Eigenstate is always related to at least one specific operator (can be more than one)
But I don't think it is a property of the entire system, perhaps you can say something else if you want to give such a state a more global characterization
@ACuriousMind One more question regarding the Gibbs paradox. Is it correct to say that by considering the indistinguishability of the particles, entropy becomes an extensive quantity? (because we add that 1/N! factor)
The paradox happens, from what I understand, because Gibbs entropy (or whatever he was using to measure it) wasn't an extensive quantity
like, Hamiltonian mechanics and phase space densities and all that make sense to me but I don't think e.g. my understanding of entropy matches up with how it's usually taught
@Amit every eigenstate of what
any eigenstate is a pure state because the definition of "eigenstate" is just that it's some vector with special properties w.r.t. some observable
and a "pure state" is just a vector
mixed states are density matrices that don't correspond to single vectors
@ACuriousMind Oh, okay, so probably even the other direction is correct to some extent, e.g. there exists some observable for which a pure state is an eigenstate... but never mind really, I only wanted to supply imbAF with a "global" characterization of an eigenstate (since pure does relate to the system, as opposed to an eigenstate which relates to observable/s).
(there's another technical issue here where you should probably think of Hilbert spaces at different temperatures as being different Hilbert spaces in the $C^\ast$-algebra formalism but let's not get into that as I don't really understand that either)
It's just that I myself got confused about this pure vs. eigen business once, I was only trying to reconstruct the resolution of the confusion ('cause I suspect that's what makes people tie eigenstates to "system" wrongly...)
No stat mech system is in a pure state, it's all density matrices, but density matrices are constructed in terms of 'pure states', so where the hell are the pure states coming from, e.g. where are the pure states in the Boltzmann distribution (and how does this relate to the Gibbs paradox)
Yes, in classical stat mech, Hamiltonian mechanics is always valid, but so what, we can't solve $\approx 10^{23}$ equations and impose the initial conditions, so we need an alternative which is stat mech
Liouville's theorem is just $\partial_t \rho = \{ H, \rho\}$
that's essentially just how all functions on phase space evolve in Hamiltonian mechanics
I understand that many courses don't teach it that way because for inscrutable reasons they never actually bother teaching proper Hamiltonian mechanics before using it, but that's just how it is
Oddly enough, we also didn't cover it in Analytical mechanics, first time I saw it was in a Susskind lecture :) and he is terrifically rigorous as we all know lol. But it was still great, totally understandable (probably even rigorously it's quite simple)
everything we did, in statistical mechanics, revolved around an isolated system, to which we associated the MCE. And since the Liouville theorem came up, when we were discussing isolated systems to which we associate the MCE, I thought that the Liouville theorem is ONLY valid or relevant for systems with conserved energy, and wouldn't be a thing for system where energy is not conserved
Liouville is basically the basis for statistical mechanics, it says the distribution function for an isolated system only depends on conserved quantities like energy, momentum, particle number etc... you can then study subsystems of this which interact with one another and frame things in terms of the distribution function (i.e. apply it to open subsystems of a larger closed system)
I talk about Liouville's theorem and why I think it's not really a theorem if you build Hamiltonian mechanics properly in this answer and the general notion of evolving functions on phase space along flow in this answer
Does anyone know anything about how "skin width" is used in character controllers?
My custom game server has voxelized regions within my terrain. When inside a region, my characters are getting stuck on things often. So I wanted to implement something like what Unity calls "skin width" where your collider can partially penetrate the obstruction. From their description, I can't figure out exactly how to implement it. (FYI: the reason the server is doing this is my game is extremely server authoritative, simulating what the client is doing)
The Unity client will slide along edges with ease with its character controller. I want my server's simulation to do the same, only given 0.1m x 0.1m voxels that say if a location is passable or not.