Stokes' Theorem: Φ (u, v) was then said to orientation-preserving when the orientation was specified by the unit vector Φ
Stokes' Theorem: Φ (u, v) was then said to orientation-preserving when the orientation was specified by the unit vector Φ
Stokes' Theorem: Φ (u, v) was then said to orientation-preserving when the orientation was specified by the unit vector Φ
Following our by now well-trodden path of extending known results by increasing dimension, we extend Greens Theorem for domains in the plane to surfaces in 3-space. This is known as Stokes Theorem. It relates a surface integral over a surface S to a line integral over its boundary S. To state the theorem it will be important to be fairly precise about the assumptions on S and S. Recall that a parametrization (u, v) for S is said to be regular when |u v | = 0 on S; if S was oriented, such a (u, v) was then said to orientation-preserving when the orientation was specied by the unit vector en (u, v) = u v . |u v |
Recall also that an orientation on a curve such as S species a positive direction on the curve. We need to relate the orientation on S to that on its boundary S. The following familiar quadric surfaces show how its done:
In all three, orientation by the outward unit normal on the surface has been chosen. But in the rst example, a sphere, the surface is closed because it has no boundary; we write S = to indicate that S is the empty set. For the second, a paraboloid, the boundary has only component, while for the third one, a hyperboloid, there are two boundary curves. In all cases, however, the orientation on S is chosen so that as one walks as a normal in the positive direction around a boundary curve the surface always lies to ones left. Such an orientation on S is called the boundary orientation.
Stokes Theorem: when S is a surface having an orientation-preserving parametrization (u, v) : D R2 S which is 1-1 and regular except possibly on D, then curl F dS = F ds
D S
for all smooth vector elds F. The line integral is taken relative to the boundary orientation on S, and the surface integral is zero if S has no boundary.
The 1-1 property is a technical condition that simplies the proof. It is satised in all standard examples. Lets see how Stokes theorem works in practice. There are three main sets of examples. I. Surface integral to a line integral: evaluate the integral I =
S
curl F dS
when F = yz i + 3zx j + xy k and S is the blue part of the sphere x2 + y 2 + z 2 = 4 as shown to the right lying inside the cylinder x2 + y 2 = 1 and oriented upwards. Solution: the sphere and the cylinder intersect when 4 = x2 + y 2 + z 2 = 1 + z 2 , i.e., when z = 3. Thus the boundary of S is a circle of radius 1 with boundary orientation counter-clockwise. It can be parametrized by r(t) = cos t i + sin t j + 3k, 0 t 2 . F(r(t)) = while r (t) = sin t i + cos t j . In this case, F(r(t)) r (t) = ( ) 3 3 cos2 t sin2 t . Now 3 sin t i + 3 3 cos t j + sin t cos t k ,
F ds
when F = 3y 2 i + 2x j + z 2 k and C is the curve oriented counter-clockwise as seen from above obtained by intersecting the plane z = 2 + y and the cylinder x2 + y 2 = 1. Solution: its possible to evaluate I directly as a line integral, but the calculations are easier if Stokes theorem is used to replace the line integral with a surface integral. As shown to the right, a natural choice of surface S is the purple portion of the plane z = 2 + y lying inside the cylinder x2 + y 2 = 1. Then S is parametrized by curl F = curl (3y 2 i + 2x j + z 2 k) = (2 6y) k , (x, y) = x i + y j + (2 + y) j, 0 x2 + y 2 1. so Since x y = j + k , the normal vector to S associated with this parametrization is the upward normal, and the boundary orientiation on S (= C) is the given counter-clockwise orientation on C. Thus Stokes theorem applies, showing that I =
x2 +y 2 1
I =
x2 +y 2 1
(curl F )
(x y ) dxdy .
But
(curl F ) Thus
(x y )
= 2 6y.
(2 6y) dxdy .
curl F dS
when F = xyz i + xy 2 j + x2 yz k and S consists of the top and four sides (but not the bottom) of the cube having vertices (1, 1, 1), oriented outwards, as shown to the right. Solution: since S is an open cube with missing bottom face, the boundary orientation on S is the one shown in red to the right when S is oriented outwards. By Stokes Theorem, I = curl F dS =
S S
ds .
(does this specify k as unit normal on T ?), while curl F = curl (xyz i + xy 2 j + x2 yz k) = x2 z i + (xy 2xyz) j + (y 2 xz) k . In this case, curl F
T
But the bottom face - call it T - will have the same red boundary as the open cube and it will have the same boundary orientation provided T is oriented upwards, i.e., the unit normal on T is k. With this orientation on T , F S ds =
S
ds . T dS .
dS T
(y 2 + x) dxdy
1 1
= curl F
T
1 [ 4 1 2 ]1 2 dy = 2y 2 dy = . y x+ x 2 3 1 1
The point is that the surface integral over T is much simpler than the one over S.