non-binary computers?

From: Eric Smith <eric_at_brouhaha.com>
Date: Fri Sep 3 12:41:42 1999

"David C. Jenner" <djenner_at_halcyon.com> wrote:
> You do need to have some way of delineating the data you are crunching.
> Feynman does this in the context of a Turing machine, where he writes
> "tape marks" of arbitrary content, but specific significance, to indicate
> the state of the process.

OK. So it's not really a unary computer after all.

Note that Turing machines actually need at least two symbols. However,
it is common for programs for binary Turing machine to represent a
positive integer as the count of the number of ones between a pair of
zeros that serve as delimiters.

When I was playing with Turing machines, I tried to use a true binary
place-value representation using two tape locations for each bit, so that
one of the tape locations of the pair was the field delimiter. However,
the code quickly got too complex and hurt my brain. (Hmmm... maybe the
damage was permanent; that would explain a lot.)
Received on Fri Sep 03 1999 - 12:41:42 BST