[seqfan] invasion of third rate programs
Joerg Arndt
arndt at jjj.de
Mon Dec 17 18:50:55 CET 2018
Dear seqfans,
here is something that really bothers me, and for quite some time.
Some contributers put in a lot of programs, in several languages,
but exclusively for sequences that are trivially to compute.
The problem is that quite a few of those contributers produce
programs of very questionable quality.
In https://oeis.org/draft/A009949
"Factorial expansion of sqrt(2) = Sum_{n>=1} a(n)/n!, using greedy algorithm."
there is (in FOUR languages, here just the Magma version)
SetDefaultRealField(RealField(250)); [Floor(Sqrt(2))] cat [Floor(Factorial(n)*Sqrt(2)) - n*Floor(Factorial((n-1))*Sqrt(2)) : n in [2..80]];
For every digit the constant is evaluated twice and the idiom
Floor(Factorial(n)*Sqrt(2)) - n*Floor(Factorial((n-1))*Sqrt(2))
is unmitigated idiocy, just as
Floor(2^n*Sqrt(2)) - 2*Floor(2^(n-1)*Sqrt(2))
would be for computing the binary expansion.
The idiom above has already spread, see
https://oeis.org/search?q=%22floor%28n%21%2ab%29%20-%20n%2afloor%28%28n-1%29%21%2ab%29%29%22&fmt=short
Another class of stupidly inefficient programs goes as follows.
To compute one term, compute the whole sequence up to that terms
and discard all terms but the final one.
Yet another is using pow and log for digit operations.
This has the added benefit of becoming wrong all the time.
There is NO excuse for such bone headed programs, they just
make the OEIS look bad.
The problem just described appears predominantly with people
jumping into other peoples edits. We simply should forbid
that (by technical means).
Hints:
1) Learning to program is fun, but don't just dump every morsel
produced on the way into the OEIS.
2) If your goal is just to put your name as often into the OEIS
as you possible can, then you are doing something VERY wrong.
Best regards, jj
P.S.: I may not have a chance to check seqfan in the next few days,
just so nobody wonders if I reply only around Christmas.
More information about the SeqFan
mailing list