3
$\begingroup$

I'm trying to prove that Holstein-Primakoff-Transformation \begin{align} S_i^{+} &= \sqrt{2S-n_i}b_i \\ S_i^{-} &= b_i^{\dagger} \sqrt{2S -n_i} \\ S_i^{z} &= S-n_i \end{align}

holds the condition $$ \vec S^2=S(S+1)$$ but I get the following (with $S_x=\frac{1}{2}( S^+ + S^-) $ und $S_y=\frac{1}{2i} (S^+-S^-)$): \begin{align*} S^{2} &= S_x^{2} + S_y^{2} + S_z^{2} = \frac{1}{4} \left( S^{+} + S^{-}\right)^{2} - \frac{1}{4} (S^{+}- S^{-})^{2} + (S-n_i)^{2} \\ &= \frac{1}{4} (2 S^{+} S^{-} + 2 S^{-} S^{+}) + S^{2} - 2Sn_i + n_i^{2} \\ &= \frac{1}{2} \left( \sqrt{2S-n_i} b_ib_i^{\dagger} \sqrt{2S-n_i} + b_i^{\dagger}(2S-n_i) b_i \right) + S^{2} - 2Sn_i + n_i^{2} \\ &= \frac{1}{2} \left( \sqrt{2S-n_i}(1+n_i) \sqrt{2S-n_i} + b_i^{\dagger}(2S-n_i) b_i \right) + S^{2} - 2Sn_i + n_i^{2} \\ &= \frac{1}{2} \left( n_i (2S-n_i) + 2S -n_i + b_i^{\dagger}(2S-n_i) b_i \right) + S^{2} - 2Sn_i + n_i^{2} \\ &= S^{2} + \frac{n_i^{2}}{2} + S - \frac{n_i}{2} - \frac{1}{2} b_i^{\dagger} b_i^{\dagger} b_i b_i \end{align*} So my question is, whether the Holstein-Primakoff satisfies this condition or not? - It should satisifies the condition, because otherwise it would be unphysically to use this transformation.

$\endgroup$
4
  • $\begingroup$ Related: physics.stackexchange.com/q/41485/2451 $\endgroup$
    – Qmechanic
    Commented Nov 16, 2017 at 19:50
  • $\begingroup$ Oh yes there is a big mistake. I try to fix it. $\endgroup$
    – Leviathan
    Commented Nov 16, 2017 at 20:23
  • 1
    $\begingroup$ @Leviathan you've been making edits every 15 seconds... why not tak5 15 minutes to work it out completely and without errors before posting? $\endgroup$ Commented Nov 16, 2017 at 20:30
  • 1
    $\begingroup$ I'm totally sorry. I thought I had calculated it right, but as I posted it I saw the first mistakes, and then it goes on an on... Next time I will handle this way. $\endgroup$
    – Leviathan
    Commented Nov 16, 2017 at 20:33

1 Answer 1

1
$\begingroup$

What you've done is treat $n_i$ as a number, which is not completely honest, since it really an operator $\hat n$. Thus for instance \begin{align} S^+ \vert n\rangle &= \sqrt{2S-\hat n}\, b\vert n\rangle = \sqrt{2S-\hat n} \sqrt{n}\vert n-1\rangle = \sqrt{(2S-n+1)n}\vert n-1\rangle\\ S^- \vert n\rangle &= b^\dagger \sqrt{2S-\hat n} \vert n\rangle = \sqrt{(2S-n)(n+1)} \vert n+1\rangle \end{align} so that \begin{align} S^+S^-\vert n\rangle &= \sqrt{(2S-n)(n+1)} S^+ \vert n+1\rangle =\sqrt{(2S-n)(n+1)} \sqrt{(2S-n)(n+1)} \vert n\rangle \\ &=(2S-n)(n+1) \vert n\rangle \\ S^-S^+\vert n\rangle &=(2S-n+1)n \vert n\rangle\, . \end{align} If my algebra is right this should take care of it.

$\endgroup$
5
  • $\begingroup$ I don't see where you get the +1 in the first term. $$S^+ \vert n\rangle = \sqrt{2S-\hat n}\, b\vert n\rangle = \sqrt{2S-\hat n} \sqrt{n}\vert n-1\rangle =\mathbf{ \sqrt{(2S-n+1)n}} \vert n-1\rangle $$ and secondly I don't think I ever treat the number operator ever as number in my calculation. I only changed the position of $n_i$ if and only if the term before is an polynomial of $n_i$, because that commutes with $n_i$. $\endgroup$
    – Leviathan
    Commented Nov 16, 2017 at 20:16
  • 1
    $\begingroup$ well.. $\hat n\vert n-1\rangle = (n-1)\vert n-1\rangle$ so $\sqrt{2S-\hat n}\vert n-1\rangle = \sqrt{2S-(n-1)}\vert n-1\rangle = \sqrt{2S-n+1}\vert n-1\rangle$ $\endgroup$ Commented Nov 16, 2017 at 20:32
  • $\begingroup$ @Leviathan maybe indeed you never treated $n$ as a number but you've edited your question so much that if you ever did this all traces of your incorrect manipulations of $n$ have by now disappeared. $\endgroup$ Commented Nov 16, 2017 at 20:34
  • $\begingroup$ I think there was one mistake with $b_i$ treated as number, not $n_i$ that I corrected. Next time should I not correct mistakes in my question? $\endgroup$
    – Leviathan
    Commented Nov 16, 2017 at 20:36
  • $\begingroup$ @Leviathan don't worry about it. No sense in trying to figure out what went wrong so many edits ago. $\endgroup$ Commented Nov 16, 2017 at 21:03

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.