How to perform vector calculus in curvilinear coordinates

This tutorial introduces some vector calculus capabilities of SageMath within the 3-dimensional Euclidean space. The corresponding tools have been developed via the SageManifolds project.

The tutorial is also available as a Jupyter notebook, either passive (nbviewer) or interactive (binder).

Using spherical coordinates

To use spherical coordinates \((r,\theta,\phi)\) within the Euclidean 3-space \(\mathbb{E}^3\), it suffices to declare the latter with the keyword coordinates='spherical':

sage: E.<r,th,ph> = EuclideanSpace(coordinates='spherical')
sage: E
Euclidean space E^3

Thanks to the notation <r,th,ph> in the above declaration, the coordinates \((r,\theta,\phi)\) are immediately available as three symbolic variables r, th and ph (there is no need to declare them via var(), i.e. to type r, th, ph = var('r th ph')):

sage: r is E.spherical_coordinates()[1]
True
sage: (r, th, ph) == E.spherical_coordinates()[:]
True
sage: type(r)
<type 'sage.symbolic.expression.Expression'>

Moreover, the coordinate LaTeX symbols are already set:

sage: latex(th)
{\theta}

The coordinate ranges are:

sage: E.spherical_coordinates().coord_range()
r: (0, +oo); th: (0, pi); ph: [0, 2*pi] (periodic)

\(\mathbb{E}^3\) is endowed with the orthonormal vector frame \((e_r, e_\theta, e_\phi)\) associated with spherical coordinates:

sage: E.frames()
[Coordinate frame (E^3, (∂/∂r,∂/∂th,∂/∂ph)),
 Vector frame (E^3, (e_r,e_th,e_ph))]

In the above output, (∂/∂r,∂/∂th,∂/∂ph) = \(\left(\frac{\partial}{\partial r}, \frac{\partial}{\partial\theta}, \frac{\partial}{\partial \phi}\right)\) is the coordinate frame associated with \((r,\theta,\phi)\); it is not an orthonormal frame and will not be used below. The default frame is the orthonormal one:

sage: E.default_frame()
Vector frame (E^3, (e_r,e_th,e_ph))

Defining a vector field

We define a vector field on \(\mathbb{E}^3\) from its components in the orthonormal vector frame \((e_r,e_\theta,e_\phi)\):

sage: v = E.vector_field(r*sin(2*ph)*sin(th)^2 + r,
....:                    r*sin(2*ph)*sin(th)*cos(th),
....:                    2*r*cos(ph)^2*sin(th), name='v')
sage: v.display()
v = (r*sin(2*ph)*sin(th)^2 + r) e_r + r*cos(th)*sin(2*ph)*sin(th) e_th
 + 2*r*cos(ph)^2*sin(th) e_ph

We can access to the components of \(v\) via the square bracket operator:

sage: v[1]
r*sin(2*ph)*sin(th)^2 + r
sage: v[:]
[r*sin(2*ph)*sin(th)^2 + r, r*cos(th)*sin(2*ph)*sin(th), 2*r*cos(ph)^2*sin(th)]

A vector field can evaluated at any point of \(\mathbb{E}^3\):

sage: p = E((1, pi/2, pi), name='p')
sage: p
Point p on the Euclidean space E^3
sage: p.coordinates()
(1, 1/2*pi, pi)
sage: vp = v.at(p)
sage: vp
Vector v at Point p on the Euclidean space E^3
sage: vp.display()
v = e_r + 2 e_ph

We may define a vector field with generic components:

sage: u = E.vector_field(function('u_r')(r,th,ph),
....:                    function('u_theta')(r,th,ph),
....:                    function('u_phi')(r,th,ph),
....:                    name='u')
sage: u.display()
u = u_r(r, th, ph) e_r + u_theta(r, th, ph) e_th + u_phi(r, th, ph) e_ph
sage: u[:]
[u_r(r, th, ph), u_theta(r, th, ph), u_phi(r, th, ph)]

Its value at the point \(p\) is then:

sage: up = u.at(p)
sage: up.display()
u = u_r(1, 1/2*pi, pi) e_r + u_theta(1, 1/2*pi, pi) e_th
 + u_phi(1, 1/2*pi, pi) e_ph

Differential operators in spherical coordinates

The standard operators \(\mathrm{grad}\), \(\mathrm{div}\), \(\mathrm{curl}\), etc. involved in vector calculus are accessible as methods on scalar fields and vector fields (e.g. v.div()). However, to allow for standard mathematical notations (e.g. div(v)), let us import the functions grad(), div(), curl() and laplacian():

sage: from sage.manifolds.operators import *

Gradient

We first introduce a scalar field, via its expression in terms of Cartesian coordinates; in this example, we consider a unspecified function of \((r,\theta,\phi)\):

sage: F = E.scalar_field(function('f')(r,th,ph), name='F')
sage: F.display()
F: E^3 → ℝ
   (r, th, ph) ↦ f(r, th, ph)

