Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose we have a quantum state, well described by its time-independent wave function $\Psi$. And we have a well-defined Hermitian (self-adjoint) operator $\hat{A}$. We successfully evaluate the expectation value of the operator $\hat{A}$. Next we derive the general formula for the higher moments of $\hat{A}$ (i.e. the expectation value of $\hat{A^n}$ for $n=2,3,4…$). Finally we scale the operator $\hat{A}$ appropriately, in order to make the result dimensionless and to remove a possible growth factor (of type $C^n$) in the moments. We obtain:

$$ <\hat{A^n}> = Cn + D $$

for $n=1,2,3,...$ and where $C$ and $D$ are constants.

Let us now define a new operator $\hat{B}$ as follows:

$$ \hat{B} = \hat{A^{n+1}} - \hat{A^n} $$

We can easily verify that the first and second moment of B are given by:

$$ < \hat{B} > = C $$

$$ < \hat{B^2} > = 0 $$

Therefore the variance of operator $\hat{B}$ is negative! In violation of statistical laws.

Should we conclude from this example that the results derived for the moments of $\hat{A}$ must be flawed? Or should we conclude that the new operator $\hat{B}$ is not a proper operator after all and therefore its strange properties are insignificant with respect to questions about the validity of $\hat{A}$?

$$\begin{align} <\hat{B}> &= <\hat{A^{n+1}}> - <\hat{A^n}> \\ &= C(n+1) + D - Cn - D = C \end{align}$$

$$\begin{align} <\hat{B^2}> &= <\hat{A^{2n+2}}> - 2<\hat{A^{2n+1}}> + <\hat{A^{2n}}> \\ &= C(2n+2) + D - 2C(2n+1) - 2D + C(2n) + D \\ &= 0 \end{align}$$

$$\begin{align} \text{Variance of }\hat{B} &= <\hat{B^2}> - \left(<\hat{B}>\right)^2 \\ &= -C^2 \end{align}$$

share|cite|improve this question
How are <B> and <B^2> calculated, can you add these? – Nikos M. Jun 13 '14 at 14:49
<B> = <A^(n+1)> - <A^n> = C*(n+1) + D - C*n - D = C. – M. Wind Jun 13 '14 at 15:09
<B^2> = <A^(2n+2)> - 2*<A^(2n+1)> + <A^(2n)> = C*(2n+2) + D - 2*C*(2n+1) - 2*D + C*(2n) + D = 0. – M. Wind Jun 13 '14 at 15:12
Thus: Variance of B = <B^2> - <B>^2 = -C^2 is negative. – M. Wind Jun 13 '14 at 15:16
i dont understand how this $<\left(\hat{A}\right)^n> = Cn + D$ holds in general? – Nikos M. Jun 13 '14 at 16:19

i think the answer to the question is that the relation of nth raw moment of the operator $A$ ($<\hat{A^n}>$)

i.e $<\hat{A^n}> = Cn + D$,

does not hold in general (not even for Gaussian random variables sometimes assumed as being simple). The rest follows from that.

Now another question is what would be the physical or mathematical meaning of a negative variance. This possibly could involve dealing directly in a complex space (e.g complex probability), however this is under discussion (or convention).

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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