2

What does $⟨ϕ|Q|ψ⟩$ physically mean, where $|ψ⟩$ and $|ϕ⟩$ are states and $Q$ is a linear operator? I know what its mathematical meaning is but I am looking for an interpretation: What does $Q$ do to $|ψ⟩$ and then the result to $⟨ϕ|$?

More specifically, what happens physically in the Stern-Gerlach experiment when you have the following expression: $⟨S_z;+|S_x|S_z;+⟩$ , where $S_x$ is the operator of the measurement including an inhomogeneous magnetic field directed along the $x$-axis and $|S_z;+⟩$ is the spin-state of the particle along the $z$-axis.

Qmechanic
  • 201,751

2 Answers2

3

In brief: in general, none.

There might be operators $O$ for which $\langle\phi|O|\psi\rangle$ has some physical meaning. These are generally unitary operators (which preserve the normalization of $|\psi\rangle$, so that $O|\psi\rangle$ is still a physical, normalized state if $|\psi\rangle$ is). For instance, if $O=U(t)$, where $U$ is the time evolution operator, then $\langle\phi|U(t)|\psi\rangle$ is the amplitude for the process $\psi\to \phi$ to happen after a time interval $t$. If $O=T_{a}$, where $T$ is the spatial translation operator, then $\langle\phi|T_{a}|\psi\rangle$ is the overlap between the state $|\psi\rangle$ translated of an amount $\Delta x=a$ and the state $|\phi\rangle$. For rotations it works similarly, etc.

Moreover, $\langle\phi|O|\psi\rangle$ might have a direct (albeit approximate) interpretation in contexts such as the time-dependent perturbation theory, where in the case of a time-independent perturbation, $H=H_{0}+O$, $\langle\phi|O|\psi\rangle$ is proportional to the amplitude for going from state $\psi$ to state $\phi$ in the presence of the perturbation (and to first order in the perturbative series).

As for your example, $\langle S_{z};+|S_{x}|S_{z};+\rangle$ is just the mean value of $S_{x}$ in the up state of $S_{z}$, meaning that if you measured $S_{x}$ in the state $|S_{z};+\rangle$ in an infinite amount of different experiments and then averaged the outcomes you would get $\langle S_{z};+|S_{x}|S_{z};+\rangle$. This is because, for any hermitian operator $Q$,

$$ \langle\psi|Q|\psi\rangle=\sum_{q}\ q\ |\langle q|\psi\rangle|^{2} $$

where $|q\rangle$ is the eigenstate of $Q$ with eigenvalue $q$ (notice that in the above $\phi=\psi$), and $|\langle q|\psi\rangle|^{2}$ does have a physical interpretation: it is the probability of finding $\psi$ in the state $|q\rangle$, hence the probability for the outcome of the measurement to be $q$.

Note: of course, not having a direct physical meaning does not mean that those matrix elements do not impact the physics of the system. Their impact, however, depends on how they enter into the equations.

BioPhysicist
  • 56,248
  • Thank you so much! But what I still do not understand is that if my particle is in a |Sz;+⟩ state and then passes through the Sx-apparatus then it has to be either |Sx;+⟩ or |Sx;-⟩ right? So how can you expect an |Sz;+⟩ outcome? Because isn't ⟨Sz;+| the outcome? – Astronaut Marbini Feb 20 '20 at 11:07
  • 1
    Thank you @AaronStevens! I tried the \bra \ket commands but StackExchange does not support them and I didn't think of \rangle and \langle. – Giorgio Comitini Feb 20 '20 at 11:14
  • @AstronautMarbini $\langle Sz;\pm|Sx;\pm\rangle \ne 0$. So there is an overlap between the two states. This is evident from the fact that the Hilbert Space is 2 dimensional (so 2 linearly independent vectors form a basis). – Superfast Jellyfish Feb 20 '20 at 11:17
  • @AstronautMarbini: the problem is that you're still thinking in terms of amplitudes, and as I explained above you shouldn't do that (unless you're in the context of time-dependent perturbation theory, which you are not). No, $\langle S_{z};+|$ is not the outcome of the experiment. When you are computing $\langle S_{z};+| S_{x} |S_{z};+\rangle$ what you're asking for is not a transition amplitude, it is the mean value of $S_{x}$ in the definite state ($|S_{z};+\rangle$). You must regard $|S_{z};+\rangle$ as the initial state in which you are making the measurement. – Giorgio Comitini Feb 20 '20 at 11:21
  • 1
    If you want to know the transition amplitude, as @FellowTraveller pointed out, you must compute it without the $S_{x}$ operator in between, as $\langle S_{x};\pm|S_{z};+\rangle$. The information about the direction of definition of the spin in the final state is encoded in the state itself ($\langle S_{x};\pm|$ rather than $\langle S_{y};\pm|$ or $\langle S_{z};\pm|$) and does not require an operator to be inserted in between the initial and final state. – Giorgio Comitini Feb 20 '20 at 11:21
  • Thank you, Mr. Comitini! What about $\langle S_{x};+|S_{x}|S_{z;+}\rangle$ ? What is the physical interpretation of that? – Astronaut Marbini Mar 04 '20 at 17:36
  • @AstronautMarbini I believe I had already answered to that (third paragraph above). Do you need any clarification? – Giorgio Comitini Mar 05 '20 at 16:07
  • My question referred to another bra this time. – Astronaut Marbini Mar 12 '20 at 09:48
  • I am sorry! I saw the ket and operator, but not the bra. As explained in the answer, there is no general physical interpretation to that number, since the bra and the ket are different. As $S_{x}|S_{x};+\rangle=\hbar/2,|S_{x};+\rangle$, that number is equal to $\hbar/2$ times the overlap between the $|S_{x};+\rangle$ and $|S_{z};+\rangle$ states. However, in my opinion, this does not count as a "physical interpretation". – Giorgio Comitini Mar 12 '20 at 17:35
0

$\langle \phi|Q|\psi\rangle$ is a mathematical statement that gives the inner (dot) product between the states $|\phi\rangle$ and $|Q\psi\rangle$. Here $Q$ is a linear operator that takes $|\psi\rangle$ to some other state $|Q\psi\rangle$.