The value of \(F\) at a point:

sage: F(p)
f(1, 1/2*pi, pi)

The gradient of \(F\):

sage: grad(F)
Vector field grad(F) on the Euclidean space E^3
sage: grad(F).display()
grad(F) = d(f)/dr e_r + d(f)/dth/r e_th + d(f)/dph/(r*sin(th)) e_ph
sage: norm(grad(F)).display()
|grad(F)|: E^3 → ℝ
   (r, th, ph) ↦ sqrt((r^2*(d(f)/dr)^2 + (d(f)/dth)^2)*sin(th)^2
    + (d(f)/dph)^2)/(r*sin(th))

Divergence

The divergence of a vector field:

sage: s = div(u)
sage: s.display()
div(u): E^3 → ℝ
   (r, th, ph) ↦ ((r*d(u_r)/dr + 2*u_r(r, th, ph)
    + d(u_theta)/dth)*sin(th) + cos(th)*u_theta(r, th, ph)
    + d(u_phi)/dph)/(r*sin(th))
sage: s.expr().expand()
2*u_r(r, th, ph)/r + cos(th)*u_theta(r, th, ph)/(r*sin(th))
 + diff(u_theta(r, th, ph), th)/r + diff(u_phi(r, th, ph), ph)/(r*sin(th))
 + diff(u_r(r, th, ph), r)

For \(v\), we have:

sage: div(v).expr()
3

Curl

The curl of a vector field:

sage: s = curl(u)
sage: s
Vector field curl(u) on the Euclidean space E^3

sage: s.display()
curl(u) = (cos(th)*u_phi(r, th, ph) + sin(th)*d(u_phi)/dth
 - d(u_theta)/dph)/(r*sin(th)) e_r - ((r*d(u_phi)/dr + u_phi(r, th, ph))*sin(th)
 - d(u_r)/dph)/(r*sin(th)) e_th + (r*d(u_theta)/dr + u_theta(r, th, ph)
 - d(u_r)/dth)/r e_ph

For \(v\), we have:

sage: curl(v).display()
curl(v) = 2*cos(th) e_r - 2*sin(th) e_th

The curl of a gradient is always zero:

sage: curl(grad(F)).display()
curl(grad(F)) = 0

The divergence of a curl is always zero:

sage: div(curl(u)).display()
div(curl(u)): E^3 → ℝ
   (r, th, ph) ↦ 0

Laplacian

The Laplacian of a scalar field:

sage: s = laplacian(F)
sage: s.display()
Delta(F): E^3 → ℝ
   (r, th, ph) ↦ ((r^2*d^2(f)/dr^2 + 2*r*d(f)/dr
    + d^2(f)/dth^2)*sin(th)^2 + cos(th)*sin(th)*d(f)/dth
    + d^2(f)/dph^2)/(r^2*sin(th)^2)
sage: s.expr().expand()
2*diff(f(r, th, ph), r)/r + cos(th)*diff(f(r, th, ph), th)/(r^2*sin(th))
 + diff(f(r, th, ph), th, th)/r^2 + diff(f(r, th, ph), ph, ph)/(r^2*sin(th)^2)
 + diff(f(r, th, ph), r, r)

The Laplacian of a vector field:

sage: Du = laplacian(u)
sage: Du.display()
Delta(u) = ((r^2*d^2(u_r)/dr^2 + 2*r*d(u_r)/dr - 2*u_r(r, th, ph)
 + d^2(u_r)/dth^2 - 2*d(u_theta)/dth)*sin(th)^2 - ((2*u_theta(r, th, ph)
 - d(u_r)/dth)*cos(th) + 2*d(u_phi)/dph)*sin(th) + d^2(u_r)/dph^2)/(r^2*sin(th)^2) e_r
 + ((r^2*d^2(u_theta)/dr^2 + 2*r*d(u_theta)/dr + 2*d(u_r)/dth + d^2(u_theta)/dth^2)*sin(th)^2
 + cos(th)*sin(th)*d(u_theta)/dth - 2*cos(th)*d(u_phi)/dph - u_theta(r, th, ph)
 + d^2(u_theta)/dph^2)/(r^2*sin(th)^2) e_th
 + ((r^2*d^2(u_phi)/dr^2 + 2*r*d(u_phi)/dr
 + d^2(u_phi)/dth^2)*sin(th)^2 + (cos(th)*d(u_phi)/dth + 2*d(u_r)/dph)*sin(th)
 + 2*cos(th)*d(u_theta)/dph - u_phi(r, th, ph) + d^2(u_phi)/dph^2)/(r^2*sin(th)^2) e_ph

