the concept of tables (not just one, surely!) mapping math and scientific
symbols to their meanings in particular fields is undoubtedly worthy,
and in fact, the question has been asked before, but, as far as
i am aware, nothing organized has ever been done about it.
in fact, accomplishing such a feat may be next to impossible.
or, as some competent mathematicians have informed me, it may be
a useless exercise. why? first, because so many symbols are
used, often with different meanings, in different areas.
also, a mathematician can define his/her own notation,
and if there's not already a well established symbol for a concept
(which would be known to a mathematician experienced in the area),
a new one will often be selected based on its shape relative to
that of symbols already used for related concepts, regardless of
the new symbol's meaning in other areas.
so, such tables would be of use mostly to newcomers in the field,
mainly graduate students, and few established mathematicians
have the incentive or interest in doing what for them would
probably be a rather menial job. and the students are usually
too busy working on their research, which will lead to the
formal recognition of a degree, while working on very useful
tex-related projects gains nothing more than appreciation.
(more than one degree has foundered on such a shoal.)
it may be instructive to consider how some existing symbol collections
were compiled. the basic cmsy
and cmex
fonts provide the symbols
that knuth needed for the art of computer programming. that's
a basic computer science collection. the additional amssymb
collection (msa
and msb
fonts) were based on what had been
used or required for ams publications prepared by earlier means,
including the symbols collection provided by the science
typographers software augmented with items from the monotype
symbols lists -- none of them identified by anything but an
access code meaningless except within the context of that
composition system. the control sequence names assigned to
the "ams symbols" were usually just the names used by proofreaders,
who were often not even mathematicians, and certainly not area
specialists.
the stix collection, which was the basis for the massive increase
of technical symbols in unicode 3 and 4, started with the cm
and
ms
fonts (no name changes), the symbols component of sgml entity
sets (iso tr 9573-13), "needed" lists compiled by the stipub
organizations (ams, acs, aip, aps, ieee, siam, and elsevier), and some
additional contributions from wolfram and design sciences. but again,
no area identification was included.
compiling area-specific lists requires specialist knowledge,
i.e., people. let's leave that aside for the moment.
what approach(es) might be considered for compiling lists by frequency?
only one comes immediately to mind: from a corpus of (la)tex publications, count
the occurrences of all control sequences, ignoring those that are clearly not symbols (\chapter
, \section
, \begin
, \end
, etc.).
but there are problems. authors often define their own macros for either individual symbols or
preformed strings of symbols, so the uses of the symbols themselves in
the body can't easily be counted. even worse, many authors, over
time, compile great collections of macros that they've used before and
might use again, and simply dump those into either the preamble or a
separate .sty file without any "weeding", so it's not easy to tell
whether a particular macro (and thus symbol) is actually used in a job.
it is possible to filter for a few specific symbols by macro name;
i've done it to document usage as required for acceptance into unicode,
but it's non-trivial. what's necessary is an automated procedure.
here are steps that might be applied to article files:
- expand all macros in the body, so that the definitions aren't needed
any more;
- remove all definitions from job;
- tex to make sure nothing is lost;
- confirm that the output is the same as the original.
undertaking such a task would need a good programmer and a really
dedicated tester and/or support from some organization (probably not
publishers, other than for access to files, since they're already hard
pressed).
now, how to identify the areas in which these symbols are used?
even if the area of an article is well defined (say by a subject
classification as defined for mathscinet), it's not clear that
it's a good example of symbol usage, so any automatic compilation
of symbols would still require manual checking.
i think the only reasonable prospect for creating area-specific
lists would be action by knowledgeable humans. any volunteers?
Best Answer
For Linux, there are two nice character selectors, KCharSelect for KDE,
and Gucharmap for GNOME,
You can browse the various characters by script or Unicode block, or search by character description. Both programs are very nice, but personally, I prefer KCharSelect, as it has an incremental search bar which makes it faster to find the characters you need, and it displays all the information about a character, without you having to switch between the glyph ťab and the character information tab.
For Windows, you can use the built-in Character Map program, which supports Unicode in later versions of Windows: