[ExI] D-Wave's Quantum Computer
Anders Sandberg
anders at aleph.se
Thu May 23 12:53:29 UTC 2013
On 2013-05-23 08:11, Giulio Prisco wrote:
> On Thu, May 23, 2013 at 8:17 AM, Anders Sandberg <anders at aleph.se> wrote:
>> Incidentally, since we are talking quantum computation. I have been trying
>> to track down a proper source on minimum energy dissipation for it. As far
>> as I get it, by virtue of reversibility QCs can run with no dissipation -
>> they do not erase bits, so the Landauer limit doesn't apply. But error
>> correction will be necessary, and error correction schemes involve dirtying
>> ancilla bits - these bits will have to be erased, at a thermodynamic cost.
> Can't the extra bits be just stored?
It gets rather expensive to have to build more and more memory. In fact,
creating a new bit in a fixed state is *at least* as expensive as
erasing a bit, since it involves erasing the unknown "bit" existing in
the original matter.
The problem with dirty ancilla bits is that they cannot be undone by
running the computation backwards, which is how you otherwise deal with
garbage in reversible computation.
> BTW what do you guys think of this?
> It seems sort of plausible that the universe may be optimally
> energy-efficient. So it should be a reversible computer. But
> apparently it isn't because information is irreversibly erased in
> wavefunction collapse. But if the information is just stored away
> instead of destroyed, the universe can still be a reversible computer.
> Assuming the MWI interpretation, the lost information is, indeed,
> still available. So the MWI follows from Landauer theorem with some
> plausible assumptions.
Well, if you view wavefunction collapse as decoherence, then all the
information is still there. It just gets embedded in the background,
like how the energy of a sound becomes thermal vibrations.
I don't think this the idea is true. You could have a non-MWI universe
running on a computer which would not be energy-efficient even if the
universe internally was perfectly reversible.
Landauer's inequality is also not really a theorem (most, if not all
derivations are flawed), although it looks like a real law of physics.
--
Dr Anders Sandberg
Future of Humanity Institute
Oxford Martin School
Oxford University
More information about the extropy-chat
mailing list