next up previous
Next: Example 2 Up: Examples for superseeker and Previous: Examples for superseeker and

Example 1

We begin with a classical analytic example, from ``Pi, Euler Numbers, and Asymptotic Expansions'' [1].

R. D. North asked for an explanation of the following fact:


The number on the right is not tex2html_wrap_inline108 to 40 places. As one would expect, the 6th digit after the decimal point is wrong. The surprise is that only the underlined digits are wrong. This is explained in detail in [1]. The discovery of the explanation is quite difficult from this result, but is somewhat easier from the following similar one:


Here we note that if we add 1, -1, 5, and -61 to the incorrect digits, we get equality (to 40 places) with tex2html_wrap_inline118 . With the help of Sloane and Plouffe (or indeed with the help of Sloane's original Handbook, as was actually the case, or the computer programs) we can identify these as the first four nonzero Euler numbers. We conjecture, then, that the error is of the form


where tex2html_wrap_inline120 is the tex2html_wrap_inline122 nonzero Euler number. We can test this conjecture by computation, and find by adding the first 80 terms in the error formula above to the sum that we get tex2html_wrap_inline118 to 500 digits. This does not tell us that our conjecture is true, but at least it encourages us that a proof might be possible. See [1] for the proof.

The point of this example was that recognition of the Euler numbers in this at first required ingenuity (to shift from tex2html_wrap_inline108 to tex2html_wrap_inline118 , because the original problem has twice the Euler numbers appearing in it). However, the case has changed: the new programs both recognize twice the Euler numbers.

If we do not give enough terms to superseeker, it fails to return anything (the heuristics of the program are not designed for short sequences, which, after all, can represent far too many things to be really useful). If we put in 7 terms, however, it returns

Report on [ 2,2,10,122,2770,101042,5405530]:
Many tests are carried out, but only potentially useful information
(if any) is reported here.


        (limited to 10 matches):

Transformation T003 gave a match with sequence A0364
Transformation T004 gave a match with sequence A0364

List of sequences mentioned:

%I A0364 M4019 N1667
%S A0364 1,1,5,61,1385,50521,2702765,199360981,19391512145,2404879675441,
%T A0364 370371188237525,69348874393137901,15514534163557086905,
%U A0364 4087072509293123892361
%N A0364 Euler numbers: expansion of sec $x$ .
%R A0364 AS1 810. MOC 21 675 67.
<some cryptic material omitted>
References (if any):

[AS1] = M. Abramowitz and I. A. Stegun, { Handbook of Mathematical 
Functions}, National Bureau of Standards, Washington DC, 1964.
[MOC] = { Mathematics of Computation} (formerly { Mathematical Tables and 
Other Aids to Computation}).

List of transformations used:
T003  sequence divided by the gcd of its elements
T004  sequence divided by the gcd of its elements, from the 2nd term

Abbreviations used in the above list of transformations:
u[j]    =       j-th term of the sequence
v[j]    =       u[j]/(j-1)!
Sn(z)   =       ordinary generating function
En(z)   =       exponential generating function

The Euler numbers appear as sequence M4019 in the book. [The code here is to the explicit tag in the book; A0364 is an internal absolute code while T003 tags the transformation used.]

next up previous
Next: Example 2 Up: Examples for superseeker and Previous: Examples for superseeker and