We get the value of an observable $A$ for a given state $|\lambda\rangle$ of a system from the eigenequation $\hat{A} |\lambda\rangle = \lambda |\lambda\rangle$ where $\hat{A}$ is the operator corresponding to observable $A$ and $\lambda$ is the required value of the observable. But this is true only when the statevector $|\lambda\rangle$ is an eigenvector of the operator $\hat{A}$. How do we assign a value to observable $A$ when the statevector $|\lambda\rangle$ is not an eigenvector of the operator $\hat{A}$?

I am a beginner in quantum physics with elementary understanding of linear algebra so please try to answer accordingly.

PS: This question is related and its accepted answer is helpful. But from it I could not understand how to assign a value to the observable.

  • $\begingroup$ Thank you @Ghoster for the edit! $\endgroup$
    – t2m
    Sep 20, 2022 at 17:07

2 Answers 2


How do we assign a value to observable $A$ when the statevector $|\lambda\rangle$ is not an eigenvector of the operator $\hat{A}$?

We don't.

One of the toughest parts of learning quantum mechanics is learning to let go — in particular, learning to let go of the requirement that everything must have a value. In QM, if you have a system in a given state $|\psi⟩$ and a given observable $\hat A$ representing a physical quantity $A$, the generic setting is that the system cannot be assigned a value for $A$.

In some lucky occasions, of course, it turns out that $A$ does have a well-defined value (and these are, as you'll have guessed, when $|\psi⟩$ is an eigenstate of $\hat A$).

But, in general, observables in quantum mechanics do not have value until and unless a projective measurement of that observable is performed on the system.

  • $\begingroup$ According to answer of @Arturodonjuan, we can find average value of observable when the state vector is not eigenvector of operator. I assume that both answers spell truth with different perspectives. But then I can't reconcile both answers. Can you please explain on that line? $\endgroup$
    – t2m
    Sep 21, 2022 at 8:05
  • $\begingroup$ The 'expectation value' of $A$ in the state $|\psi⟩$ is what you get if you take the system in the state $|\psi⟩$ and then (i) perform a projective measurement of $A$, which will then (ii) change the state of the system to one of the eigenstates of $\hat A$ and produce the corresponding eigenvalue as the value of $A$, and then (iii) you repeat this procedure multiple times (typically on a large ensemble), which will then typically (iv) give a wide spread of different values of $A$ with different frequency; and then finally you (v) take the average of the value of $A$ over this distribution. $\endgroup$ Sep 21, 2022 at 16:57
  • $\begingroup$ However, this does not mean that the expectation value of $A$ is in any way assignable as "the value of $A$ prior to the measurement". That value does not exist, and cannot be assigned. $\endgroup$ Sep 21, 2022 at 16:57
  • $\begingroup$ I get your point that there is an 'expectation value' but don't get that it is not assignable as a value of $A$ prior to the measurement. I think, even the definite outcome given by the eigenvalues too is what we get after measurement. Isn't it? $\endgroup$
    – t2m
    Sep 21, 2022 at 17:13
  • 1
    $\begingroup$ Thank you for helping with insightful answers. $\endgroup$
    – t2m
    Sep 22, 2022 at 12:25

A pure physical state is comprised of a linear sum of eigenstates:

$$|\psi\rangle = \sum_{i} c_i |\lambda_i\rangle$$

where $|\lambda_i\rangle$ are complete eigenstates of a self-adjoint operator $A$, $\langle \lambda_i |\lambda_j\rangle = \delta_{ij}$ and $\sum_i |c_i|^2=1$.

The probability that we will find $|\psi \rangle$ in the $|\lambda_i\rangle$ state (i.e. having value $\lambda_i$ when acted upon by operator $A$) is given by $|\langle \lambda_i | \psi\rangle|^2=|c_i|^2$.

The average value of the operator $A$ in the state $|\psi\rangle$ is given by $\langle \psi | A |\psi \rangle = \sum_i |c_i|^2\lambda_i$. This is the average value of $A$ you would measure if you prepared an infinite number of identical systems described by the state-vector $|\psi\rangle$.

  • $\begingroup$ Is there a reason for saying a state as 'pure' physical state? $\endgroup$
    – t2m
    Sep 21, 2022 at 8:54
  • 1
    $\begingroup$ Yes because a physical state is in general described by as a 'mixed' state, represented by a density matrix/operator. A special case is when the physical state/system can be represented by a single state vector, i.e. it is a 'pure' state. See the wikipedia article here. $\endgroup$ Sep 21, 2022 at 17:02
  • $\begingroup$ Can you explain in the post, or point me to an easy to understand online resource, why the average value (expectation value) is as you have written? $\endgroup$
    – t2m
    Sep 22, 2022 at 9:39
  • 1
    $\begingroup$ @Aarone Do you mean why the expectation value of an observable $\hat A$ in a physical state $|\psi\rangle$ is given by $\langle \psi | \hat A |\psi\rangle$? This is sort of by definition, an axiom of the framework of quantum mechanics. With this, the relation $\langle \psi | \hat A |\psi\rangle = \sum_i |c_i|^2 \lambda_i$ follows by direct substitution of $|\psi\rangle = \sum_i c_i |\lambda_i\rangle$, and using that the set of eigenstates $|\lambda_i\rangle$ is orthonormal, i.e. $\langle \lambda_i |\lambda _j\rangle = \delta_{ij}$. $\endgroup$ Sep 22, 2022 at 18:06
  • $\begingroup$ I was asking about how $\langle \psi | \hat A |\psi\rangle$ is equal to $\sum_i |c_i|^2 \lambda_i$. Your hint is sufficient. Thank you. $\endgroup$
    – t2m
    Sep 23, 2022 at 2:08

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.