CEST)

### "Heritage" pages

A closely related problem: some historic pages clearly require very serious facelift, sometimes splitting into several pages after proper disambiguation etc. My favorite example is Singular point. Still the revision might well spread over long time, besides, the pages in their current form might be of historical interest. I would suggest creation of a limited number of "historical", or "heritage" pages and moving selected pages into this category, while commissioning the work on the modern revision... - Sergei Yakovenko 13:49, 28 April 2012 (CEST)

The idea is nice, but are you aware that all preceding versions of pages are maintained anyway and accessible under "View history" ? --Ulf Rehmann 23:41, 28 April 2012 (CEST)
Yes, I understand that any page can, if necessarily, be rolled back to any point in the past. My suggestion was two-fold: (1) to stress that the current page is "out of fashion" and undergoes a serious revisions, and (2) stress that the "vintage" page has its own value, albeit historic rather than informational. Note that some of the pages of the original EoM were written by great Russian mathematicians and deserve to be archived in a more explicit way than just one of the past revisions. One (quick fix) possibility could be to add the postfics "(historical)" (or something like that) to the current page name (by "moving") and post a link to it at the top of the newly edited page. - Sergei Yakovenko 08:05, 29 April 2012 (CEST)
Thus, "heritage" pages should be signed, in contrast to usual wiki pages. Also, they can reside in a separate, not user-writable namespace. --Boris Tsirelson 10:35, 29 April 2012 (CEST)
Ok, I will propose this to Springer. I think I can't arrange this myself. As was already realized here, the originator entry disappears even from the history page after its first revision. One way out of this could be: I could block that page, put it into a history category, and then set up a copy which is further edited. However, if a page were edited before having been blocked, the "originator" entry were gone. I have asked Springer to change this behaviour, and they promised to do so, but only after they will have set up the asymptote graphics software, since that is what they are working on right now. (We only have small time slots at their system admins workload). --Ulf Rehmann 11:52, 29 April 2012 (CEST)
But hopefully this encyclopedia (as it was before wiki) is still available elsewhere, and we can always restore the author name manually. --Boris Tsirelson 11:56, 29 April 2012 (CEST)

## How to cite ...

It seems that the "how to cite" URL at the bottom of each page is incorrect. For example, on the page Triangular summation method, the URL is given as "http://www.encyclopediaofmath.org/index.php?title=Triangular_summation_method&oldid=25646" which is not only uglier than what is in the address bar of my browser: "http://www.encyclopediaofmath.org/index.php/Triangular_summation_method", it also does not link to "this page", instead to this, which seems to be a "most recent revision of this page" page.

It might also be useful to have the URL text contain a link to the same place as this would mean that one could use the "copy this link" facility in browsers rather than cutting-and-pasting the URL text.

--Jjg 12:44, 28 April 2012 (CEST)

I think the bottom entry is correct. It refers to a versioned page (and not to the "most recent version"). After all,by quoting one refers to a specific text which is fixed (as kept in the history list of a given page) and not to its possible future versions, which may not even contain the material to which the quoting wants to point to.
The proposal to have the URL as a link is a good idea, will try to achieve this somehow. --Ulf Rehmann 12:21, 29 April 2012 (CEST)

## "It's All Text" extension

I've just installed the "It's All Text" extension to Firefox. It works on Wikipedia but does not work on our wiki. The "Edit" button does not appear (while I am typing this in), and right-click gives pull-down menu without "It's All Text!" However, just now, in another tab, it does - on a Wikipedia page. Any advice? --Boris Tsirelson 19:54, 1 May 2012 (CEST)

It also works on two other wikis: Citizendium and Knowino. Just not here! I tried changing my preferences; it did not help. --Boris Tsirelson 20:13, 1 May 2012 (CEST)

Same story on my side. :-( -- Sergei Yakovenko 06:51, 2 May 2012 (CEST)

Well, I got an advice from Ulf (by email): just to wait; something is wrong now with this wiki; and in fact, "log in" does not work (thus, do not log out...); system people are informed. --Boris Tsirelson 21:16, 1 May 2012 (CEST)

