Vector eld theory

Scalar elds. Vector elds

Throughout this section, D R3 will be a domain. Denition 1.1 A scalar eld in D is a real-valued function dened in D, that is : D R. The value of at an arbitrary point M (x, y, z) D is denoted by (M ) or (x, y, z) .

Denition 1.2 A vector eld in D is a vector-valued function v dened in D, that is v associates to each point M (x, y, z) D, a vector denoted by v (M ) or v (x, y, z) .

Remark 1.1 Observe that for every vector eld v in D, there are three scalar elds P , Q, R in D such that v (x, y, z) = P (x, y, z) i + Q (x, y, z) j + R (x, y, z) k, Example 1.1 (i) The position vector eld is r = xi + y j + z k, (x, y, z) R3 . (x, y, z) D.

(ii) The modulus (or the length) of r is the following scalar eld:

r = |r| =

x2 + y 2 + z 2 ,

(x, y, z) R3 .

Denition 1.3 Let be a scalar eld in D. The level surface of at the level c R is the following set: {M (x, y, z) D : (x, y, z) = c} . The level surface of through the point M0 (x0 , y0 , z0 ) D is the following set: {M (x, y, z) D : (x, y, z) = (x0 , y0 , z0 )} . Example 1.2 When is the temperature or pressure eld, the level surfaces of are called isotherms or isobars, respectively.

Denition 1.4 Let v be a vector eld in D. A eld line of v is a curve D such that for all M , the vector v (M ) is tangent to at the point M . Remark 1.2 Let v (x, y, z) = P (x, y, z) i + Q (x, y, z) j + R (x, y, z) k, (x, y, z) D.

Since dr = dx i + dy j + dz k, the condition for v to be tangent to at each point M (x, y, z) can be written as v dr v dr = 0 dy dz dx = = . P (x, y, z) Q (x, y, z) R (x, y, z) (1)

Formula (1) is a dierential system under symmetric form and is called the dierential system of the eld lines of v. Denote by 1 (x, y, z) = c1 , 2 (x, y, z) = c2 , c1 R, c2 R,

two functionally independent rst integrals of the symmetric system (1). Then, the set of the eld lines of v is the following family of curves in space which depends on two parameters c1 , c2 R: 1 (x, y, z) = c1 , c1 ,c2 : 2 (x, y, z) = c2 . Example 1.3 Find the eld lines of the following vector eld: v = (y z) i + (z x) j + (x y) k. Answer. The eld lines of v are the following:

c1 ,c2 :

x + y + z = c1 , x 2 + y 2 + z 2 = c2 ,

c1 , c2 R.

They are circles situated on spherical surfaces centered at O (0, 0, 0). Denition 1.5 Let v be a vector eld in D. A eld surface of v is a surface generated by the eld lines of v. Remark 1.3 Let v = P i + Qj + Rk. Let 1 (x, y, z) = c1 , 2 (x, y, z) = c2 , c1 R, c2 R,

be two functionally independent rst integrals of system (1). Then, the equation of a eld surface of v is (1 (x, y, z) , 2 (x, y, z)) = 0, where : R2 R is an arbitrary function.

Dierential operators

Throughout this section, D R3 will be a domain. Denition 2.1 Let be a scalar eld of class C 1 in D. Let s = s1 i + s2 j + s3 k be a unit vector, i.e. a vector having modulus 1:

|s| =

s2 + s2 + s2 = 1. 1 2 3

The directional derivative of along the unit vector s is d = s1 + s2 + s3 . ds x y z Denition 2.2 Let v be a vector eld of class C 1 in D. Let s = s1 i + s2 j + s3 k be a unit vector. The directional derivative of v along the unit vector s is dv v v v = s1 + s2 + s3 . ds x y z Remark 2.1 If v = P i + Qj + Rk, then Q R P v i+ j+ k, = x x x x v Q R P i+ j+ k, = y y y y v P Q R = i+ j+ k. z z z z Consequently, dv dP dQ dR = i+ j+ k. ds ds ds ds Denition 2.3 Let be a scalar eld of class C 1 in D. The gradient of is the vector eld in D i+ j+ k. grad = x y z Proposition 2.1 (Properties of the gradient) Let , be two scalar elds of class C 1 in D, let c R be a constant and let F : R R be a function of class C 1 . We have: (i) grad ( + ) = grad + grad ; (ii) grad (c) = c grad ; (iii) grad () = grad + grad ; (iv) grad

grad grad ; 2

(v) grad (F ()) = F () grad .

