• Security incident: ISF was recently accessed by intruders. Please change your password, and change it anywhere else you used it. Read more

What causes entropy?

rocketdodger

Philosopher
Joined
Jun 22, 2005
Messages
6,946
It is hard for me to understand the wikipedia entries regarding entropy and thermodynamics but it seems that the fundamental cause has something to do with probability.

Is this correct?

Could someone with more knowledge in the subject elaborate, in an easy to understand manner?
 
It is hard for me to understand the wikipedia entries regarding entropy and thermodynamics but it seems that the fundamental cause has something to do with probability.

Is this correct?

Yes, pretty much. Entropy is a measure of the disorder of a system, and there is a greater probability that any system will be in a disordered, rather than an ordered, system. For example, consider a tray completely covered by billiard balls, of which half are black and half are white. There are a few possible arrangements that show a high degree of order - for example, all the black balls on one side and the white balls on the other side of the tray - but very, very many that are random and chaotic. Shake the tray, let the balls settle, and look at the pattern, and, if it started with any discernible pattern, it will tend to be more chaotic afterwards than before. If it started chaotic, it's very unlikely that the balls will have rearranged themselves into an ordered pattern. The entropy of the chaotic pattern is higher, and statistically it's overwhelmingly likely that random changes will increase the entropy.

Dave
 
I really dislike anything that uses statistics and probability to define something because neither are actual measures, they represent accumulated bits of data and weigh the likely hood of something.
 
Last edited:
Yes, pretty much. Entropy is a measure of the disorder of a system, and there is a greater probability that any system will be in a disordered, rather than an ordered, system. For example, consider a tray completely covered by billiard balls, of which half are black and half are white. There are a few possible arrangements that show a high degree of order - for example, all the black balls on one side and the white balls on the other side of the tray - but very, very many that are random and chaotic. Shake the tray, let the balls settle, and look at the pattern, and, if it started with any discernible pattern, it will tend to be more chaotic afterwards than before. If it started chaotic, it's very unlikely that the balls will have rearranged themselves into an ordered pattern. The entropy of the chaotic pattern is higher, and statistically it's overwhelmingly likely that random changes will increase the entropy.

Dave

Is there a way to frame the energy of a system in terms of probability as well?
 
My (very limited) understanding as well is that any work creates entropy. So even when order is created in a local environment (such as a well constructed artifact or an unfolding leaf), a price in disorder is paid elsewhere and the net entropy in the universe still increases.
 
My (very limited) understanding as well is that any work creates entropy. So even when order is created in a local environment (such as a well constructed artifact or an unfolding leaf), a price in disorder is paid elsewhere and the net entropy in the universe still increases.

Exactly - the leaf unfolds and grows but it gets it's energy from the sun, which is one season closer to using up it's store of hydrogen.
 
It is hard for me to understand the wikipedia entries regarding entropy and thermodynamics but it seems that the fundamental cause has something to do with probability.

Is this correct?

Yes.

Here's a simple example (this is called the "Ehrenfest urn model"). Consider two containers and N coins, labelled 1 through N. N is a large number. Every instant you randomly pick a number between 1 and N, and move the corresponding coin from whichever container it's currently in to the other. The quantity of interest is the number of coins in each container, or more conveniently, the fraction f of the coins that are in container A at any given time (i.e. f={number of coins in A}/N).

What will happen to f as a function of time? It's probably clear that f has an equilibrium value, namely 1/2 (if that's not clear, I can explain). It may also be clear that any state where f differs significantly from 1/2 will not last long (i.e. not much longer than N instants), and that once f gets near 1/2. it will experience only small fluctuations around it. All of those statements are true if N is large (bear in mind that this is a simple model for a physical system where N is often of order 1023 or larger).

What does this have to do with entropy? Entropy is simply (the log of) the number of states that share some macroscopic property, like a value of f. For example, suppose N is even. Then there are N choose N/2 (that's a very large number) states with f=1/2. But there are only N states with f=1/N, and only 1 with f=0. That means that with very high probability, low entropy states (like f=1/N) will evolve into high entropy states (like f=1/2 or near it) rather than lower entropy states (f=0), simply because there are many more high entropy states (by definition).

What makes this a useful concept is that when N is very large, the number of states near f=1/2 is vastly, mind-bogglingly larger than the number of states near f=0. Therefore the chance of the entropy decreasing is absurdly small, so small it may never have happened in the history of the universe. As a result, the 2nd law of thermodynamics is called a "law", even though it is in fact "merely" probabilistic.