Still waiting. --Boris Tsirelson 07:38, 2 May 2012 (CEST)
I can login now --Jjg 11:11, 2 May 2012 (CEST)
Ok, server seems back again: Both "It's all text" (Firefox) and "Edit with Emacs" (chrome/chromium) seem to work. --Ulf Rehmann 11:18, 2 May 2012 (CEST)
Really? For me "It's all text" does not work (still? or again? on WP it does). --Boris Tsirelson 13:52, 2 May 2012 (CEST)
Still, not. (Trying from another platform. Login successful.) --Boris Tsirelson 16:50, 2 May 2012 (CEST)
This is edited via "It's all text" and firefox 12.0 --Ulf Rehmann 18:11, 2 May 2012 (CEST)
My Firefox is 12.0; and on Wikipedia it works. --Boris Tsirelson 18:33, 2 May 2012 (CEST)
Having no better conjectures I wonder, could it be related to your admin status? Did some plain user succeed with "It's all text" here? Did you try it on a non-admin account? Or maybe some preferences? Custom *.js files? --Boris Tsirelson 19:02, 2 May 2012 (CEST)
This is from my test account with standard privileges, using firefox + "its all text." The little blue "edit" box is here under the lower right corner of the edit area, and it vanishes (sometimes) as soon as the cursor is outside of the edit area, so in order to fetch it you have to approach that corner with the cursor from inside the edit area. The preferences for "Editing" might be essential as well. I have marked
• Show preview before edit box
• Show preview on first edit
• Enable section editing via  links
• Mark all edits minor by default
• Enable dialogs for inserting links, tables and more
--TestUser 19:58, 2 May 2012 (CEST)
Wow, it works! It was because of the preferences! Now I'll check, which one exactly. --Boris Tsirelson 20:09, 2 May 2012 (CEST)

It was because of the "Show edit toolbar" preference. It was "on". --Boris Tsirelson 20:20, 2 May 2012 (CEST)

Now my edit preferences are

• Edit area font style: Browser default
• Show preview before edit box - no
• Show preview on first edit - no
• Enable section editing via  links - yes
• Enable section editing by right clicking on section titles (requires JavaScript) - yes
• Edit pages on double click (requires JavaScript) - no
• Show edit toolbar (requires JavaScript) - NO
• Mark all edits minor by default - no
• Use external editor by default - no
• Use external diff by default - no
• Prompt me when entering a blank edit summary - no
• Use live preview (requires JavaScript) (experimental) - no
• <vector-editwarning-preference> - no
• Enable dialogs for inserting links, tables and more - yes

and "It's all text" works. --Boris Tsirelson 20:35, 2 May 2012 (CEST)

That fixes it for me too -- great work Boris! --Jjg 22:36, 2 May 2012 (CEST)
Nice. Happy editing!
Ulf, you can mention it on our help page. --Boris Tsirelson 07:03, 3 May 2012 (CEST)
Done. --Ulf Rehmann 13:46, 3 May 2012 (CEST)
Nice. By the way, about "an even more convincing Example" there: also the possibility of automatic numbering is worth to mention, see the end of Section "Hadamard theorem" above. --Boris Tsirelson 17:20, 3 May 2012 (CEST)
The mathjax link on our help page does point directly to a tex example with labeling. Nevertheless.--Ulf Rehmann 20:27, 3 May 2012 (CEST)

## The language of this encyclopedia is ...

From searching for "distinguishing words" (fibre, fiber, centre, center, generalised, generalized) I think that the language is BE (British English), or as we Brits call it, English :-), with -ize endings. Also known as Oxford or OED English see here. Could others confirm this? --Jjg 11:18, 6 May 2012 (CEST)

The text of pages is (very) mostly British, while American English essentially seems to appear in bibliographic titles only. As an example, I did count 1219 files containing the word "generalized", where there are only 18 pages with the word "generalised", and all the latter do occur within in the references. --Ulf Rehmann 13:23, 6 May 2012 (CEST)
I feel a bit embarrassed (my mother tongue is pidgin English), yet probably we need to take into account the statistical habits of the readers. A modern student would, it looks, with greater probability search for center, not for centre. For the terms in the page names the problem should be better solved by redirects. And as for the language in which the bodies of the pages are (or will be) written, I suggest the freedom of choice ;-) otherwise we'll eventually end up with generalised centres - Sergei Yakovenko 14:24, 6 May 2012 (CEST).

## Orphaned pages

We have many orphaned pages. This should not be so, as the value of an encyclopedia like this grows with a higher number of inter-pages links.

