4
$\begingroup$

What's the difference between a change of frame and a change of coordinates? I feel like both are transformations on the coordinates but change of frame changes also the vectors.

$\endgroup$

2 Answers 2

3
$\begingroup$

There's a difference because your coordinates do not always specify your frame. It's true that coordinates always give you one particular choice of a frame, but you can choose to use a different set of basis vectors to represent the vector space at any particular point.

The first example of this you'll probably run into is the difference between the coordinate basis and the orthonormal basis in spherical coordinates. They're based on the exact same coordinate system, but they're totally different frames. If $\mathbf{p}(r, \theta, \phi)$ is the position vector to the point with spherical coordinates $r, \theta, \phi$, then the coordinate basis vectors are defined as \begin{align} \mathbf{r} &= \frac{\partial \mathbf{p}} {\partial r} \\ \boldsymbol{\theta} &= \frac{\partial \mathbf{p}} {\partial \theta} \\ \boldsymbol{\phi} &= \frac{\partial \mathbf{p}} {\partial \phi}. \end{align} Now, remember that we can express the point in terms of the usual orthonormal Cartesian basis as \begin{equation} \mathbf{p} = r\sin\theta\cos\phi\, \hat{\mathbf{x}} + r\sin\theta\sin\phi\, \hat{\mathbf{y}} + r\cos\theta\, \hat{\mathbf{z}}, \end{equation} which we can differentiate to find \begin{align} \mathbf{r} &= \sin\theta\cos\phi\, \hat{\mathbf{x}} + \sin\theta\sin\phi\, \hat{\mathbf{y}} + \cos\theta\, \hat{\mathbf{z}} \\ \boldsymbol{\theta} &= r\cos\theta\cos\phi\, \hat{\mathbf{x}} + r\cos\theta\sin\phi\, \hat{\mathbf{y}} - r\sin\theta\, \hat{\mathbf{z}} \\ \boldsymbol{\phi} &= -r\sin\theta\sin\phi\, \hat{\mathbf{x}} + r\sin\theta\cos\phi\, \hat{\mathbf{y}}. \end{align} So already, we see that there are two different bases for the vector space at this point: one that uses the usual $(\hat{\mathbf{x}}, \hat{\mathbf{y}}, \hat{\mathbf{z}})$ basis, and another that uses $(\mathbf{r}, \boldsymbol{\theta}, \boldsymbol{\phi})$. Any vector can be expanded in either basis, even though they correspond to the same point.

In fact, we can even find another basis based on spherical coordinates. Using the expressions above, it's a simple exercise to see that we have \begin{align} \mathbf{r} \cdot \mathbf{r} &= 1 \\ \boldsymbol{\theta} \cdot \boldsymbol{\theta} &= r^2 \\ \boldsymbol{\phi} \cdot \boldsymbol{\phi} &= r^2 \sin^2 \theta. \end{align} So this frame is not orthonormal. If you want an orthonormal basis, you need \begin{align} \mathbf{e}_r &= \mathbf{r} \\ \mathbf{e}_\theta &= \frac{1}{r}\boldsymbol{\theta} \\ \mathbf{e}_\phi &= \frac{1}{r\sin\theta} \boldsymbol{\phi}. \end{align} You'll see and use both of these frames. They are clearly different frames, even though they're based on exactly the same coordinates.

More generally, you're free to define your frame at any point in any way you want (as long as you keep the same number of vectors, and keep them linearly independent). So a change of coordinates implies a change in coordinate frame, but it is not equivalent to a change in frame generally.

