Am Mittwoch, 21. September 2016 13:49:57 UTC+2 schrieb Johan S. R. Nielsen:
> With MR numbers, do you mean a link of the type [MR3352496]? 
Yes!  (Except, that in a compiled document such links could be transformed 
into a link such as [1].

> > well, for preprints clearly there is of course the arXiv number and for 
> > sciences without a good database, there is doi. 
> Neither arXiv nor DOI completely catalogues all publications. I don't 
> know how many such cases appear in Sage's bibliography of course. 

Exactly, that's why findstat uses all of them (and more), but identifies 
references with the same (up to some noise) title and authors. 

> > concerning readability, there is a well known justification for using 
> > sequential numbers 
> Can you clarify? How would sequential numbers work? The documentation of 
> Sage is never read in sequence but more like random access. 

What I meant is the common referencing scheme in many mathematics 
publications.  Of course, if you read something like

    See [1] and [2,4] for background and recent developments.

that's terrible.  However, if the author cares at all, (s)he would have 

    See the textbook by Foo [1] and the recent papers by Harry and John 
[2,4] for background and recent developments.

A reference like [Tho2000] is to me much more recognisable than 
> [MR1794692]. Having two or three of the latter kind of references in a 
> text, it takes brain-effort simply to distinguish if they are different 
> or not. 

That is very true. 

In articles and books, [Tho2000] is a much more popular format, and I 
> guess for exactly this reason. Of course in such publication sizes the 
> scalability problems don't show well, which could be the case for 
> Sage.

> I just don't think so: in the current master bibliography, there's 1130 
> references. There's *2* collisions with the current naming scheme 
> (broken by appending 'a', 'b', etc.)! 


> I'm not making this up, I used this to organise the references for 
> >, and I'm very happy with the result. 
> Can you elaborate? When I look at e.g. 
> then the references are [KTT04], [Lo04] and [Sta99].

Yes, things are more complicated than advertised.

So, part of the truth is as follows 
or for examples

* Alice submits a statistic using, and in the 
reference field, she is (strongly) encouraged to type something like

    [[MathSciNet:3338726]] [[arXiv:1304.4309]]

* FindStat retrieves bibtex from the mathscinet and arxiv catalogues and 
checks that they indeed refer to the same paper (by definition: authors and 
title must match).  The identifier from the most preferred catalogue is 
then used as bibtex key in our bibtex file.  If somebody references this 
paper again, the bibtex file will not be modified.

Within a statistic entry, the references are simply numbered [1], [2], [3], 
...  Currently, there is no automatic linking between numbers used in the 
description field of the statistic and the list of references.  In this 
case, I think it would be overkill.

* in case Alice did not use an identifier but rather free text, FindStat 
sends this to and tries to obtain an MR 
number.  Failing that, the reference is left as Alice types it.

* On the webpage, we show then author, title, and identifiers.

I would have suggested to write "As shown by Chern et al. 
[[MathSciNet:3338726]] one can do this at that" because it works for me (I 
also use this in my LaTeX files, using reftex), but I also admit that I do 
not care enough to advertise it more :-)

The main advantage is that one then has automatically unique identifiers, 
which can (but need not) be transformed easily into whatever you want for 
the end user (eg., the one reading docstrings).


You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
To post to this group, send email to
Visit this group at
For more options, visit

Reply via email to