[seqfan] Re: Maple program

Brendan McKay Brendan.McKay at anu.edu.au
Sun Oct 25 14:11:06 CET 2020


This raises a question that applies not only to Maple but other
languages:  which versions of the software should they work
with?  Is this stated in the instructions anywhere?  I can't find it.

Especially with costly packages like Maple and Mathematica,
many people keep running the same version for quite a few
years before upgrading.  I know several colleagues who are
still using Maple 2015 -- they can't run Peter's code because
the NumberTheory package was introduced in 2016.  (The
old package still works with current Maple, though.)  Even
for free packages, using brand new features will force people
to upgrade (or, more likely, not use the code given) and this
isn't nice or, usually, necessary.

So, is there a policy?  If not, should there be?

Incidentally, while poking around I found
https://oeis.org/wiki/Maple
which says "The latest version of Maple, as of January 2014,
is version 17."  The latest version is Maple 2020, but I
suggest removing the sentence altogether as it will soon
be out of date again.

Brendan.

On 25/10/20 9:24 pm, Peter Luschny wrote:
> A program for the OEIS is also a publication on which further
> criteria, such as legibility and comprehensibility, are applied.
> Often, one-liners do not meet these criteria but are intended
> to show the writer's technical mastery and tend towards CodeGolf.
>
> So it is advantageous to read the notes of Charles Greathouse [1]:
> "Programs should be kept to a minimum reasonable size (but
> please don't golf)."
>
> Good readability still allows a wide range of styles. Mathar
> seems to prefer the procedural style, the loops. I'm
> more of a fan of Leslie Lamport, who once said in a lecture
> in Heidelberg: "As soon as you write 'for' you have lost."
> The functional style breaks a program down into small individual
> functions, which are then put together.
>
> My solution looks like this:
>
>      gpf := n -> max(NumberTheory:-PrimeFactors(n)): # A006530
>      isA081258 := n -> evalb(gpf(n^3 - 1) <= n):
>      A081258List := upto -> select(isA081258, [`$`(2..upto)]):
>      A081258List(400);
>
> According to Charles, the first two lines, the is-function,
> are enough, while the list-function and the actual call should
> not be listed. I add them mostly because I would like to
> generate the sequence after a copy-and-paste without any
> further action on my part.
>
> The type of program, as described here, is generally applicable
> for so-called 'lists.' Please note that these lists have the
> offset '1' by definition, so this is /not/ the lower limit of
> the selection range! In fact, in the present case, this is 2,
> not 1. (Such distinctions are an eternal source of confusion
> in the OEIS.)
>
> So we have indeed found a bug in Mathar's program: Since his
> for-loop starts with 1, he also generates a '1', but this is
> not a term of the list!
>
> Moreover, this does not lead to an error due to one (of several)
> bugs in Maple's 'numtheory' module. Bugs hide bugs. (You should
> switch to the new 'NumberTheory' module as soon as possible,
> or, better, to SageMath.) The current version quits Mathar's
> program with the error message:
>
>      Error, (in NumberTheory:-PrimeFactors) cannot represent
>      all prime factors of 0.
>
> Cheers, Peter
>
> [1] http://oeis.org/wiki/User:Charles_R_Greathouse_IV/Programs
>
> --
> Seqfan Mailing list - http://list.seqfan.eu/


More information about the SeqFan mailing list