6
$\begingroup$

Sorry if this question is too soft or off-topic, but I wasn't sure where else to get feedback from a broad range of subfields.

Short version: How should one condense the 32 existing arXiv math subject tags into approximately 10 "subject areas"? There can be overlaps between the subject areas, but ideally researchers within a given area would have similar preferences for specialized journals.

Long version: After discussing with some other early-career researchers, I'm thinking about putting together a ranking of math journals based on survey responses from established researchers. This would by no means be a definitive ranking, but it would at least allow one to get a sense of "expert opinions" about various journals.

As a feature, I would like to allow one to filter results based on the primary research area(s) of the respondents. However, I do not plan on stipulating how many journals a given respondent should rank, so I will have to use something like Plackett--Luce to generate an overall ranking. The difficulties arise when generating this ranking:

  • If I generate this ranking for all possible combinations of arXiv tags, I need to run a Plackett--Luce implementation $2^{32}$ times and store the resulting rankings. That's not going to work, but this approach should be just fine for around $2^{10}$ combinations of subject areas.

  • Given a choice of subject areas, I could generate this ranking live. However, the graphical display of the rankings would take a non-user-friendly amount of time to update. Also, the only Plackett--Luce implementatons I've seen are in R, and I really should be getting ready for job applications instead of trying to figure out how to run R on a website.

One option is the classification used in the AMS annual survey on "doctoral recipients by field of thesis," but it's not clear to me where e.g. math physics would belong. These categories are:

  1. Algebra / Number Theory
  2. Real, Complex, Harmonic, and Functional Analysis
  3. Geometry / Topology
  4. Discrete Math / Combinatorics / Logic / Computer Science
  5. Probability
  6. Statistics
  7. Biostatistics
  8. Applied Math
  9. Numerical Analysis / Approximations
  10. Linear and Nonlinear Optimization / Control
  11. Differential, Integral, and Difference Equations
  12. Math Education
  13. Other
Improve this question
$\endgroup$
9
  • 8
    $\begingroup$ The AMS division seems verrry weird to me. For instance, breaking probability and statistics into two different groups, but grouping discrete math and logic together? $\endgroup$
    – Sam Hopkins
    Commented Jul 28, 2021 at 21:05
  • 3
    $\begingroup$ Btw, you might be interested in this older MO question: mathoverflow.net/questions/3512/top-specialized-journals $\endgroup$
    – Sam Hopkins
    Commented Jul 28, 2021 at 21:06
  • 7
    $\begingroup$ @SamHopkins ... In large universities: statistics is often a separate department from math, whereas probability is (a small) part of the math department. $\endgroup$
    – Gerald Edgar
    Commented Jul 28, 2021 at 21:13
  • 2
    $\begingroup$ Conveniently, the NSF Division of Mathematical Sciences has exactly 11 disciplinary research programs. If you exclude statistics (which is often a different department) , that brings it down to 10. nsf.gov/funding/programs.jsp?org=DMS $\endgroup$
    – Willie Wong
    Commented Aug 28, 2021 at 1:34
  • 4
    $\begingroup$ One problem with any attempt to split mathematics into fixed areas is that a reasonably good classification at some time can become awkward 20 years later. $\endgroup$
    – YCor
    Commented Aug 28, 2021 at 10:31

1 Answer 1

Reset to default
1
$\begingroup$

Technical solution to an analytic problem, sounds my style: The littler package, to be read as "little R" is a CLI interface to R, so running R on a website should be as straightforward as a shellout to that. It's installable as a system package in Debian (and would you run a web-server on anything else?).

Improve this answer
$\endgroup$
1
  • $\begingroup$ Thanks for the tip! There’s still the issue of the update time when computing the ranking, but this will be very helpful if I need to go this route $\endgroup$
    – Stephen McKean
    Commented Jul 28, 2021 at 21:55

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .