Re: Reversible Computers

Michael Nielsen (
Sun, 28 Jun 1998 11:37:54 -0600 (MDT)

On Sat, 27 Jun 1998, John K Clark wrote:

> On Sat, 27 Jun 1998 Michael Nielsen <> Wrote:
> >Our solid state computing devices have error rates in the
> >range, while single particle systems are lucky to get down to
> I don't know what you mean by "single particle systems", or where you got
> that figure.

10^{-18} is an estimate from a condensed matter physicist and
colleague at IBM. 10^{-2} is from scores of research talks by people
working on this kind of thing, in many different architectures. Most of
the talks have been oriented towards quantum computing, but a good
number have been oriented towards conventional computing as well.
10^{-2} may be slightly pessimistic; 10^{-3} is not, in my opinion, and
10^{-4} is wildly optimistic.

I'm not sure what the problem is with "single particle
systems". Admittedly, they are not the only subject of
discussion here, but they provide a convenient reference point
at one end of the size spectrum.

> >Could you please send me a reference to Merkle's work on this?
> Drexler discuses Merkle's work on pages 82-85 in Nanosystems, he also lists
> many other articles by Merkle and others in the field of Reversible

I'll have a look. Thanks also to Anders Sandberg for the link to Merkle's
page. In the "interesting facts" file, I'll mention that while a 1973
paper by Bennett in the IBM Journal of Research and Development is
usually credited as the discovery of reversible computation,
the following paper:

title={Machines de {Turing} r\'eversibles},
journal={Comptes Rendus},

appears to be where reversible computing first appeard. I say "appears"
because I don't read French; I have been told that it does contain an
architecture for a universal, reversible Turing machine. The Bennett
article is

title={Logical reversibility of computation},
journal={IBM Journal of Research and Development},

Michael Nielsen