Physicists carry out Maxwell's "demon" thought experiment | Hacker News Reader

Physicists carry out Maxwell's "demon" thought experiment

scientificamerican.com

43 pointswaterhouse15 comments

This is cool. But it should be noted that it has already been proven that there is a minimum amount of entropy associated with a single irreversible bit flip, and as a result Maxwell's demon cannot actually violate the second law of thermodynamics.

That's the trouble with exciting phrases like "This seems to violate the second law of thermodynamics". It is known that Maxwell's demon does not violate anything, and the article does explain this later. But that's not quite the point.

I think this is a better explanation of the experiment: http://blog.joerg.heber.name/2010/11/14/the-demon-is-out-of-...

It's direct empirical support for the idea that "there is energy in information", and it demonstrates how even supposedly tried-and-trusted science is still subject to empirical validation.

This is the part I never understood about Maxwell's Demon:

> If the demon itself doesn’t use up any energy (which can be done), entropy would decrease, right?

Would anyone care to explain? Energy isn't spent moving the door around? (Particularly since the demon has to make sure the fast-moving molecules don't go right back through it... though I suppose with a sufficiently large box for fast-moving molecules, this may not happen...)

One nice way to think about is to imagine a very light but strong door, and very massive particles. The energy it would take to move the door back and forth would be trivial compared to the energy of the system.

You can even use a reversible process (e.g. the doors on a spring so the energy put into it to close the door is regained by letting the spring's energy reopen it) to make it use effectively zero energy.

Then all the "energy" in the system is in the demon's observation of the system, and not in the actual mechanics of opening and closing the door.

"Energy isn't spent moving the door around?"

Energy is a tricky thing, and you have to pay really close attention to the concept in physics class or you'll miss out on it, because it is not at all intuitive. As non-intuitive things go it's actually not that difficult to understand, but it doesn't match with our experience, because our bodies are set up in such a way that even doing nothing requires a certain amount of exertion, and our intuitive attempts to match "exertion" to "energy" produces very bad results.

How much energy does it cost to suspend an object one mile in the air, without it moving up or down? None. The problem is not one of energy, it's of an inability to counter the force of gravity. A hovering helicopter is indeed using a lot of energy, but 0% of the energy is doing work against gravity, it's all going other places.

How much energy does it take to keep an already-in-orbit object in orbit? None. Motion isn't energy. (It can't be, that would crack relativity wide open.)

How much energy is it to work against gravity and go uphill? X. How much energy is it to work against gravity and go back downhill to the same location? -X. Whats the net energy used against gravity? 0. Again, if we try to do this with our bodies, we are using lots of energy to do it, you can trace the chemical energy going out of the ATP and being used for things, but if you sum it all up after a round trip it is all going to things other than gravity; every joule of gravitational potential energy you gained going uphill was precisely returned to you on the return trip.

So yes, Maxwell's Demon can, in principle, use no energy manipulating the door. It'll need a bit to open and close it, but since we're in a thought experiment, whichever direction is increasing the potential energy of the system, it will get 100% of it back moving the other way if you set it up with a force to work against, like a 100% perfect spring. (Or it is balanced, moving across the set point, and since it's 100% efficient we never lose any.) Movement is not energy.

You are referring to Landauer’s principle with the "minimum amount of entropy associated with a single irreversible bit flip". While it is proven (or "trivial" if you believe some scientists on the subject), its applicability to all possible computation systems is not clear.

Recent theory and experiment have re-emerged to study this field both in the classical and quantum regimes. See http://arxiv.org/PS_cache/physics/pdf/0210/0210005v2.pdf for a nice review and "Breakdown of the Landauer bound for information erasure in the quantum regime" (http://pre.aps.org/abstract/PRE/v64/i5/e056117) for a specific case.

I wonder how much energy was used to manipulate the stuff around the bead, and how much energy would be used to just directly move it up.

"The bead is driven as a mini-rotor, with a information-to-energy conversion efficiency of 28%."

What exactly are the units of information in this case? The statement doesn't make much sense as I think the OP just took the quote out of context.

A simpler way to visualize what's going on: imagine your trained dog picks up a ball and tosses it up in your hands and you catch it. Now the ball is 1.5 meters higher off the ground than it was before. E = m * g * h = 1 kg (it's a heavy ball, all right?) * 9.8 m/s^2 * 1.5 m = 14.7 J. That's 14.7 J you got for free. But the dog expanded more energy than that to pick up the ball, overcome gravity and air resistance and toss it up. The total system (= you + ball + dog) lost energy and increased entropy.

I may be missing the point here... but this seems like a simple feedback system, just with a slightly more complex energy exchange (i.e. the experimenters observation and action) which hides the fact and allows it to be presented as a "paradox".

In actual fact this is just bad reporting, really. The experiment is about showing energy exists in information, which it seems to do successfully.

s/energy/entropy/ and then proof read.

Reply on news.ycombinator.com