Maxwell's Demon ?

TallDave · Post by **TallDave** » Wed Jul 15, 2009 6:58 am

That's the point, you don't have to say "At precisely 9:02:01.738274983, entropy was X." It's enough for my purposes to say "over time t, there is probability Y of state Z." From a low-entropy state's nonzero probability I can deduce that entropy sometimes declines. I don't have to measure.

MSimon · Post by **MSimon** » Wed Jul 15, 2009 7:17 am

kcdodd wrote:How can you say how much entropy is in the box if you never measure the box.

You can't even say how much entropy is in the box if you measure it.

Since measurements changes what is measured it is impossible to know or measure anything.

I'll stick to hand waving. It doesn't require a thermometer.

</snark>

kcdodd · Post by **kcdodd** » Wed Jul 15, 2009 7:18 am

That's the point, you don't have to say "At precisely 9:02:01.738274983, entropy was X." It's enough for my purposes to say "over time t, there is probability Y of state Z." From a low-entropy state's nonzero probability I can deduce that entropy sometimes declines. I don't have to measure.

That is a very classical viewpoint. For better or worse.

TallDave · Post by **TallDave** » Wed Jul 15, 2009 2:05 pm

Ah, I see. You're viewing the box as full of quantum indeterminacy, which disappears when we open it and measure.

But even if that's a valid measure of entropy for atoms bouncing around in a box, it can't tell us anything about entropy intrinsic to the distribution in the box after we measure it, it can only tell us if we opened the box and measured. Different kinds of entropy, I guess.

What's interesting from the article they appear to have found a tiny system in which they can selectively use indeterminacy to violate Landauer's bound. But I have my doubts this could actually function in a way that defeats or reverses the heat-death of the Universe.