# Reif statistical and thermal physics

• orthovector
Indeed, but in case of statistical mechanics it is just so unphysical to consider systems that are exactly described by classical mechanics that you don't want to define your fundamental concepts based on that.In summary, the reif textbook is an advanced level book that is good for thermal physics. There are newer editions, but the 1965 edition is good. It is part of the Berkeley Physics Course series.f

#### orthovector

I've recently bought the 1965 copy of the reif textbook by mcgraw hill, fundamentals of statistical and thermal physics.

It is considered an advanced undergrad/beginning grad student book. Zemansky and Dittmann is somewhat a lower level text, but both are good.

I have a F. Reif Statistical Physics text in my book shelf. It's part of the Berkeley Physics Course series; it's volume 5. I believe I used it in an undergradute course.

There are two books by Reif on stat mech. The Berkeley on and the McGraw-Hill one. This causes all sorts of confusion.

I've recently bought the 1965 copy of the reif textbook by mcgraw hill, fundamentals of statistical and thermal physics.

This is one of the best book on the subject that exists. It is one of the few books that teaches thermal physics in the correct way. Other books make deliberate errors for the sake of simplifying things. Take e.g. the definition of the entropy. Entropy arises by a coarse graining procedure. You have to define some small but macroscopic energy resolution and count the number of energy eigenstates that lies in that small energy interval. This is the only correct definition of entropy.

Many books completely bypass this and simply pretend that energy levels are exactly degenerate and define a multiplicity function. They then consider systems of harmonic oscillators and let the students do problems in which you have to compute the degeneracy. They then quickly move on the the canonical ensemble.

I doubt that many students who take a first course in thermodynamics know that if you were to exactly specify the internal energy of a closed system, the entropy would be exactly zero. So, while they have learned a few formal techniques, they have missed out on a very important part of thermal physics.

I doubt that many students who take a first course in thermodynamics know that if you were to exactly specify the internal energy of a closed system, the entropy would be exactly zero

How is that? More than one microstates correspond to a given energy in any example I can think of.

Last edited:
How is that?

You would have specified the exact energy eigenstate the system is in. So, if the system is not exactly degenerate, you have specified the exact microstate the system is in. In a realistic system, any symmetries that lead to degeneracies of the exited states will be broken.

This is why I really dislike books like the one by Kittel where they consider systems which are exactly degenerate to define the entropy via a multiplicity function. They then bypass the fact that the entropy is fundamentally a macroscopic property of a system that is obtained by a coarse graining procedure. In the thermodynamic limit the entropy becomes independent of the energy resolution over which you do the coarse graining.

You would have specified the exact energy eigenstate the system is in. So, if the system is not exactly degenerate, you have specified the exact microstate the system is in. In a realistic system, any symmetries that lead to degeneracies of the exited states will be broken.

But we never consider macroscopic bodies to be in stationary states. Also, your statement is false in classical statistics.

But we never consider macroscopic bodies to be in stationary states. Also, your statement is false in classical statistics.

Yes, but that means that we never specify the energy to infinite accuracy in the first place. And "classical statistical mechanics" is unphysical. This is why Reif starts with quantum systems right from the start. And his Omega function has a finite energy resolution build in right from the start. He then writes quite a lot about it so that you gain deep understanding of what entropy really is.

In books like the one by Kittel much less time is spent on explaining the fundamentals of real systems. Simple unrealistic models are used to define the entropy.

And "classical statistical mechanics" is unphysical.

No less physical than classical mechanics, classical electrodynamics and general relativity.

No less physical than classical mechanics, classical electrodynamics and general relativity.

Indeed, but in case of statistical mechanics it is just so unphysical to consider systems that are exactly described by classical mechanics that you don't want to define your fundamental concepts based on that.

I'm not sure which concepts you're talking about. Statistical mechanics and it's concepts (microstates, entropy, ensembles ...) are logically independent of whether you're applying it to the classical or quantum case. In fact, it is best to think of it as a special type of probability theory.

Last edited:
Ok, you can build up the theory in a more or less axiomatic way also for classical statistical mechanics. But then, you are automatically forced to distinguish between the fine grained and coarse graind entropy.

My point is that all this is more easy to explain intuitively in a way that is suitable in a first course in thermal physics in the approach followed in Reif's book. Kittel avoids these issues by using misleading examples which causes students to not fully understand what entropy is.

i am mesmerized by this conversation...hope I can do the same in a couple of months

Take e.g. the definition of the entropy. Entropy arises by a coarse graining procedure. You have to define some small but macroscopic energy resolution and count the number of energy eigenstates that lies in that small energy interval. This is the only correct definition of entropy.

Many books completely bypass this and simply pretend that energy levels are exactly degenerate and define a multiplicity function. They then consider systems of harmonic oscillators and let the students do problems in which you have to compute the degeneracy. They then quickly move on the the canonical ensemble.

I doubt that many students who take a first course in thermodynamics know that if you were to exactly specify the internal energy of a closed system, the entropy would be exactly zero. So, while they have learned a few formal techniques, they have missed out on a very important part of thermal physics.

dude...you are awesome!