Re Recaman

[N. J. A. Sloane]

Eric Rains and I concluded yesterday that probably lim sup R(n) = constant n log n
but that still leaves it possible for every term to eventually appear

David, can you hack a C program that would go out to 10^10 or further?
I'll send you my fortran program, which stores 1 term of the "hit" sequence
per byte.  Of course with C you can store one term per bit, so we could go at least 8 times as far in C.
The fortran program calls for 15 gigabytes, but very little of that needs
to be in memory at any one time - the "hit" array is
very sparse.  So i was able to run it on our 8 gigabyte machine and get a billion terms.

If you don't have that much memory then i could run it here.
In any case, if you /do/ feel inclined to write the C version,
i would like to get a copy, because i am still trying to analyze
this sequence.

But the main question that the supercollider program would try to
answer is whether 1355 occurs

Neil