Since this expression is quite lengthy, we may ask for a display component by component:

sage: Du.display_comp()
Delta(u)^1 = ((r^2*d^2(u_r)/dr^2 + 2*r*d(u_r)/dr - 2*u_r(r, th, ph) + d^2(u_r)/dth^2
 - 2*d(u_theta)/dth)*sin(th)^2 - ((2*u_theta(r, th, ph) - d(u_r)/dth)*cos(th)
 + 2*d(u_phi)/dph)*sin(th) + d^2(u_r)/dph^2)/(r^2*sin(th)^2)
Delta(u)^2 = ((r^2*d^2(u_theta)/dr^2 + 2*r*d(u_theta)/dr + 2*d(u_r)/dth
 + d^2(u_theta)/dth^2)*sin(th)^2 + cos(th)*sin(th)*d(u_theta)/dth
 - 2*cos(th)*d(u_phi)/dph - u_theta(r, th, ph) + d^2(u_theta)/dph^2)/(r^2*sin(th)^2)
Delta(u)^3 = ((r^2*d^2(u_phi)/dr^2 + 2*r*d(u_phi)/dr + d^2(u_phi)/dth^2)*sin(th)^2
 + (cos(th)*d(u_phi)/dth + 2*d(u_r)/dph)*sin(th) + 2*cos(th)*d(u_theta)/dph
 - u_phi(r, th, ph) + d^2(u_phi)/dph^2)/(r^2*sin(th)^2)

We may expand each component:

sage: for i in E.irange():
....:     s = Du[i].expand()
sage: Du.display_comp()
Delta(u)^1 = 2*d(u_r)/dr/r - 2*u_r(r, th, ph)/r^2
 - 2*cos(th)*u_theta(r, th, ph)/(r^2*sin(th)) + cos(th)*d(u_r)/dth/(r^2*sin(th))
 + d^2(u_r)/dth^2/r^2 - 2*d(u_theta)/dth/r^2 - 2*d(u_phi)/dph/(r^2*sin(th))
 + d^2(u_r)/dph^2/(r^2*sin(th)^2) + d^2(u_r)/dr^2
Delta(u)^2 = 2*d(u_theta)/dr/r + 2*d(u_r)/dth/r^2 + cos(th)*d(u_theta)/dth/(r^2*sin(th))
 + d^2(u_theta)/dth^2/r^2 - 2*cos(th)*d(u_phi)/dph/(r^2*sin(th)^2)
 - u_theta(r, th, ph)/(r^2*sin(th)^2) + d^2(u_theta)/dph^2/(r^2*sin(th)^2)
 + d^2(u_theta)/dr^2
Delta(u)^3 = 2*d(u_phi)/dr/r + cos(th)*d(u_phi)/dth/(r^2*sin(th))
 + d^2(u_phi)/dth^2/r^2 + 2*d(u_r)/dph/(r^2*sin(th))
 + 2*cos(th)*d(u_theta)/dph/(r^2*sin(th)^2) - u_phi(r, th, ph)/(r^2*sin(th)^2)
 + d^2(u_phi)/dph^2/(r^2*sin(th)^2) + d^2(u_phi)/dr^2

As a test, we may check that these formulas coincide with those of Wikipedia’s article Del in cylindrical and spherical coordinates.

Using cylindrical coordinates

The use of cylindrical coordinates \((\rho,\phi,z)\) in the Euclidean space \(\mathbb{E}^3\) is on the same footing as that of spherical coordinates. To start with, one has simply to declare:

sage: E.<rh,ph,z> = EuclideanSpace(coordinates='cylindrical')

The coordinate ranges are then:

sage: E.cylindrical_coordinates().coord_range()
rh: (0, +oo); ph: [0, 2*pi] (periodic); z: (-oo, +oo)