As the examples of Rigidity or Isotopy (see their discussion pages) show, there are many more pages using some page title without a a link to the respective page (see entries for those pages under "What links here").

So editing/modifying pages should also include installing links for keywords which refer to other pages.

To some extent this could be done by a robot, but then there would be some danger of getting inappropriate links.

Any comments? --Ulf Rehmann 17:51, 6 May 2012 (CEST)

Could a robot create a page of "links to be added": a list of "page : list of links" entries, then humans could take those pages which are close their fields and add those links by hand (if appropriate) and delete the entry in the "links to be added" page.
--Jjg 18:03, 6 May 2012 (CEST)
Yes, it must be made selectively, by editor (not robot). Surely "Real number" (and "Quantifier") should not be linked every time when "for all real $x$" is written. Also, usually a link to article Y should be made only once (within article X), on the first occurrence. On Wikipedia links are selective, but our readers probably are assumed to have better background, thus, links should be even more selective. --Boris Tsirelson 20:59, 6 May 2012 (CEST)
And on the other hand, yes, notable connections between topics should be indicated. --Boris Tsirelson 21:01, 6 May 2012 (CEST)

## Isotopy (IN algebra)

For style uniformity we should choose between "X (Y)" and "X (in Y)", as "Isotopy (algebra)" and "Isotopy (in algebra)". I do not mind to switch to "X (in Y)" and move "Measure algebra (measure theory)" into "Measure algebra (in measure theory)", but first I am asking here. --Boris Tsirelson 09:44, 7 May 2012 (CEST)

Also, is it now our style, to make back links to disambiguation pages, as "For other meanings of the term see the disambiguation page Isotopy" on the top of "Isotopy (in algebra)"?

Let me also reproduce here a related question of Sergei from Talk:Isotopy (in algebra):

May be, we need standard template for uniform disambiguations? Something that would say "The term may mean one of the following..." and "For other meanings of this term see the page ...", while automatically updating the corresponding technical categories? Or I am inventing a bicycle? ;-)

--Boris Tsirelson 09:51, 7 May 2012 (CEST)

As with other style/bibliography/... issues, I think we can only cheerleader but not impose standards, as they will inevitably be violated when (hopefully) the broad popular masses of mathematicians will rush to contribute ;-) Creating convenient templates may be conducive to the spread of good manners whatever of them will be declared good. -- Sergei Yakovenko 10:00, 7 May 2012 (CEST)

## Commutative diagrams in TeX/LaTeX

How the TeXperts recommend to typeset them? I tried \includepackage{amscd} from AMS, but failed... -- Sergei Yakovenko 07:56, 8 May 2012 (CEST)

A good question... We have Help:Fancy diagram, but of course it is considerably more tedious than amscd. --Boris Tsirelson 10:12, 8 May 2012 (CEST)
Wow. Arrows are non-stretchable, everything manually aligned in direct contradiction of the original TeX spirit (to describe formulas, not to draw them like in WYSIWYG). Isn't it possible to fine tune MathJax to preload some standard packages? (I know that it is easier said than done :-( ) -- Sergei Yakovenko 10:41, 8 May 2012 (CEST)
Well, the above example is the "original" TeX spirit, as it is from Knuth's book. I drafted that in order to demonstrate the plain tex compatibility of mathjax. Admittedly it is not "the" convenient way to draw diagrams.
How is this:

$$\def\mapright#1{\xrightarrow{#1}} \def\mapdown#1{\Big\downarrow\rlap{\raise2pt{\scriptstyle{#1}}}} \begin{array}{ccc} A& \mapright{f} & B \\ \mapdown{g} & & \mapdown{g'} \\ A^*& \mapright{f^*} & B^* \end{array}\phantom{h} \qquad \qquad \def\mapdiag#1{\vcenter{\diagdown\kern-.4em\lower.63em{\searrow} \llap{\raise2pt{\scriptstyle #1\kern2pt}}}} \begin{array}{ccc} A& \mapright{f} & B \\ &\mapdiag{g} & \mapdown{g'} \\ & & B^* \end{array}$$


A bit more here, from which the above is quoted.
Another source:

$$\begin{array}{ccccccccc} 0 & \xrightarrow{i} & A & \xrightarrow{f} & B & \xrightarrow{q} & C & \xrightarrow{d} & 0\\ \downarrow & \searrow & \downarrow & \nearrow & \downarrow & \searrow & \downarrow & \nearrow & \downarrow\\ 0 & \xrightarrow{j} & D & \xrightarrow{g} & E & \xrightarrow{r} & F & \xrightarrow{e} & 0 \end{array}$$

