I'm studying introductory vector calculus and need to confirm/clarify my concepts. The definition of the derivative of a vector (for example in $\mathbb{R}^2$) if the unit vectors are constant throughout the 2D space is in terms of its components: if we have $\mathbb{r}(t)=(x(t), y(t))$ in the standard Cartesian basis, then


Now if we move to polar coordinates $\rho, \phi$, then the unit basis vectors $\mathbb{e}_{\rho},\mathbb{e}_{\phi}$ will change direction depending on the location in 2D space. To define the derivative in this case, the book that I'm studying gives the following quick method: we see that $\mathbb{r}=\rho \mathbb{e}_{\rho}$ (where $\rho$ is the distance of the vector's endpoint from the origin), which means $$\frac{d\mathbb{r}}{dt}=\frac{d\rho}{dt}\mathbb{e}_{\rho}+\rho\frac{d\mathbb{e}_{\rho}}{dt}$$ enter image description here

So far, so good: $\frac{d\rho}{dt}$ can be calculated since we can express $\rho$ in terms of $x(t)$ and $y(t)$, and differentiate that expression w.r.t. $t$. In this specific case, we can also express $\mathbb{e}_{\rho}=(\cos\phi)\mathbb{e}_x + (\sin\phi)\mathbb{e}_y$. It turns out that $$\frac{d\mathbb{e}_{\rho}}{dt}=\frac{d\phi}{dt}\mathbb{e}_{\phi}$$ because of the specific way $\mathbb{e}_{\rho}$ and $\mathbb{e}_{\phi}$ are defined in terms of $\mathbb{e}_x$ and $\mathbb{e}_y$.

Expressing the same vector $\mathbb{r}$ in a general curvilinear coordinate system $u,v$, enter image description here

To even start differentiating $\mathbb{r}$, we need to find the components of $\mathbb{r}$ in the new system. I'm assuming the way to identify $\mathbb{r}$ is to identify it as the intersection of two coordinate curves $u=c_1$ and $v=c_2$ - in this case, $u=5$ and $v=4$. Is my understanding correct? Is this the way to identify the components of a vector in a curvilinear system?

So if we have some differentiable functions $f,g$ such that $u=f(x,y)$ and $v=g(x,y)$ and $\mathbb{r}=u\mathbb{e}_u+v\mathbb{e}_v$, then $$\frac{d\mathbb{r}}{dt}=\frac{du}{dt}\mathbb{e}_u+u\frac{d\mathbb{e}_u}{dt}+\frac{dv}{dt}\mathbb{e}_v+v\frac{d\mathbb{e}_v}{dt}$$

$\frac{du}{dt}$ can be identified as $\frac{df(x(t),y(t))}{dt}$ and can be evaluated. How does one, in general, express basis vectors $\mathbb{e}_u$ and $\mathbb{e}_v$ in terms of $\mathbb{e}_x$ and $\mathbb{e}_y$? And even if we do manage to define curvilinear basis vectors in terms of $\mathbb{e}_x,\mathbb{e}_y$, it's not necessary that we'll get a nice expression for $\frac{d\mathbb{e}_u}{dt}$ and $\frac{d\mathbb{e}_v}{dt}$ in terms of $\mathbb{e}_u$ and $\mathbb{e}_v$. How do we get the curvilinear components of $\frac{d\mathbb{r}}{dt}$ in that case?

  • $\begingroup$ A comment on your notation: Usually, the position vector is $\mathbf{r}$ while the length of the position vector is $r$. So, for example, it would be $\frac{d\mathbf{r}}{dt} = \frac{dx}{dt}\mathbf{e}_x + \frac{dy}{dt}\mathbf{e}_y$ and $\mathbf{r} = \rho\mathbf{e}_\rho$ and $r = \sqrt{x^2+y^2} = \rho$ $\endgroup$ – Hal Hollis Dec 26 '19 at 19:31
  • $\begingroup$ @HalHollis: Thanks! I'll edit the question asap $\endgroup$ – Shirish Kulhari Dec 26 '19 at 19:31
  • $\begingroup$ Math mods: Please merge. $\endgroup$ – Qmechanic Dec 29 '19 at 5:35

This may or may not help, but we can write $$d \vec r=\frac{\partial \vec r}{\partial u}du+\frac{\partial \vec r}{\partial v}dv+\frac{\partial \vec r}{\partial w}dw$$ But by definition of the basis vectors in the $u, v, w$ system, $$d \vec r=\vec e_udu+\vec e_vdv+\vec e_wdw$$ Therefore we have $$\vec e_u=\frac{\partial \vec r}{\partial u}$$ and so on. But $$\frac{\partial \vec r}{\partial u}=\frac{\partial x}{\partial u}\frac{\partial \vec r}{\partial x}+\frac{\partial y}{\partial u}\frac{\partial \vec r}{\partial y}+\frac{\partial z}{\partial u}\frac{\partial \vec r}{\partial z}$$ So $$\vec e_u=\frac{\partial x}{\partial u}\vec e_x+\frac{\partial y}{\partial u}\vec e_y+\frac{\partial z}{\partial u}\vec e_z$$ and similarly for $\vec e_v$ and $\vec e_w$.

