Some remarks by a (solid state) physics student.
On Wed, 09 Apr 1997 16:30:01 -0400, "Jeffrey S. Dutky" <·····@BellAtlantic.net>
wrote:
>Robert Harley wrote:
>>
>> ······@netcom.com (Henry Baker) writes:
>> > [...]
>> > The current assumption is that fast necessarily means 'hot' (in
>> > temperature and dissipation), although I can find nothing in my
>> > physics books that even remotely implies this connection. It's
>> > time to wipe the slate clean, and get some new ideas into this
>> > industry.
Fast does not necessarily mean hot.
Under special circumstances, though, fast can imply hot or vice versa, e.g. in
a classical ideal gas, higher temperature is directly related to a higher mean
velocity of the particles of the gas.
On the other hand, there are very cold, yet very fast systems. In fact,
classical thermodynamics does not even refer to speed: It holds in any inertial
frame.
>Computation is currently achieved by moving physical objects
>from one place to another as a means of symbolizing information.
>Moving these symbols around takes WORK since the symbols are
>physical things (beads, cams, relay contacts, electrons) and
>that work generates heat due to friction (see the first and
>second laws of thermodynamics).
It does not necessarily take work to move things around. It often does, but
not in general. If it takes work, this work does not necessarily generate
heat.
The laws of thermodynamics do not tell anything about friction. They can be
applied to systems with friction, but they do not make statements about the
existence of friction.
>The friction in a system can be minimized but not eliminated.
>Because information must by physically symbolized to be processed the production of
>heat by information processing devices is unavoidable.
There is no principal lower bound to minimizing friction other than zero.
Consider superfluid systems: You stir the liquid in a pot, go for your
vacation, and three weeks later, when you return, you see the liquid still
moving. So, if you thought, a liquid will not move by itself for longer than a
day, due to friction, this is a counterexample.
>Note: this may not hold true in superconducting or quantum
>devices. I don't know enough about superconduction or quantum
>mechanics to say. However, I suspect, that getting the information
>out of the quantum or superconducting computer will incur some
>amount of frictional work and thus will still generate some
>amount of heat. (although that amount might be much less heat
>per computational unit than we see today)
I think, what you have in mind is the following:
There are classical systems with friction, whose quantum counterpart does not
show friction: Consider Bloch electrons traversing a perfect ionic crystal:
Their states are stationary, which means they are not scattered. If you throw
a ball at a perfect lattice, you might observe a different behaviour.
Another aspect: There are means to produce heat other than friction. Have you
thought of them ?
>> The only way to avoid the connection is to do reversible
>> computations. That this is possible in both classical and
>> quantum physics was worked out by the late Jan van de Snepscheut
>> and Richard Feynman. But their ideas are very much theoretical
>> and come nowhere even remotely close to being realistic.
Does anyone remember the papers ?
>I think that the second law of thermodynamics will have something
>to say about heat production even in the case of reversible
>computing.
Indeed. It says, that the entropy (which you might consider a synonym of heat)
is constant for reversible computing.
>In the long run entropy always increases.
No, see above.
>> I probably don't know what I'm talking about so feel free to
>> correct me!
>>
>> -- Rob.
Same for me.
>I probably have something wrong so feel free to correct me as well
>(not that anyone NEEDS my permission).
>
>- Jeff Dutky
So do I.
Bye,
Dirk