--Ulf Rehmann 14:29, 8 May 2012 (CEST)
Here is a LaTeX encoding of the above "Fancy Diagram":

$$\def\cO{\mathcal{O}} \def\cC{\mathcal{C}} \def\cE{\mathcal{E}} \def\cL{\mathcal{L}} %%\def\ra#1{\kern-1.5ex\xrightarrow{\ \ #1\ \ }\phantom{}\kern-1.5ex} \def\ra#1{\xrightarrow{\ \ #1\ \ }} \def\da#1{\bigg\downarrow\raise.5ex\rlap{#1}} \begin{array}{ccccccccc} & & & & & & 0 &&\\ & & & & & &\da{} &&\\ 0 &\ra{} & \cO_C & \ra{\iota}& \cE & \ra{\rho} & \cL & \ra{} & 0\\ & & \bigg\| & &\da{\phi} & & \da{\psi} & & \\ 0 &\ra{} & \cO_C & \ra{} & \pi_*\cO_D &\ra{\delta}& R^1f_*\cO_V(-D) & \ra{} & 0\\ & & & & & & \da{\theta_i\otimes\gamma^{-1}} & & \\ & & & & & & R^1f_*(\cO_V(-iM))\otimes\gamma^{-1} & & \\ & & & & & & \da{} & & \\ & & & & & & 0 & & \\ \end{array}$$

--Ulf Rehmann 19:44, 8 May 2012 (CEST)

So, we have a list of predefined TeX macros, but we could also have a list of predefined LaTeX packages, llike AMSmath... --Boris Tsirelson 18:36, 8 May 2012 (CEST)
As it seems, the \require{AMSmath} isn't necessary, since it is collected automatically from the mathjax extensions. Anyway, the list of extensions can be seen in the source of any EoM page by searching for 'extensions':
extensions: ["tex2jax.js","TeX/AMSmath.js","TeX/AMSsymbols.js"]
--Ulf Rehmann 19:44, 8 May 2012 (CEST)

## Spam: what happens?

Discouraging and frustrating: the chief editor fights manually an increasing gang of spambots. Is it our stationary state?? Nonsense. It cannot work this way. In order to survive, the wiki must be protected.

Our situation is much easier than that of Wikipedia: (1) only registered users can contribute; (2) we are not interested in "innumerate" (that is, mathematically illiterate) contributors; (3) thousands of contributors are not expected these days (and hardly ever). Thus, a reasonably reliable anti-spammer barrier can be installed at registration. Moreover, if we use our specific features (above), no one will take the burden of creating a custom spambot just for this wiki.

