I've been reading Tuckerman's statistical mechanics textbook, in Feynman's path-integral chapter, the book introduced a method to convert from thermodynamic to virial energy estimators by using Euler's homogeneous function theorem:

$\alpha(x_1, \cdots, x_P) = \frac{1}{2} m \omega_P^2 \sum_{k=1}^{P} (x_k - x_{k+1})^2$ the function $\alpha$ is a homogeneous function of degree 2. Hence, applying Euler's theorem, which we can write as:

$\alpha(x_1, \cdots, x_P) = \frac{1}{2} \sum_{k=1}^{P}x_k \frac{\partial \alpha}{\partial x_k}$.

I know the general theorem: $n f(x_1, \cdots, x_P) = \sum_{k=1}^{P} x_k \frac{\partial f}{x_k}$

To prove this is correct, I first let $p=1$, then we arrive $\alpha = \frac{1}{2}m \omega_P^2 (x_1^2 - x_1 x_2 )$, which isn't same as the outcome of Euler's theorem? I'm quite confused about this part, thanks in advance for any help!

  • $\begingroup$ What do you mean "include"? $\frac12m\omega_P^2$ is a multiplicative constant — it does not involve any of the variables. It will be there, on both sides of your equation. $\endgroup$ Feb 11 '17 at 15:30
  • $\begingroup$ What about trying the dim 1 (i.e $p=1$ case), just to check ? $\endgroup$
    – Thomas
    Feb 11 '17 at 16:55
  • $\begingroup$ @Thomas, when $P=1$, $\alpha(x_1, \cdots, x_P) = \frac{1}{2} m \omega_P^2 (x_1^2 - x_1x_2)$ instead, it doesn't reveal its original expression, I think it's because of this $x_{k+1}$ term being troublesome $\endgroup$
    – Gvxfjørt
    Feb 11 '17 at 23:22

The confusion here is purely because I ignored the physical meaning of $P$ in Feynman's path-integrals. In which case, the resemblance of the partition function to that of a classical cyclic polymer chain of $P$ points to coin the term classical isomorphism and to exploit the isomorphism between the classical and approximate quantum partition functions. In regard to our current question, $P \ge 2$ must be devised in order to form a cyclic polymer, otherwise, we only construct a single particle (reveals classical system with single point).
When $P=2$, $\alpha = \frac{1}{2} m \omega_P^2 [(x_1 - x_2)^2 + (x_2 - x_3)^2] = \frac{1}{2} m \omega_P^2 [(x_1 - x_2)^2 + (x_2 - x_1)^2] = m \omega_P^2(x_1 - x_2)^2$, then the theorem can be easily proved.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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