I understand the wave vector for plane waves in Cartesian coordinates. Along the direction of propagation of wave, $$k = \sqrt{k_x^2+k_y^2+k_z^2} = \frac{2\pi}{\lambda}$$ If $(k_x, k_y, k_z) \neq 0$, this would mean we can find troughs and crests moving along $(x,y,z)$ directions respectively. The dot product in wave equation in Cartesian coordinates, $\vec{dr}=(dx,dy,dz)$ would be $$\vec{k}\cdot\vec{dr} = k_x dx + k_y dy + k_z dz$$

What would the interpretation be in case of spherical coordinates? Can we still have wave vector, $k = (k_{r},k_{\theta},k_{\phi})$ along $(r,\theta,\phi)$ directions such that $$k = \sqrt{k_r^2+k_{\theta}^2+k_{\phi}^2} = \frac{2\pi}{\lambda}$$ If $(k_{r},k_{\theta},k_{\phi}) \neq 0$, would this mean we can find troughs and crests moving along $(r,\theta,\phi)$ directions respectively. Is the following dot product in wave equation in spherical coordinates, $\vec{dr}=(dr,d\theta,d\phi)$ correct? $$\vec{k}\cdot\vec{dr}= k_{r}dr + k_{\theta}rd\theta + k_{\phi}r\sin\theta d\phi$$

  • $\begingroup$ The "dot product" $\vec{k}\cdot\vec{dr}= k_{r}dr + k_{\theta}rd\theta + k_{\phi}r\sin\theta d\phi$ doesn't make much sense. It could be appropriate in a small fraction of circumstances, but you'd have to be careful to explain what's happening around it. $\endgroup$ Feb 1, 2021 at 16:49
  • 1
    $\begingroup$ $\vec{dr}=(dr,d\theta,d\phi)~$ this is not correct $\endgroup$
    – Eli
    Feb 1, 2021 at 17:50
  • $\begingroup$ Be careful with position vectors in polar coordinate systems. The coordinates of a point in spherical for example are (r,th,ph), but the position vector is r alone. The angles are contained within the unit vector for r, because this is a curvilinear system. Look up 'position vector in spherical coordinates' and you should find plenty of information about it. $\endgroup$ Feb 5, 2021 at 12:23

2 Answers 2


As a general rule, having a wave vector $\vec k$ implies that you have a travelling wave, i.e. one which transports momentum and energy.

This is particularly relevant to part of your question,

would this mean we can find troughs and crests moving along $(r,\theta,\phi)$ directions respectively,

because you cannot have a travelling wave along the $\theta$ direction $-$ the energy would accumulate at the north axis and you'd need a source along the south. A travelling wave along the $\phi$ axis is less of a problem (you can have energy circling around indefinitely) but it's still a problem (you would have energy circling around indefinitely).

That said, the general answer to your question is: yes. The way to get at it is to rephrase the Cartesian wavevector into saying that we want to solve the wave equation, $$ \nabla^2 f = \frac{1}{c^2}\partial_t^2 f, $$ first by separating out $f(\mathbf r,t)=f(\mathbf r)e^{-i\omega t}$ into a space- and time-dependent parts, and then handling the resulting Helmholtz equation for the spatial part by using Cartesian coordinates, $$ \frac{\partial^2 f}{\partial x^2} + \frac{\partial^2 f}{\partial y^2} + \frac{\partial^2 f}{\partial z^2} = k^2 f, $$ and using separation of variables. Here the wavenumber arises as a straight definition $k^2=\omega^2/c^2$, and once you solve the ODE of each individual coordinate into individual exponentials, the result can be folded back into $f(\mathbf r)=e^{i\mathbf k\cdot \mathbf r}$, with a wavevector $\mathbf k$ whose norm coincides with $k$.

Now, if you want the wave behaviour of $f(\mathbf r)$ to go along the spherical coordinate surfaces, then you do the same, but now you address the Helmholtz equation $$ \nabla^2 f = k^2 f $$ by expressing $\nabla^2$ in spherical coordinates, and you separate the variables as $f(\mathbf r)=R(r)\Theta(\theta)\Phi(\phi)$. The resulting procedure is standard, so I won't reproduce it here (try e.g. Jackson's Classical Electrodynamics for a thorough solution, or any book on advanced mathematical methods for physics). The upshot is that:

  • The solution for the angular variables is that $\Theta(\theta)\Phi(\phi) = Y_{lm}(\theta,\phi)$ must be one of the family of spherical harmonics, which are functions which "wave" exclusively in the angular directions, with nodes along the $\theta=\rm const$ and $\phi=\rm const$ directions. However, because of the problems mentioned at the start, particularly in the $\theta$ direction, they are not travelling waves $-$ they are standing waves.

    Moreover, the spherical-harmonics solutions do not have any associated "wave vector". Instead, the description of the wave moves over to the two indices $l$ and $m$, which describe the total number of nodes and how they're split into the azimuth and altitude directions.

  • For the radial variable, the Helmholtz equation becomes the spherical Bessel equation, $$ r^{2}{\frac {d^{2}R}{dr^{2}}}+2r{\frac {dR}{dr}}+\left(k^2r^{2}-l(l+1)\right)R=0, $$ and its solutions are the spherical Bessel functions. As a general rule, all of these are spherical waves, with crests and troughs along the radial direction, but there's several different types:

    • The spherical Bessel functions of the first kind, $j_l(kr)$, describe standing waves, and are regular at the origin.
    • The spherical Bessel functions of the second kind, $y_l(kr)$, describe standing waves, but they have a singularity at the origin, so they can only be used if the origin is not part of the region of interest.
    • The spherical Hankel functions, $h_l^{\pm}(kr)$, describe travelling waves, with a source (or sink) at the origin, where they have a singularity.

\begin{align*} &\text{with}\\ \vec{R}&=r\,\left[ \begin {array}{c} \cos \left( \phi \right) \sin \left( \theta \right) \\ \sin \left( \theta \right) \sin \left( \phi \right) \\ \cos \left( \theta \right) \end {array} \right] \\ \vec{dR}&=\left[ \begin {array}{c} \cos \left( \phi \right) \sin \left( \theta \right) \\ \sin \left( \theta \right) \sin \left( \phi \right) \\ \cos \left( \theta \right) \end {array} \right] \,dr\\&+\left[ \begin {array}{c} -r\sin \left( \theta \right) \sin \left( \phi \right) \\ r\cos \left( \phi \right) \sin \left( \theta \right) \\ 0\end {array} \right] \,d\phi\\&+ \left[ \begin {array}{c} r\cos \left( \phi \right) \cos \left( \theta \right) \\ r\cos \left( \theta \right) \sin \left( \phi \right) \\ -r\sin \left( \theta \right) \end {array} \right] \,d\theta \end{align*}

thus $~\vec{dR}\ne (dr~,d\phi~,d\theta)$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.