So, what is happening?? --Boris Tsirelson 21:17, 18 May 2012 (CEST)   :-(

Oh well, I'm in contact with the sysadmins and have already proposed some measures (which I can't install myself, as I don't have server access). I am (as a maintainer) in charge of | this wiki as well. Here I did install the | "Checkuser" extension as well as this | Asirra module, The latter immediately stopped the actual spam rush for that wiki. The "Checkuser" allows to chase and remove spammers and their traces more efficiently than as it is possible by these inefficient standard web interfaces for blocking and deleting.
For this EoM wiki, I have proposed to install as well the "Checkuser" extension, but rather than taking Asirra I have proposed this QuestyCaptcha, possibly with some set of questions to be edited by us, so that they can be modified in case the spammers find out about them. We could ask questions concerning say MSC numbers and their areas etc. Of course, in the long run this may have to be changed as well, as the spammers seem to do everything to place their links into high ranked collections in order to improve their visability.
Also, I am preparing to set up an API access bot in order to detect and remove spam (sort of) immediately by means of a cron job or so. --Ulf Rehmann 23:43, 18 May 2012 (CEST)
I modestly repeat my suggestion: use subscription for MathSciNet as "pons asinorum" (via QuestyCaptcha or some other extension), tacitly assuming that no sane spammer will have it. For the few people not affiliated with a university to provide MSN, a "manual" option of writing a few words to the Editor will do the job - I wish we have more than a couple of new "genuine" contributors daily... -- Sergei Yakovenko 06:55, 19 May 2012 (CEST)
"Asirra" could give a remission for several months, hardly more, till the next crisis. ("Picasa" is already able to distinguish between photos of my friends; also, spammers can find human helpers; but hardly among mathematicians.) Typical solutions tend to be broken by newer spambots. The spammers seem to do everything to place their links into many wikis, but I do not think they will work hard for our wiki exclusively. It should be less profitable to them; then, they'll seek easier targets. --Boris Tsirelson 09:02, 19 May 2012 (CEST)
Today, Asirra was installed for account creation. Hopefully this will keep spammers out for a while. As far as I can tell, at this moment we have to run through this test for each edit also. I have proposed to the sysadmins to remove the test condition for edits, as my experience from another wiki shows that it is sufficient (at this moment) to have it just for account creation. --Ulf Rehmann 19:19, 23 May 2012 (CEST)
Another test: As it seems, only new page creation requires another test. --Ulf Rehmann 19:25, 23 May 2012 (CEST)
In fact, today there was another spam event, despite of asirra's cats/dogs. This may have been special, as the links pointed to an IT specialists company (probably by a single person, not a bot (unless they are testing some image recognition software). Anyway, Springer suggested to switch to registered accounts. I told them to watch the current situation for a few more days, since this drastic move would change the character of the whole site. Also, individual account admission would probably cause at least as much work as spam traces removal (every time a person has to act), and does not prevent fraud either. Any further opinions hereon? --Ulf Rehmann 13:12, 24 May 2012 (CEST)
"Another spam event"? It could be also treated rather as a new user that did not prove his helpfulness yet. No real harm to EoM (since he did not spoil articles). Instead of bocking him immediately, we could just warn him that only helpful users have the privilege to maintain userpages here. Probably, ultimately the result will be the same, but nevertheless it looks better (at least to me).
The page above was just openend with two commercial links. A convenient automated way to get the initial info mentioned above to a new user would be to have the user page initially opened up with appropriate hints. Does anybody know how to achieve this? --Ulf Rehmann 17:13, 24 May 2012 (CEST)
Maybe, MediaWiki:Welcomecreation, like that?
Thanks, good hint (I just did find that page as well), that could do. We could embed our adjusted policy page there. --Ulf Rehmann 21:17, 24 May 2012 (CEST)
Yes, "every time a person has to act"; at least two better (I think so) approaches were proposed here (one by Sergei and one by me). --Boris Tsirelson 14:41, 24 May 2012 (CEST)
Oh well, unless I missunderstood, both these proposals amount to setting up a QuestyCaptcha (which I suggested to the admins). But as soon as the questions all have been answered, the robots will know the answers and one has to change the setup. So I proposed to set it up in a way that the questions can be modified without sysadmin access. Apparently this was not possible. Also, at least one of our active users mentioned to me to have no access to mathscinet. There may be more. --Ulf Rehmann 17:13, 24 May 2012 (CEST)
The questions evidently can be modified, since they may be written on separate pages (and linked, or transcluded). Do you mean that the expected answers cannot be modified without sysadmin access? --Boris Tsirelson 20:16, 24 May 2012 (CEST)
Well, looking more closely on QuestyCaptcha I see that it hardly fits. Rather, I guess, MediaWiki PHP code should be tweaked. But only a little; at some point of the login process, to ask for a number and check the input. Everything else can be made on the wiki (help pages etc), I guess. --Boris Tsirelson 20:41, 24 May 2012 (CEST)
The code I know requires the questions and answers to be encoded in the file LocalSettings.php, which sits on the server machine and is not accessible to me. The problem is that there is only a finite number of questions, which will be solved soon and then quickly known to the robots. The last spam storms used similar access modi but did come from various ips. There seems to be an efficient network of infected machines out there.--Ulf Rehmann 21:17, 24 May 2012 (CEST)
OK, leave the MediaWiki code as is; install QuestyCaptcha; the question is fixed: "what is the answer to the math problem you saw on the relevant help page?"; and make a cron job that edits LocalSettings.php once a day, changing a single number within it; the new number is taken from a wiki page (its name being fixed once and for all); the remaining problem then is, whether a wiki page can be writable by you but neither writable nor readable by others. Can it be so protected?
Probably not: See this. --Ulf Rehmann 23:25, 24 May 2012 (CEST)
Or use a well-known cryptographic protocol: everyone can see the encrypted number, but the script can decipher it, since it knows your key. Or maybe the script can instead extract the number from your email sent to an appropriate "user" on the server.
Some variant of DES you are talking about? So you have to tell LocalSettings.php (or the local machine environment) how to either en- or decrypt. Sounds like writing another Captcha extension. --Ulf Rehmann 23:25, 24 May 2012 (CEST)
No. It is not difficult to write a program that does some well-formulated action. It is difficult to understand a big program written by others and modify it correctly (and MediaWiki is indeed not easy to understand). To tell the server how to decrypt? What is a problem? Such programs are widely used, and are contained by default in many Linux installations (including mine; I use it for entering my university account from home by ssh); see [5] and [6]. But after all, you could also send a new answer by email to a sysadmin, and he could edit LocalSettings.php accordingly; no more than 5 min of his time. --Boris Tsirelson 07:48, 25 May 2012 (CEST)
Yes, ssh and RSA/DSA based stuff is easy to use, the principles are well known and usable as programming snippets. But a setup which requires immediate interaction of some admin is not possible here. The things you suggest I could all do on a machine on which I have admin access, but not on the remote EoM server. I usually have to wait several days until I get something done. I had asked for a setup which allows me to modify questions and answers, but no practical solution was found so far. --Ulf Rehmann 10:00, 25 May 2012 (CEST)
OK, you reject the "send a new answer by email to a sysadmin" approach. But still, one approach is not (yet?) rejected: you put the encoded new answer to a dedicated wiki page (writable by you only), and a script on the server automatically (once a day) reads it, decodes it, and replaces it on LocalSettings.php. This is about the answer. About the question, even simpler: you replace it on a dedicated wiki page (writable by you only). And somewhere on help pages it is explained that everyone wishing to get an account must solve it first. --Boris Tsirelson 10:31, 25 May 2012 (CEST)
But wait, one more approach is not (yet?) rejected: you send the new answer by email to a script on the server (this is technically possible!), and then the script updates LocalSettings.php. Here the problem is that an enemy may also send such email. But I see some possibilities to reliably verify that the email is indeed yours (even if the enemy can read your emails; the more so if he cannot). If you find this way interesting, maybe some secret details should be discussed in private. --Boris Tsirelson 10:39, 25 May 2012 (CEST)
Yes, some people can solve such question. I only hope that spammers will find it not worth to do so just for spamming our wiki during several days. Why not seek an easier target instead? --Boris Tsirelson 22:04, 24 May 2012 (CEST)
And by the way, we could try to warn potential spammers (humans, not bots, of course) that our external links are "nofollow" (yes they are, I've checked) and therefore do not increase their Google rank. --Boris Tsirelson 14:45, 24 May 2012 (CEST)
Everybody knows that (at least those guys who are after ranking manipulation), because our source pages publicly tell. However, the effect of the nofollow attribute to ranking robots is very much disputed. [7], [8] --Ulf Rehmann 17:13, 24 May 2012 (CEST)

