[seqfan] Re: metadata for Mathematica code

[israel at math.ubc.ca]

There's also potentially an issue with taking up large amounts of real 
estate on an OEIS page with, perhaps, only slightly different variants of 
code in various languages (today Mathematica, but Maple, Pari, and all the 
rest won't want to be left out).

Eventually, it would be nice to have OEIS pages displayed in a customizable 
way with not all fields visible unless you click on them. Thus someone 
who's not a Mathematica person might not, by default, see Mathematica code. 
I guess this could be implemented using cookies.

Cheers,
Robert

On Nov 2 2015, hv at crypt.org wrote:

>It's great to see someone thinking along these lines; I think it would
>be valuable to state (or develop) the definitions in one or more pages
>on the OEIS wiki.
>
>Is it intended that these special comments would be significant iff they
>appear on their own as a comment, or even if the label appears as a
>word in a longer comment? (Given the examples TermFunction and
>TermFunctionDefinition, I guess "substring" already doesn't cut it;
>I'm in two minds about whether that's wise.)
>
>I think I understand what most of the labels represent, but I'd be more
>comfortable if they were expressed in a more formal maths/computing style.
>I've never used Mathematica, so the given example is of limited value
>(and a range of examples in other languages would be useful).
>
>I'm confused about TermFunction and TermFunctionDefinition: I'd have
>expected the example of the former would require that the example of
>the latter define a function called 'CatalanNumber'.
>
> Is it intended that the comment, along with the choice of language, be 
> sufficient in themselves to let you automatically invoke a function 
> appropriately? Or is it expected that invokers will still need to inspect 
> the source code or additional documentation to do that? I believe the 
> former would be incredibly useful, but the mechanism to get from comment 
> to valid invocation would probably need to be specified on a 
> language-by-language basis.
>
>Note that some languages have explicit mechanisms to specify metadata
>separate from comments (eg Java); there'd be some (limited) value in
>specifying the use of comments as the fallback only when the language
>doesn't already provide a better mechanism, rather than sanctifying it
>as the one true way.
>
>I think it may also be useful to define a mechanism to tag characteristics
>of a particular implementation: you might have, for example, a simplest
>possible algorithm for didactic purposes, and multiple additional versions
>that make different trades between CPU and memory use. I could imagine
>that developing over time into machine-readable performance characteristics
>that are simultaneously human-readable (variants of) big-O notation.
>
>Hugo
>
>Eric Rowland <rowland at lacim.ca> wrote:
>:Hi all,
>:
>: Wolfram Research has a mentorship student, JungHwan Min, who is going to 
>: be contributing Mathematica / Wolfram Language code to OEIS entries. I 
>: think this is a great opportunity to start adding metadata identifying 
>: what each program does (computes the nth term vs. computes the first n 
>: terms, etc.). Having more structured data in the OEIS will make it that 
>: much more useful, especially for programmatic tasks. You can imagine 
>: eventually being able to query for a list of all functions that define 
>: the nth term of a 2-automatic sequence, for example.
>:
>: Charles Greathouse and I discussed this at length about a year ago and 
>: came up with an initial list of labels. It makes sense to put labels in 
>: comments, so for example the Mathematica field for A000108 would look 
>: like this:
>:
>: (* TermFunction *) CatalanNumber (* TermFunctionDefinition *) A000108[ 
>: n_ ] := (2 n)!/n!/(n+1)! (* TermFunctionDefinition *) A000108[n_] := 
>: Hypergeometric2F1[1 - n, -n, 2, 1] (* Richard L. Ollerton, Sep 13 2006 
>: *) (* TermListByIndexFunctionDefinition *) A000108list[nmax_] := Table[ 
>: CatalanNumber@ n, {n, 0, nmax}] (* Robert G. Wilson v, Feb 15 2011 *) (* 
>: TermListByIndexFunctionDefinition *) A000108list[nmax_] := 
>: CoefficientList[InverseSeries[Series[x/Sum[x^n, {n, 0, nmax+1}], {x, 0, 
>: nmax+1}]]/x, x] (* Mats Granvik, Nov 24 2013 *)
>:
>:Here is an initial list of labels:
>:
>: * TermFunction - function that computes the nth term (respecting the 
>: offset) * TermListByIndexFunction - function that computes the list of 
>: terms up to the nth term * TermListByMaximumFunction - function for 
>: computing the list of terms <= m for a non-decreasing sequence * 
>: TermListFunction - other function for computing a list of terms (where 
>: it may not be clear in advance how many terms it computes) * 
>: MembershipTestFunction - predicate function for testing membership, 
>: returning True or False
>:
>:For a function definition, append "Definition" to the label (getting
>:TermFunctionDefinition / TermListByIndexFunctionDefinition /
>:TermListByMaximumFunctionDefinition / TermListFunctionDefinition /
>:MembershipTestFunctionDefinition).  For example,
>:   CatalanNumber
>:and
>:   Function[n, CatalanNumber[n]]
>:would get the label TermFunction, whereas
>:   a[n_] := CatalanNumber[n]
>:is a TermFunctionDefinition.
>:
>:Some entries have programs with hard-coded parameters, so they are not
>:functions or function definitions.  These could get labels such as the
>:following.
>:
>:* TermList - code that produces some list of terms
>:* Visualization - code that produces related graphics
>:
>:Undoubtedly there will be others.
>:
>:Comments?  Where is the right place to document a list of "official"
>:metadata labels?
>:
>:Eric