I'm working my way through Griffith's Introduction to Electrodynamics. In Ch. 10, gauge transformations are introduced. The author shows that, given any magnetic potential $\textbf{A}_0$ and electric potentials $V_0$, we can create a new set of equivalent magnetic and electric potentials given by:

$$ \textbf{A} = \textbf{A}_0 + \nabla\lambda \\ V = V_0 - \frac{\partial \lambda}{\partial t}. $$

These transformations are defined as a "gauge transformation". The author then introduces two of these transformations, the Coulomb and Lorenz gauge, defined respectively as:

$$ \nabla \cdot \textbf{A} = 0 \\ \nabla \cdot \textbf{A}= -\mu_0\epsilon_0\frac{\partial V}{\partial t}. $$

This is where I am confused. I do not understand how picking the divergence of $\textbf{A}$ to be either of these two values actually constitutes a gauge transformation, as in it meets the conditions of the top two equations. How do we know that such a $\lambda$ even exists for setting the divergence of $\textbf{A}$ to either of these values. Can someone convince me that such a function exists for either transformation, or somehow show me that these transformations are indeed "gauge transformations" as they are defined above.


Comment to the question (v1): It seems OP is conflating, on one hand, a gauge transformation

$$ \tilde{A}_{\mu} ~=~ A_{\mu} +d_{\mu}\Lambda $$

with, on the other hand, a gauge-fixing condition, i.e. choosing a gauge, such e.g., Lorenz gauge, Coulomb gauge, axial gauge, temporal gauge, etc.

A gauge transformation can e.g. go between two gauge-fixing conditions. More generally, gauge transformations run along gauge orbits. Ideally a gauge-fixing condition intersects all gauge orbits exactly once.

Mathematically, depending on the topology of spacetime, it is often a non-trivial issue whether such a gauge-fixing condition is globally well-defined and uniquely specifies the gauge-field, cf. e.g. the Gribov problem. Existence and uniqueness of solutions to gauge-fixing conditions is the topic of several Phys.SE posts, see e.g. this and this Phys.SE posts.

  • $\begingroup$ +1 for the excellent referencing to related questions as well as related topics - as always $\endgroup$ – Danu Aug 7 '14 at 0:33

The Coulomb and Lorenz gauges are gauge fixing conditions, not gauge transformations. Your question still makes sense, but it should be phrased more like this: how can you prove that for arbitrary $\mathbf A_0$ and $V_0$ there always exists a gauge transformation to fields $\mathbf A$ and $V$ that satisfy these conditions? In fact many such transformations always exist (these conditions don't completely fix the gauge).

For the Coulomb gauge condition, we need $0 = \nabla\cdot\mathbf A = \nabla\cdot(\mathbf A_0 + \nabla\lambda) = \nabla\cdot\mathbf A_0 + \nabla^2 \lambda$. This is Poisson's equation, which can be solved for $\lambda$ with Green's functions. The Lorenz case is easier if you use four-vectors, in which case you get a 3+1 dimensional version of Poisson's equation.

  • $\begingroup$ For the sake of completeness, could you expand your answer to cover the Lorenz gauge in a more comprehensive way? $\endgroup$ – Danu Aug 7 '14 at 0:28

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.