Comrades, correct me if I'm wrong reading the logs, but what I see is dozens of daily spam accounts manually deleted by Ulf, and hardly one or two new "real" people coming onboard in a week. In my view, at least until we see scores of new humans rushing to contribute to EOM, opening of a new account can be made "moderated". Consider the following scheme: each new human will be instructed to write (in a free style) a page describing his/her mathematical interests or other reasons why (s)he wants to be able to edit EOM pages, and then by one simple click Ulf will approve the right of the person to make visible choices. Until such approval the newly created page will remain in some DMZ where it is not visible to anybody except the candidate himself and the moderator. Once approving the new accounts created this way would become a major burden, the task of approval could be distributed among the members of the editorial board. Thus far I do not see hordes of genuine volunteers, only the spambots... -- Sergei Yakovenko 10:32, 25 May 2012 (CEST)

A "human moderation" scheme is used by the FoM (foundations of mathematics) mailing list. One applies for membership on a web-form, and then a few days later receive a curt email from the list manager asking for your mathematical background and interest in FoM, all rather intimidating for a neophyte, never mind a bot. There has never been, to my knowledge, a spam message on the FoM list. --Jjg 12:50, 25 May 2012 (CEST)
How to Cite This Entry:
EoM:This project. Encyclopedia of Mathematics. URL: http://www.encyclopediaofmath.org/index.php?title=EoM:This_project&oldid=26837