The default vector frame is the orthonormal frame \((e_\rho,e_\phi,e_z)\) associated with cylindrical coordinates:

sage: E.default_frame()
Vector frame (E^3, (e_rh,e_ph,e_z))

and one may define vector fields from their components in that frame:

sage: v = E.vector_field(rh*(1+sin(2*ph)), 2*rh*cos(ph)^2, z,
....:                    name='v')
sage: v.display()
v = rh*(sin(2*ph) + 1) e_rh + 2*rh*cos(ph)^2 e_ph + z e_z
sage: v[:]
[rh*(sin(2*ph) + 1), 2*rh*cos(ph)^2, z]

sage: u = E.vector_field(function('u_rho')(rh,ph,z),
....:                    function('u_phi')(rh,ph,z),
....:                    function('u_z')(rh,ph,z),
....:                    name='u')
sage: u.display()
u = u_rho(rh, ph, z) e_rh + u_phi(rh, ph, z) e_ph + u_z(rh, ph, z) e_z
sage: u[:]
[u_rho(rh, ph, z), u_phi(rh, ph, z), u_z(rh, ph, z)]

Differential operators in cylindrical coordinates

sage: from sage.manifolds.operators import *

The gradient:

sage: F = E.scalar_field(function('f')(rh,ph,z), name='F')
sage: F.display()
F: E^3 → ℝ
   (rh, ph, z) ↦ f(rh, ph, z)
sage: grad(F)
Vector field grad(F) on the Euclidean space E^3
sage: grad(F).display()
grad(F) = d(f)/drh e_rh + d(f)/dph/rh e_ph + d(f)/dz e_z

The divergence:

sage: s = div(u)
sage: s.display()
div(u): E^3 → ℝ
   (rh, ph, z) ↦ (rh*d(u_rho)/drh + rh*d(u_z)/dz + u_rho(rh, ph, z) + d(u_phi)/dph)/rh
sage: s.expr().expand()
u_rho(rh, ph, z)/rh + diff(u_phi(rh, ph, z), ph)/rh + diff(u_rho(rh, ph, z), rh)
 + diff(u_z(rh, ph, z), z)

The curl:

sage: s = curl(u)
sage: s
Vector field curl(u) on the Euclidean space E^3
sage: s.display()
curl(u) = -(rh*d(u_phi)/dz - d(u_z)/dph)/rh e_rh + (d(u_rho)/dz - d(u_z)/drh) e_ph
 + (rh*d(u_phi)/drh + u_phi(rh, ph, z) - d(u_rho)/dph)/rh e_z

The Laplacian of a scalar field:

sage: s = laplacian(F)
sage: s.display()
Delta(F): E^3 → ℝ
   (rh, ph, z) ↦ (rh^2*d^2(f)/drh^2 + rh^2*d^2(f)/dz^2 + rh*d(f)/drh
    + d^2(f)/dph^2)/rh^2
sage: s.expr().expand()
diff(f(rh, ph, z), rh)/rh + diff(f(rh, ph, z), ph, ph)/rh^2
 + diff(f(rh, ph, z), rh, rh) + diff(f(rh, ph, z), z, z)

The Laplacian of a vector field:

sage: Du = laplacian(u)
sage: Du.display()
Delta(u) = (rh^2*d^2(u_rho)/drh^2 + rh^2*d^2(u_rho)/dz^2 + rh*d(u_rho)/drh
 - u_rho(rh, ph, z) - 2*d(u_phi)/dph + d^2(u_rho)/dph^2)/rh^2 e_rh
 + (rh^2*d^2(u_phi)/drh^2 + rh^2*d^2(u_phi)/dz^2 + rh*d(u_phi)/drh
 - u_phi(rh, ph, z) + d^2(u_phi)/dph^2 + 2*d(u_rho)/dph)/rh^2 e_ph
 + (rh^2*d^2(u_z)/drh^2 + rh^2*d^2(u_z)/dz^2 + rh*d(u_z)/drh
 + d^2(u_z)/dph^2)/rh^2 e_z

sage: for i in E.irange():
....:     s = Du[i].expand()
sage: Du.display_comp()
Delta(u)^1 = d(u_rho)/drh/rh - u_rho(rh, ph, z)/rh^2 - 2*d(u_phi)/dph/rh^2
 + d^2(u_rho)/dph^2/rh^2 + d^2(u_rho)/drh^2 + d^2(u_rho)/dz^2
Delta(u)^2 = d(u_phi)/drh/rh - u_phi(rh, ph, z)/rh^2 + d^2(u_phi)/dph^2/rh^2
 + 2*d(u_rho)/dph/rh^2 + d^2(u_phi)/drh^2 + d^2(u_phi)/dz^2
Delta(u)^3 = d(u_z)/drh/rh + d^2(u_z)/dph^2/rh^2 + d^2(u_z)/drh^2 + d^2(u_z)/dz^2

Again, we may check that the above formulas coincide with those of Wikipedia’s article Del in cylindrical and spherical coordinates.

Changing coordinates

Given the expression of a vector field in a given coordinate system, SageMath can compute its expression in another coordinate system, see the tutorial How to change coordinates