I was working out the math of Gauge Invariance of Schrodinger Equation. Gauge transformations on my $\psi$, and the fields $\vec{A}$, and $A_0$ are \begin{eqnarray*} \psi(x)&\rightarrow& e^{i\alpha(x)} \psi(x)\\ \vec A &\rightarrow& \vec A + \frac{1}{q} \vec\nabla \alpha(x,t)\\ A_0 &\rightarrow& A_0 - \frac{1}{q} \partial_0 \alpha(x,t) \end{eqnarray*}

I need to show that \begin{eqnarray*} i \frac{\partial \psi'}{\partial t}&=& H' \psi'\\ \implies i \frac{\partial \psi}{\partial t}&=& H \psi \end{eqnarray*} where my Hamiltonian is given by \begin{eqnarray*} H=\frac{1}{2m}(p-qA)^2 + qA_0 \end{eqnarray*}

First I try the kinetic energy term, \begin{eqnarray*} (p-qA')^2 \psi' &=& \left(-i \vec\nabla -qA-\vec\nabla\alpha\right)^2 (e^{i\alpha} \psi)\\ &=&\left(i \vec\nabla +qA+\vec\nabla\alpha\right)\left(i \vec\nabla +qA+\vec\nabla\alpha\right) (e^{i\alpha} \psi)\\ &=& \left(i \vec\nabla +qA+\vec\nabla\alpha\right) \left[ ie^{i\alpha}\vec\nabla\psi +qAe^{i\alpha}\psi \right]\\ &=& %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% - \vec\nabla \left( e^{i\alpha}\vec\nabla\psi\right) +i \vec\nabla\left(qAe^{i\alpha}\psi \right) +\left(qA+\vec\nabla\alpha\right) \left[ ie^{i\alpha}\vec\nabla\psi +qAe^{i\alpha}\psi \right]\\ \end{eqnarray*} Now, I am worried about the second term in the last line above, $i \vec\nabla\left(qAe^{i\alpha}\psi\right)$. I assume that should be \begin{eqnarray*} i \vec\nabla\left(qAe^{i\alpha}\psi\right) = iq\left(\vec\nabla A\right)e^{i\alpha}\psi +iqA\left(\vec\nabla e^{i\alpha}\right)\psi +iqAe^{i\alpha}\left(\vec\nabla\psi\right) \end{eqnarray*} My problem isn't solved unless $\vec \nabla A$ is zero. Now, is it zero? What is it anyway? Divergence of $\vec A$ or gradient of $\vec A$? If it's divergence, then that is zero only in electrostatics. If it is a gradient, then it has some other tensor component.


The problem is also solved if $\vec\nabla A$ is non-zero. Your last line equals $$ \mathrm e^{\mathrm i\alpha} \left( -\Delta \psi + \mathrm i q\, (\vec\nabla \cdot \vec A)\, \psi + 2 \mathrm i q\, \vec A \cdot \vec\nabla \psi + (q\vec A)^2\, \psi\right) \;. $$

Note that this is the same as $$ \mathrm e^{\mathrm i\alpha}\, ( \vec p - q\vec A ) ( \vec p - q\vec A )\, \psi \;. $$

Edit to respond to comment.

You already did everything correctly in your question, $\vec p - q\vec A$ acts as follows on a wave function $\psi(x)$: $$ (( \vec p - q\vec A ) \psi)(x) = -\mathrm i\hbar\, \vec \nabla \psi(x) - q\vec A(x)\, \psi(x) $$ If you act again from the left with $\vec p - q\vec A$, you just take the inner product. Thinking about this in terms of gradients, divergences etc is possible but not too helpful, it's better to just think in components: $$ (( \vec p - q\vec A )^2 \psi)(x) = \sum_{i=1}^3 \bigl( -\mathrm i\hbar\, \partial_i - q A_i(x) \bigr) \bigl( -\mathrm i\hbar\, \partial_i \psi(x) - q A_i(x)\, \psi(x) \bigr) $$

  • $\begingroup$ thanks, understood. But the primary doubt is only clear, and not the second. can you please also state as to how we should treat this exactly? a gradient or a divergence? since $\vec p - q\vec A$ is itself a vector operator, how should it act on $\vec p - q\vec A$ which has already acted on $e^{i\alpha} \psi$? $\endgroup$
    – MycrofD
    Mar 24 '18 at 14:30
  • $\begingroup$ @MycrofD I edited my reply $\endgroup$
    – Noiralef
    Mar 25 '18 at 9:27

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.