A non-degenerate two-level system is described by a Hamiltonian $\hat H$ with $\hat H|n\rangle = \epsilon_n|n\rangle$, where $n = 1, 2$. An observable $\hat B$ has eigenvalues $\pm 1$, with the corresponding orthonormal eigenfunctions $|\pm\rangle$ related to $|n\rangle$ by

$|+\rangle = (|1\rangle + 2|2\rangle)/\sqrt{5}$

$|−\rangle = (2|1\rangle − |2\rangle )/\sqrt{5}$

What are the possible outcomes, together with their probabilities, of a measurement of $\hat B$ in a state of the system with the energy $\epsilon_2$? Would you get the same result if you made a second measurement of $\hat B$ some time after the first?

I think i can do the first part:

P(+) = |<2|+>|^2 = 4/5

P(-) = |<2|->|^2 = 1/5

But after making a measurement the wavefunction collapses into either one or the other so should the the probability be 1 and 0, how does this depend on time?

  • $\begingroup$ Does $\varepsilon$ depend on $n$? $\endgroup$ – probably_someone May 9 '18 at 13:50
  • $\begingroup$ Yes, should be ε subscript n, i.e ε1 doesnt equal ε2 $\endgroup$ – jjp1996 May 9 '18 at 13:53

If you measurement yields $+$, your state will be in the eigenstate of $\hat B$ with eigenvalue $+$. Thus, for instance, if you obtain $+$ at $t=0$ your system will collapse to $\vert +\rangle$. This becomes your new initial state for the evolution. The time-evolution is obtained by expanding $$ \vert\Psi(0)=\vert +\rangle = \vert 1\rangle \langle 1\vert +\rangle +\vert 2\rangle \langle 2\vert +\rangle $$ from which $$ \vert\Psi(t)\rangle= e^{-i\epsilon_1 t}\vert 1\rangle \langle 1\vert +\rangle +e^{-i\epsilon_2 t}\vert 2\rangle \langle 2\vert +\rangle $$ Since the time-evolution does not produce an eigenstate of $\hat B$, a second measurement of $\hat B$ would produce $+$ with probability $$ P(+,t)=\vert \langle +\vert\Psi(t)\rangle\vert^2 =\vert e^{-i\epsilon_1 t}\langle +\vert 1\rangle \langle 1\vert +\rangle +e^{-i\epsilon_2 t}\langle +\vert 2\rangle \langle 2\vert +\rangle\vert^2 \ne 1 $$ and likewise for $P(-,t)$.

| cite | improve this answer | |
  • $\begingroup$ so this will give: P(+,t) = |exp(-iE1t/h)/5 + 4exp(-iE2t/h)/5|^2, i take it then that the exp are not cacelled by the mod squared as this would just give a probability of 1. $\endgroup$ – jjp1996 May 9 '18 at 15:34
  • $\begingroup$ correct. $\vert A+B\vert^2 \ne \vert A\vert^2+ \vert B\vert^2$ so the exponentials will "interfere" and will produce a time-dependent probabilities. $\endgroup$ – ZeroTheHero May 9 '18 at 15:51

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.