What it means is what inner products usually mean. It is a measure of overlap between the two states $|\phi\rangle$ and $|Q\psi\rangle$. How much of $|Q\psi\rangle$ is in $|\phi\rangle$, loosely speaking.

Now for the physical interpretation of this we need to make use of completeness and the Born postulate. Say $|q_i\rangle$s are the eigenstates of the operator $Q$ with eigenvalues $q_i$, then we can rewrite: $$|\psi\rangle=\sum_i |q_i\rangle\langle q_i|\psi\rangle=\sum_i \alpha_i|q_i\rangle$$ where $\alpha_i$s are the overlap between $|q_i\rangle$ and $|\psi\rangle$. Using this in our expression we get: $$\langle \phi|Q|\psi\rangle=\langle\phi|Q \sum_i \alpha_i|q_i\rangle\\ =\langle\phi|\sum_i \alpha_iQ|q_i\rangle\\ =\langle\phi|\sum_i \alpha_iq_i|q_i\rangle $$

This here as of now doesn’t really make much sense. But consider the case when $|\phi\rangle$ is an eigenstate $|q_j\rangle$ of $Q$, then our expression simplifies to: $$\langle q_j|Q|\psi\rangle=\langle q_j|\sum_i \alpha_iq_i|q_i\rangle\\ =\sum_i\alpha_iq_i\langle q_j|q_i\rangle=\sum_i\alpha_i q_i\delta_{i,j}=\alpha_j q_j $$ This physically means that when we measure $Q$ on a state $|\psi\rangle$, we get the outcome $q_i$ with a probability of $|\alpha_i|^2$, assuming the states are normalised.

Now consider the general case where $|\phi\rangle=\sum_n \beta_n|q_n\rangle$. Now the inner product looks like: $$ \langle \phi|Q|\psi\rangle=\sum_{i,n}\langle q_n|\beta^*_n \alpha_iq_i|q_i\rangle\\ =\sum_{i,n}\beta^*_n\alpha_iq_i\langle q_n|q_i\rangle=\sum_{i,n}\beta^*_n\alpha_i q_i\delta_{i,n}=\sum_n\beta^*_n \alpha_n q_n $$

You can easily extend this argument to your system at hand.

  • Your content is correct, but your second part seems off. Usually you just look at $\langle q_j|\psi\rangle=\alpha_j$ in order to say the probability of measuring $q_j$ is $|\alpha_j|^2$. You don't need to bring in $Q$ explicitly, and this specific example doesn't really help illuminate what $\langle\phi|Q\psi\rangle$ means in my opinion. – BioPhysicist Feb 19 '20 at 01:14
  • Yup. Added a section to explicitly talk about this inner product. – Superfast Jellyfish Feb 19 '20 at 03:45
  • In your second to last statement, are you talking about taking two measurements? – BioPhysicist Feb 19 '20 at 03:56
  • Yes. And the way to observe it is by looking at the average value of $q$ over repeated measurements. – Superfast Jellyfish Feb 19 '20 at 04:10
  • Oh Ok. Your wording confused me because it makes it seem like you are saying the state could be $|\phi\rangle$ after measuring $Q$. You instead mean that you measure $Q$ and then whatever observable $|\phi\rangle$ is an eigenstate of? – BioPhysicist Feb 19 '20 at 04:16
  • Precisely. If not an eigenstate then we need to repeat measurements to converge to the coefficients in some basis (that we are measuring). – Superfast Jellyfish Feb 19 '20 at 04:50
  • As usual, I would like to know the reason for the downvote as it might help me better my understanding. – Superfast Jellyfish Feb 19 '20 at 10:58
  • 2
    I downvoted because your statement "If we have an initial state $|\psi>$ and we measure $Q$ what’s the probability (amplitude) of finding it in the state $|\phi>$" is simply false. That would be $<\phi|P_{q}|\psi>$, where $P_{q}$ is the operator that projects the states onto the eigen-subspace of $Q$ with eigenvalue $q$, provided that the measurement has resulted in the observation of the value $q$. Which is exactly what @AaronStevens pointed out, only in terms of probabilities rather than amplitudes. The wording does not make the statement true. – Giorgio Comitini Feb 19 '20 at 12:41
  • Ah okay. I get why the way I have written it is misleading. I meant to say that at the end we are projecting onto the $\phi$ state and the amplitude for the same is given by the matrix element as pointed out in my response to @AaronStevens comment. I shall remove the statement so as to not misinform. Thank you for the feedback. – Superfast Jellyfish Feb 19 '20 at 13:35
  • Yes, now it makes much more sense. – Giorgio Comitini Feb 20 '20 at 11:12
  • But still thank you all so much! – Astronaut Marbini Feb 20 '20 at 11:13