  • $\begingroup$ Thanks for the answer! The math makes sense, but intuitively what does it mean for $\mathbb{e}_u$ to be equal to $\frac{\partial r}{\partial u}$. Could you explain it geometrically, if possible? $\endgroup$ – Shirish Kulhari Dec 27 '19 at 6:42
  • $\begingroup$ I think it must help! Since you can calculate those derivatives by inverting the Jacobian matrix. $\endgroup$ – Guillermo BCN Dec 27 '19 at 9:49
  • $\begingroup$ @Shirish Kulhari In my first equation, ∂𝑟⃗/∂𝑢 is clearly a vector and is the vectorial amount by which displacement changes per unit excursion in the direction such that $u$ changes without $v$ or $w$ changing, so it is the basis vector along the $u$ axis. $\endgroup$ – Philip Wood Dec 27 '19 at 9:50
  • $\begingroup$ I would add that this is in fact the definition of the basis vectors. $\endgroup$ – Guillermo BCN Dec 27 '19 at 10:05
  • $\begingroup$ @PhilipWood: So you expressed the differential vector $d\mathbf{r}$ in terms of $\mathbf{e}_u, \mathbf{e}_v, \mathbf{e}_w$. That's probably because the basis vectors themselves are only defined locally. Does that mean it's not possible to express the non-differential vector $\mathbf{r}$ (as you see in the 2nd figure of the question) in terms of basis vectors in curvilinear coordinates? i.e. Does $\mathbf{r}=u\mathbf{e}_u + v\mathbf{e}_v+w\mathbf{e}_w$ make any sense? $\endgroup$ – Shirish Kulhari Dec 27 '19 at 13:05

To get the unit vectors in cartesian coordinates, define $\boldsymbol{r} = x e_x + y e_y$ and calculate the partial derivatives $$\frac{\partial \boldsymbol{r}}{\partial x} = e_x \quad \text{and} \quad \frac{\partial \boldsymbol{r}}{\partial y} = e_y.$$ Now suppose the coordinate transformation is given by a locally invertible functions $$\begin{pmatrix} x \\ y \end{pmatrix} = \begin{pmatrix} f^1(u,v) \\ f^2(u,v)\end{pmatrix}.$$ By plugging the transformation functions into $\boldsymbol{r}$ and taking the derivatives, one obtaines $$\frac{\partial \boldsymbol{r}}{\partial u} = \frac{\partial f^1(u,v)}{\partial u}e_x + \frac{\partial f^2(u,v)}{\partial u}e_y =\boldsymbol{r}_u \quad \text{and} \quad \frac{\partial \boldsymbol{r}}{\partial v} = \frac{\partial f^1(u,v)}{\partial v}e_x + \frac{\partial f^2(u,v)}{\partial v}e_y = \boldsymbol{r}_v.$$ Now you only need to normalize $\boldsymbol{r}_u$ and $\boldsymbol{r}_v$ to get the orthonormal basis vectors $e_u$ and $e_v$, which are expressed in terms of the canonical basis.

The curvilinear components of $\dot{\boldsymbol{r}}$ are quite easy to obtain. The trick is to use the chain rule: $$\dot{\boldsymbol{r}} = \left(\frac{\partial f^1}{\partial u} \dot{u} + \frac{\partial f^1}{\partial v}\dot{v}\right)e_x + \left(\frac{\partial f^2}{\partial u}\dot{u} + \frac{\partial f^2}{\partial v} \dot{v}\right)e_y.$$ Rearranging yields $$\dot{\boldsymbol{r}} = \left(\frac{\partial f^1}{\partial u} e_x + \frac{\partial f^2}{\partial u}e_y\right) \dot{u} + \left(\frac{\partial f^1}{\partial v} e_x + \frac{\partial f^2}{\partial v}e_y\right) \dot{v} = \boldsymbol{r}_u \dot{u} + \boldsymbol{r}_v \dot{v}.$$

  • $\begingroup$ IMO, \dot{\mathbf{r}} looks better than using \dot{\boldsymbol r}. $\endgroup$ – Kyle Kanos Dec 27 '19 at 0:18
  • $\begingroup$ For reference, I'm on iOS 13.3 & am seeing a massive gap between the dot and the r. $\endgroup$ – Kyle Kanos Dec 27 '19 at 0:19
  • $\begingroup$ One doubt: Why do you equate $\frac{\partial \mathbb{r}}{\partial u}$ to $\mathbb{r}_u$? I can't understand the intuition behind that. $\endgroup$ – Shirish Kulhari Dec 27 '19 at 6:28
  • $\begingroup$ I just used $\mathbf{r}_u$ and $\mathbf{r}_v$ as shorthand notation for the partial derivatives. $\endgroup$ – Tobi7 Dec 27 '19 at 9:42

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.