Re: [Fis] Szilard's Engine and Information

From: John Collier <[email protected]>
Date: Tue 13 Apr 2004 - 09:03:49 CEST

Some brief comments (I'd like to say more, but I have a lotto do this week).

At 12:40 AM 2004/04/13, Viktoras Didziulis wrote:

 How would behave a mixture of oil and water if Earth's field of gravitation was absent - would it mix or separate ?

I would guess that it would separate. Reasoning is as following: oil is hydrophobic, so it tends to expel water. Globules of oil would form. Gradually the rest of the water would migrate from the oil, leaving a pure oil region (it gets out, but it can't get back in). A similar thing happens to water in marginal permafrost zones (one of my areas of expertise in my past career as a geophysicist). What happens is that random motion of water takes it to low temperature regions, where it freezes out. This can lead to, e.g., gas pipelines popping (ripping?) out of the ground due to accumulated ice. Once they are captured in ice, the water molecules can't easily get away again. Another factor is surface tension. I am told that Benard cells can form even in space, due to surface tension, with no gravity. In this case, surface tension would tend to keep the water molecules in.

On the other hand, the process of separation would be slower (gravity absent means that the globules of water left in the oil would not fall). So perhaps the formulation below is too simple.

Can this be described by a conditional sentence like the one below ?
 
if (gravity present) {separation}
else {mixing}

if(gravity present) {rapid separation}
else (slower separation)

Then it seems like an algorithmic process where entropy produces opposite results given different initial conditions. Thousands, millions, mirriads of similar conditional branches take place every moment everywhere in the Universe. Thence everything can be described/modelled as causal chains of events. That's how our (natural) and artificial inteligence (rule based or case (experience) based reasoning) work. Just interesting parralels... 

I think the real problem is that there need not be an effectively computable number of branches. I have argued in several (as yet) unpublished papers that determinism is compatible with non-predictability and effective randomness (up to other constraints on the system). One place where I mention this (rather than argue for it) is in CAUSATION IS THE TRANSFER OF INFORMATION Howard Sankey (ed) Causation, Natural Laws and Explanation (Dordrecht: Kluwer, 1999). You can find a copy at http://www.nu.ac.za/undphil/collier/papers/causinf.pdf. Interestingly, I get this result even assuming that causation is like a deduction, not in spite of it. The connection between causation and logical inference is to be found in Rosen (1991) What is Life?, though his account of non-mechanical systems (in his version of 'mechanical', what I call non-predictable, non-reducible, or complexly organized systems) is defective in not giving an explanation of how they can occur. Howard Pattee claims that the idea originates with Hertz, which seems likely to me.

Some more remarks follow:

-------Original Message-------
 
From: Stanley N. Salthe
Date: 2004 m. balandis 10 d. 07:47:56
To: [email protected]
Subject: Re: [Fis] Szilard's Engine and Information
 
Shu-Kun said:
 
>Regarding the related entropy of mixing (Delta S), it is certain that
>the entropy of mixing is an information theoretical entropy
>because there is no heat involved. It should not be taken as a typical
>thermodynamic entropy (Delta S). Mixing of two chiral molecules
>gas R and gas L you mentioned cannot be a thermal process. Therefore,
>it is not a thermodynamic process in an heat engine. Mixing of R
>and L cannot be used to generate mechanical work. This is a fact.
>When we discuss the engine and related possibility of energy
>conservation, this fact must be kept in mind.
>
>If the mixing of gas R and gas L would create work (a kind of mechanical
>energy calculated as distance times force), one should be able to also create
>mechanical work by mixing red color and black color.
 
I find this interesting in regard to the understanding ofphysical
entropy as disorder. This interpretation has been disputed because of
examples like mixtures of oil and water, which seem to spontaneously
separate, making a more orderly result than was present in the mixed state.

The separation is only possible with dissipation. With no dissipation, the
molecules would just bounce back (i.e., the no dissipation condition is
reversible at the macro level).

But Shu-Kun's posting here suggests why this understanding is specious.
What is neglected in this view is that energy-utilizing work had to be done
to mix the oil and water to begin with, in a thermal process. This process
set up a curious kind of dispersed energy gradient, which then dissipated,
producing the separation, and giving off heat again. That is to say, the
unmixing of oil and water is a kind of work (not unlike the unwinding of
many wound up rubber bands!), and, as such, would not be expected to
produce disorder. Put another way, the unmixing of oil and water is NOT
SPONTANEOUS, but is instead a massive amount of microscopic work.

Just doing work is reversible, so this argument that the separation involves
work doesn't work (no pun intended). However, we can measure the work
done relative to the forces that have to be overcome. In this case, there
is a thermodynamic gradient promoting mixing. It must be overcome by
the forces of separation (gravity, surface tension, hydrophobia). Typically,
the end state in such cases is lower than required to overcome the
thermodynamic gradient of mixing (which, I note, is only one factor of
the overall thermodynamic gradient). The rest, can be used for work,
contrary, I think, to what Shu-Kun says. I agree with him, however, that
mixing is best understood as an information process (so is sorting).
Such processes need not involve heat (I can give you models of this,
from Charles Bennett, using work on reversible computation by
Toffoli, or you can look at my paper on Maxwell's demon on my
web site). Since heat transfer is just a mixing of higher and lower
kinetic energies, I see no good reason to distinguish thermodynamic
and "infodynamic" (I prefer "morphodynamic) processes at the
fundamental level, though it may be convenient in particular cases.

Cheers,

John

-------------------------------------------------------------------- Please find our disclaimer at http://www.ukzn.ac.za/disclaimer -------------------------------------------------------------------- <<<>>> Received on Sun Apr 18 04:18:07 2004