$\endgroup$
7
  • $\begingroup$ "If $p(r, \theta,\phi)$ is the position vector to the point...", "remember that we can express the point in terms of the usual orthonormal Cartesian basis" it means that we assume a priori that the space is a vectorial space in which we know one basis (the orthonormal Cartesian basis). Is it correct? Then, are the coordinates $r,\theta,\phi$ just numbers that indicate a certain position vector? $\endgroup$
    – SimoBartz
    Commented Jul 3, 2019 at 14:22
  • $\begingroup$ Well, in this particular case, I'm implicitly assuming three-dimensional Euclidean space, so we can assume that we know one basis. And in this case, it is true that $r, \theta, \phi$ are just numbers that indicate a certain position vector. But this is just one particularly simple (and hopefully familiar) example to illustrate my point. More generally, in differential geometry, a position is not given by a vector — though it's still very true that the coordinates only give you one particular frame, and you're free to use a different frame. $\endgroup$
    – Mike
    Commented Jul 3, 2019 at 14:34
  • $\begingroup$ What if, for example, I want to traslate the frame? In this example if I apply the corresponding change of coordinates the position vector doesn't change so it looks impossible to have a translation $\endgroup$
    – SimoBartz
    Commented Jul 3, 2019 at 14:41
  • $\begingroup$ I'm not actually sure what you mean by "translate a frame". The word "frame" really means just a basis for a vector space (in a particular order), and there's no concept of translation of a basis. There's a separate notion of a physical space — a "manifold". At each point of this manifold, there is a separate vector space associated with how points move in the manifold. But as you move through the manifold, the vector spaces all stay where they are. So if you translate points of the manifold, you change the vector spaces that they see, and whatever frames you might have chosen at those points. $\endgroup$
    – Mike
    Commented Jul 3, 2019 at 16:15
  • $\begingroup$ "if I apply the corresponding change of coordinates the position vector doesn't change" What change of coordinates are you using?! Unless you're just changing $\phi$ when $\theta=0$ or $\pi$, or if you're changing $\theta$ and $\phi$ when $r=0$, that $\mathbf{p}$ vector will change. $\endgroup$
    – Mike
    Commented Jul 3, 2019 at 16:16
1
$\begingroup$

It depends what you mean with 'change'. If 'change' is 'rate of change' i.e. the operator $\frac{d}{dt}$ acting on a vector $\vec v$, then the following is the case:

Any vector that is the element of a vector space can be split into its basis vectors $\vec e_i$ and its scalar components $v^i$, so that $\vec v = \sum_{i=1}^{n=3} v^i \vec e_i \equiv v^i \vec e_i$, where the last definition just uses the summation convention for indices.

If you now take $\frac{d}{dt}\vec v$ then according to the product rule you get $$\frac{d}{dt}\vec v = \left(\frac{d}{dt}v^i\right) \vec e_i + v^i\left(\frac{d}{dt}\vec e_i\right).$$ Now $\left(\frac{d}{dt}v^i\right)$ is what I would call a change of coordinates, and $\left(\frac{d}{dt}\vec e_i\right)$ a change of frame. The change of coordinates is just the change of the scalar numbers denoting some position in the given coordinate system $\vec e_i$. Should the coordinate system be accelerated, or, you just wish to change the 'grid' with which the $v^i$ are computed, then you need to change the basis vectors accordingly.

Mathematically there is no difference between a change of coordinates at a given time, or an accelerated coordinate system w.r.t time, they give you only different transformation Jacobians.

On another note, the change in basis vectors can impact our understanding of physics fundamentally. Simply changing from cartesian to spherical coordinates in Newtons equation of motion (at a given time) already introduces a term that behaves like a centrifugal force, although we only changed our 'viewpoint'.

$\endgroup$
5
  • $\begingroup$ "The change of coordinates is just the change of the scalar numbers denoting some position in the given coordinate system 𝑒⃗𝑖" does this mean that you are assuming the position to be a vector in a 3D vectorial space where the coordinates are 3 numbers that identify the vector? $\endgroup$
    – SimoBartz
    Commented Jul 3, 2019 at 14:26
  • $\begingroup$ @SimoBartz: In this example, yes. But it doesn't have to be the position vector, nor does the space need to be restricted to 3D space. Note that $\frac{d}{dt} \vec e_i$ in the end must evaluate to some quantity in your co-moving frame, so that $\frac{d}{dt} \vec e_i = \hat{v}^i \vec e_i$, where the numbers $\hat{v}^i$ then account for the effect of change-of-frame in your co-moving frame. $\endgroup$ Commented Jul 3, 2019 at 14:31
  • $\begingroup$ I don't understand how it's possible to derive in time $\vec e_i$ since we don't have it to be time dependent $\endgroup$
    – SimoBartz
    Commented Jul 3, 2019 at 14:46
  • $\begingroup$ @SimoBartz: In general, $\vec e_i$ can of course be time-dependent! For circular motion they are exactly that. Even for linear, nonaccelerated motion the $\vec e_i$ are time-dependent, as the frame moves through space, but in this case there appear no fictitious forces in Newton's law, so we don't notice it. $\endgroup$ Commented Jul 3, 2019 at 20:43
  • $\begingroup$ your answer an the one of Mike lead me to another fundamental question that I did here, give check if you like physics.stackexchange.com/questions/489587/… $\endgroup$
    – SimoBartz
    Commented Jul 3, 2019 at 20:46