7
$\begingroup$

If I have two projectors $\pi_1, \pi_2$ such that for some $|{\phi}\rangle$:

$\langle {\phi}| \pi_1 |{\phi}\rangle \geq e$ and $\langle {\phi}| \pi_2 | {\phi}\rangle \geq e$

What can I conclude about the following quantity?

$\langle {\phi} | \pi_1 \pi_2 |{\phi}\rangle$

Is it also $\geq e$?

$\endgroup$
5
$\begingroup$

Based on those relations there's nothing more that you can conclude. Consider the two extremes.

At one extreme $\pi_1$ and $\pi_2$ project onto the same subspace, in which case: $$\langle {\phi} | \pi_1 \pi_2 |{\phi}\rangle = \langle {\phi} | \pi_1 |{\phi}\rangle = \langle {\phi} | \pi_2 |{\phi}\rangle \ge e, \;\; \pi_1=\pi_2=\pi_1 \pi_2.$$

At the other extreme $\pi_1$ and $\pi_2$ project onto perpendicular subspaces, in which case $$\langle {\phi} | \pi_1 \pi_2 |{\phi}\rangle = 0, \;\; \pi_1 \pi_2 = 0.$$

Based on the stated relationships, the projectors could exist anywhere between and including these two extremes. If $e>0$, you could at least say that $\vert \phi \rangle$ is not perpendicular to either the $\pi_1$ or $\pi_2$ subspaces, but that alone still doesn't tell you any more about the value of $\langle {\phi} | \pi_1 \pi_2 |{\phi}\rangle$.

EDIT: DaftWullie's answer makes an important point that I missed. If $e>\frac{1}{\sqrt{2}}$ (with $\vert \phi \rangle$ normalized), $\pi_1$ and $\pi_2$ cannot be orthogonal projectors, in which case $e$ imposes the lower bound $2e^2-1$.

$\endgroup$
  • $\begingroup$ Thanks for explaining. I asked a related question at: physics.stackexchange.com/questions/594565/… Would appreciate it if you can share your insights. $\endgroup$ – islamfaisal Nov 17 '20 at 23:07
  • $\begingroup$ @islamfaisal My pleasure. Looks like Chiral Anomaly provided a solid answer while I was thinking it through. That's similar to the example I had in mind. $\endgroup$ – Jonathan Trousdale Nov 18 '20 at 2:09
4
$\begingroup$

Let's assume $$ \pi_i|\phi\rangle=e_i|\phi\rangle+\sqrt{1-e_i^2}|\phi_i^\perp\rangle, $$ where $\langle\phi|\phi_i^\perp\rangle=0$ and, for simplicity, let's assume the $e_i$ are real. We can immediately expand $$ \langle\phi|\pi_1\pi_2|\phi\rangle=\left(f_1\langle\phi|+\sqrt{1-f_1^2}\langle\phi_1^\perp|\right)\left(f_2|\phi\rangle+\sqrt{1-f_2^2}|\phi_2^\perp\rangle\right) $$ This simplifies to $$ f_1f_2+\sqrt{1-f_1^2}\sqrt{1-f_2^2}\langle\phi_1^\perp|\phi_2^\perp\rangle, $$ which is bounded between $$ f_1f_2\pm\sqrt{1-f_1^2}\sqrt{1-f_2^2}. $$ Thus, we can say $$ \min_{f_1,f_2\geq e}f_1f_2-\sqrt{1-f_1^2}\sqrt{1-f_2^2}\leq\langle\phi|\pi_1\pi_2|\phi\rangle\leq\max_{f_1,f_2\geq e}f_1f_2+\sqrt{1-f_1^2}\sqrt{1-f_2^2}. $$ Clearly the right-had side of this bound is just 1. But it's the left side we're really interested in. To perform the minimisation, let $f_i=\cos\theta_i$. Then, we want $$ \min_{\theta_1,\theta_2}\cos(\theta_1+\theta_2). $$ If we assume that $\theta_1+\theta_2<\pi$, the minimum is achieved by setting both $\theta_1$ and $\theta_2$ as large as possible, corresponding with $f_i=e$. Thus, provided $e>0$, the minimum value is $2e^2-1$. $$ \langle\phi|\pi_1\pi_2|\phi\rangle\geq 2e^2-1 $$

$\endgroup$
2
$\begingroup$

$$ \pi_1 = \begin{pmatrix} 1&0\\ 0&0\\ \end{pmatrix}\\ \pi_2 = \begin{pmatrix} 0&0\\ 0&1\\ \end{pmatrix}\\ \rho = \frac{1}{\sqrt{2}} \begin{pmatrix} 1\\ 1\\ \end{pmatrix}\\ e = \frac{1}{2} $$

$\endgroup$
  • $\begingroup$ Thanks. Can you please elaborate on how one can generalize from this example? $\endgroup$ – islamfaisal Nov 17 '20 at 17:45
  • $\begingroup$ @islamfaisal If you calculate $\langle \phi | \pi_1 \pi_2 | \phi \rangle $ in this case, what would you get? This is an example to show the last statement in your question is not true. $\endgroup$ – KAJ226 Nov 17 '20 at 19:47
  • $\begingroup$ @KAJ226 Thanks. and are there other non-trivial counterexamples? $\endgroup$ – islamfaisal Nov 17 '20 at 20:57

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.