Review: A048799
David Wilson
davidwwilson at attbi.com
Thu May 30 16:22:55 CEST 2002
A048799 needs some work. The few values of A048799 are inaccurate, but I
cannot fix the sequence without the help of extended-precision arithmetic.
Let S(n) = A002034(n) = smallest k >= 1 with n | k!. A048799 purports to be
the decimal expansion of the constant
%F A048799 Sum (1/S(n)!), where S(n) is the Smarandache function A002034
and n >= 2.
For simplicity, I would like to change this formula to
%F A048799 SUM(j = 1 to inf; 1/A002034(j)!)
which serves only to increase the constant by 1.
To compute the constant efficiently, do as follows:
Let z be the smallest number such that z! has 200+ digits (this
is probably overkill, but hey).
Now, compute a(n) = A027423(n) = d(n!) for 1 <= n <= z. I
suspect that Mma could compute d(n!) directly with fair
efficiency, if not, you can use the formula:
a(n) = PROD(p <= n, p prime; I(n, p)+1)
where
I(n, p) = (n-S(n, p)) / (p-1)
where
S(n, p) = sum of base-p digits of n.
As a check, a() should start with
1,2,4,8,16,30,60,96,160,270,...
Now, compute b(n) for 1 <= n <= z as the differences of a(n):
b(n) = 1 if n = 1; a(n)-a(n-1) if n > 1.
b() should start with
1,1,2,4,8,14,30,36,64,110,...
b(n) gives the number of values k with A002034(k) = n.
Finally, we compute the constant
c = SUM(j = 1 to z; b(n)/n!).
Our choice of z should make this accurate to 108 digits. These 108
digits constitue A048799. As a check, you should get
c = 2.093170459...
So we should have
%S A048799 2,0,9,3,1,7,0,4,5,9,...
And the 108 digits should fill out the sequence.
More information about the SeqFan
mailing list