We have Gibbs Entropy which is defined as $S_G(U,V,N)=k_B \ln \Omega_0$, where $\Omega_0$ is the number of microscopic states with energy equal or smaller than $U$. Note that this definition differs from the definition of Boltzmann Entropy, $S_B(U,V,\delta U, N) = k_B \ln \Omega$, where $\Omega = (\partial \Omega_0 / \partial U)\delta U$. Consider the different definitions of tempratures: $$T_B = (\partial S_B / \partial U)^{-1}$$ $$T_G = (\partial S_G / \partial U)^{-1}$$

I want to show that: $$T_B/T_G = 1/(1+k_B/C)$$ where $C = (\partial T_G /\partial E)^{-1}$ is the total heat capacity associated with $T_G$.

I am not sure how to show it, after some algebraic manipulations I got to: $$T_B/T_G = \frac{\Omega / \Omega_0}{(\partial \Omega \partial U) / (\partial \Omega_0 / \partial U)} = $$ $$ = \frac{\delta U (\partial \Omega_0 / \partial U)/\Omega_0}{\partial^2 \Omega_0 /\partial U^2\delta U / (\partial \Omega_0 /\partial U)} = $$ $$ = \frac{(\partial \Omega_0 / \partial U)^2 1/ \Omega_0}{\partial^2 \Omega_0 / \partial U^2}$$

And $$C^{-1} = \frac{-1}{(\partial S_G / \partial U)^2} \frac{\partial^2 S_G}{\partial U \partial E}$$

I am stuck, does someone have a detailed reference where this is shown or knows how to show it?

  • $\begingroup$ @CountTo10 how is exact duplicate? they don't show the relation I asked to prove. $\endgroup$ Commented Nov 29, 2016 at 17:03
  • $\begingroup$ physics.stackexchange.com/questions/141321/… I had this as an exact duplicate but it is not, my apologies for that mistake. $\endgroup$
    – user108787
    Commented Nov 29, 2016 at 17:03

1 Answer 1


Okay, let primes be partial derivatives with respect to $U$ and $\sigma = S_B$ and $S = S_G$ and let me redefine $\Omega$ as $\Omega_0$ so that I don't have to keep writing these subscript-zeros on every $\Omega_0$ (in other words I will not refer to your $\Omega$ explicitly at all in this answer). Then you have the corresponding temperature definitions $$\begin{array}{rcl} \tau^{-1} &= \sigma' =& k ~\Omega'' /\Omega',\\ T^{-1} &= S' =& k ~\Omega'/\Omega. \end{array} $$ You have defined $$C^{-1} = T' = (1/S')' = \frac{-S''}{(S')^2} = -\frac{k \big[\Omega''/\Omega ~-~(\Omega')^2/\Omega^2\big]}{k^2 (\Omega')^2 / \Omega^2}.$$ Seems pretty straightforward to me, though I can see why you might have been put off by the complexity of the symbols, especially if you failed to unite conceptually $E$ and $U$ in your expressions. After cancellation above you should get:$$k/C = 1 - \frac{\Omega'' ~\Omega}{(\Omega')^2} = 1 - \frac{T}{\tau}.$$The only variation I'm seeing from your expression is that I get a $\tau/T = 1/(1 - k/C)$ expression, and I don't see a way to fix that difference in our signs; that minus sign seems morally destined to be there unless I'm missing something like $E$ is some energy flow into a system and therefore $E = - U$ and $C^{-1} = - T'$ or so.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.