[seqfan] verifying A153063 with a(n)=a^2-n^a

[Richard Mathar]

Help! I am running into problems with A153063 http://oeis.org/A153063 .
I thought a(-1)=1, a(n)=floor(a(n-1)^2-n^a(n-1)) with offset 0 would work,
with a glance at the Mma code. Basically the step 4 -> -2092 means one
would need to have some 16-(...)^4 to generate -2092, and that seems
to be impossible.

RJM