This is the reason drops of ink dissolve in water and never re-form, it's why you never see people leaping up out of swimming pools and landing on the diving board above an undisturbed water surface, it's the essence of life, and it explains why time has a direction. It's one of the simplest, most beautiful, and most profound ideas in physics.
 
It's the principle that wires are more likely to get tangled than untangled. Untangling requires intelligent intervention.
 
I really dislike anything that uses statistics and probability to define something because neither are actual measures, they represent accumulated bits of data and weigh the likely hood of something.

Could yu elaborate? This does not make sense to me. Staristics are measures.
 
What makes this a useful concept is that when N is very large, the number of states near f=1/2 is vastly, mind-bogglingly larger than the number of states near f=0. Therefore the chance of the entropy decreasing is absurdly small, so small it may never have happened in the history of the universe. As a result, the 2nd law of thermodynamics is called a "law", even though it is in fact "merely" probabilistic.

Based on this model, though, it seems like entropy should actually fluctuate very slightly and actually decrease almost as much as it increases when the current state of the system is near f = 1/2.

So when you say the chance of entropy decreasing is absurdly small, do you mean decrease at all, even to a state at f = 1/2.000000000000000000 .... 00001, or do you mean 'decrease to any appreciable extent'?
 
It is hard for me to understand the wikipedia entries regarding entropy and thermodynamics but it seems that the fundamental cause has something to do with probability.

Is this correct?

Could someone with more knowledge in the subject elaborate, in an easy to understand manner?

Arrange 100 coins so that they are all heads up.

There's exactly one way of doing this. HHHHH... etc.

Now arrange them so that 99 are heads up and one is tails.

You're free to choose which one is the tails. So there are 100 different arrangements to choose from.

Now arrange them so that 98 are heads up and two are tails.

Again you're free to choose which on of the remaining 99 heads to flip to tails. Heck you can even move the existing tails too. So there's 9,900 different arrangements to choose from.

Entropy is a concept which describes this level of freedom. The highest entropy level is where 50 are heads and 50 are tails.

The tendency of entropy to increase is statistical.

Imagine now that one coin is picked at random every second and flipped. It might be one of the 2 coins which have flipped to tails already in which case the entropy of the system will decrease but the chances are that it will be one of the 98 remaining heads in which case entropy increases.

Over time the system will tend towards it's maximum entropy state of 50 heads and 50 tails.
 
I really dislike anything that uses statistics and probability to define something because neither are actual measures, they represent accumulated bits of data and weigh the likely hood of something.

You do understand that deep down, everything is a statistical measurement? If not for any other reason, for quantum uncertainty?

That explains a lot...
 
Exactly - the leaf unfolds and grows but it gets it's energy from the sun, which is one season closer to using up it's store of hydrogen.

OT, I know, but I never fail to be amazed when I think of the fact that four million tons of hydrogen vanish every second in our Sun, converted into energy.
 
It also needs to be considered that order and disorder are subjective, they don't have objective definitions.

Or am I wrong?
 
Based on this model, though, it seems like entropy should actually fluctuate very slightly and actually decrease almost as much as it increases when the current state of the system is near f = 1/2.

Yes, that's correct.

So when you say the chance of entropy decreasing is absurdly small, do you mean decrease at all, even to a state at f = 1/2.000000000000000000 .... 00001, or do you mean 'decrease to any appreciable extent'?

That depends. If you start away from equilibrium, any decrease is extremely unlikely (i.e. you will tend towards equilibrium, which one can define by the state with maximum entropy). If you start at equilibrium a decrease can be likely, but its magnitude is tiny.
 
Can you elaborate on that? I don't understand how causality could be explained by entropy...

Go back to the urn example. At a small scale, any movement from a coin to an urn is reasonable to see. If the system were at equilibrium and you saw a chart (or a movie) of the coins as they move around, you couldn't tell which line was the initial state and which line was the final state. You could run the thing backward and it would appear very similar.

But if there is a low-entropy state (all the coins in one urn), then it's obvious that it's the initial state and the system proceeds to the high-entropy state. It is unreasonable to see the system produce that state randomly (and as the number of states increase, it becomes *far* less likely). You can now see which direction things ran where you could not beforehand.
 
My simple mind has always thought of entropy as a system trying to reach an equilibrium.
 

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom