My professor of statistical mechanics stated briefly that the probability density of the "microcanonical ensemble" is given by the Dirac delta function $\delta (E-E_0)$ where $E_0$ is the total energy of the given system.
I don't understand why. Can anyone elaborate on this point please?

P.S. I can see that the normalization condition is satisfied due to the identity $$\int\limits_{-\infty}^{\infty}\delta (E-E_0)dE=1$$

  • $\begingroup$ Some context would help. What was the topic of discussion when it came up? What (if any) limits had been placed on the situation? $\endgroup$ Aug 12, 2017 at 2:37
  • 3
    $\begingroup$ In the microcanonical ensemble, the energy of the system has a fixed definite value. $\endgroup$
    – garyp
    Aug 12, 2017 at 3:07
  • 1
    $\begingroup$ As garyp said the mean energy is fixed, it's because the temperature of the system is fixed from the outside. $\endgroup$
    – Ismasou
    Aug 12, 2017 at 7:25

1 Answer 1


Your professor was probably looking at the microcanonical distribution from the standpoint of the canonical distribution.

Consider the following expectation value of a variable A in the canonical distribution: $$\langle A(\beta,X)\rangle = \frac{\int A(E,X)\Omega(E,X)exp(-\beta E)dE}{\int \Omega(E,X)exp(-\beta E)dE}.$$ The denominator is the canonical partition function $Z(\beta,X)$, which is the $E \to \beta$ Laplace transform of the microcanonical distribution $\Omega(E,X)$. The X variable represents extensive variables other than E.

The expectation value can also be written as an integral with respect to a probability density function $p(\beta,E,X)$ as follows: $$\langle A(\beta,X)\rangle = \int A(E,X)p(\beta,E,X)dE,$$where, from the first equation, p is given by:$$ p(\beta,E,X) = \frac{\Omega(E,X)exp(-\beta E)}{\int \Omega(E,X)exp(-\beta E)dE}.$$ Now introduce the delta function $\delta(E-E_o)$ into the integrals. This selects for a particular energy $E_o$. As a result, the weights $\Omega(E_o,X)exp(-\beta E_o)$ in the first equation cancel out.

In particular, the reduced canonical pdf can simply be written $ p(\beta,E,X) = \delta(E-E_o)$, which, as asserted by your professor, leads to the (fixed) value of A in the microcanonical distribution:$$A(E_o,X)=\int A(E,X)\delta(E-E_o)dE.$$


Your Answer

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

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