Example 2.1 Let r = xi+y j +z k be the position vector eld and let a = a1 i+a2 j +a3 k be a constant vector. Remember that

r = |r| = x2 + y 2 + z 2 , a r = a1 x + a2 y + a3 z. Then, we have

grad r =

Denition 2.4 Let v = P i + Qj + Rk be a vector eld of class C 1 in D. The divergence of v is the scalar eld in D div v = The curl of v is the vector eld in D
        

Q R P + + . x y z

curl v =

i x P

j y Q

k z R

R Q P R Q P i+ j+ y z z x x y

k.

Proposition 2.2 (Properties of the divergence and curl) Let u, v be two vector elds of class C 1 in D, let be a scalar eld of class C 1 in D and let c R be a constant. We have: (i) (ii) (iii) div (u + v) = div u + div v, curl (u + v) = curl u + curl v; div (cv) = c div v, curl (cv) = c curl v; div (v) = grad v + div v, curl (v) = curl v v grad ; div (u v) = v curl u u curl v, du dv . curl (u v) = u div v v div u + dv du

(iv)

Example 2.2 Let r = xi+y j +z k be the position vector eld and let a = a1 i+a2 j +a3 k be a constant vector. Remember that i j k a r = a1 a2 a3 x y z Then, we have div r = 3, curl r = 0, div (a r) = 0, curl (a r) = 2a. 4
             

= (a2 z a3 y) i + (a3 x a1 z) j + (a1 y a2 x) k.

Denition 2.5 We call the nabla or del or Hamilton operator the vector dierential operator k. i+ j+ = x y z Proposition 2.3 (Properties of ) Let be a scalar eld of class C 1 in D, let v be a vector eld of class C 1 in D and let s be a unit vector. We have: (i) grad = (ii) div v = (iii) curl v = (iv) (v) d = (s ds dv = (s ds ; v; v; ) ; ) v.

Denition 2.6 We call the Laplace operator or the Laplacian the dierential operator = 2 2 2 + 2 + 2. x2 y z

Remark 2.2 If is a scalar eld of class C 2 in D and v = P i + Qj + Rk is a vector eld of class C 2 in D, then = 2 2 2 + 2 + 2, x2 y z 2 2 v v 2v v = + + , x2 y 2 z 2 2P 2Q 2R 2P 2Q 2R 2P 2Q 2R = i+ j+ k+ i+ j+ k+ i+ j+ k x2 x2 x2 y 2 y 2 y 2 z 2 z 2 z 2 = P i + Q j + R k. Proposition 2.4 Let be a scalar eld of class C 2 in D and let v be a vector eld of class C 2 in D. We have: (i) div (curl v) = 0; (ii) curl (grad ) = 0; (iii) div (grad ) = ; (iv) curl (curl v) = grad (div v) v.

Integral formulas

Throughout this section, D R3 will be a domain. Denition 3.1 Let v be a continuous vector eld in D and let D be a smooth surface in space. The ow (or the ux ) of v through the surface is the number
ZZ

v n d,

where n is a unit normal (vector) eld to the surface . Remark 3.1 If v = P i + Qj + Rk and n = n1 i + n2 j + n3 k, then the ow can be expressed as follows:
ZZ ZZ

(P n1 + Qn2 + Rn3 ) d =

P dydz + Qdzdx + Rdxdy.

Thus, the ow is a surface integral of the rst or second kind. Remark 3.2 In mechanics, the ow could represent (for example) the amount of uid owing through the surface per unit time. Denition 3.2 Let v be a continuous vector eld in D and let D be a smooth curve in space. The circulation of v along the curve is the number
Z

C=

v dr.

Remark 3.3 If v = P i+Qj +Rk, then, taking into account that dr = dx i+dy j +dz k, we infer that Z C = P dx + Qdy + Rdz.

Thus, the circulation is line integral of the second kind. Remark 3.4 In mechanics, the circulation C could represent (for example) the mechanical work done by a force eld on a particle moving along the curve . Theorem 3.1 (Divergence theorem) Let D R3 be a bounded smooth domain, let be the closed surface which is the boundary of D, let n be the outer unit normal (vector) eld to the surface and let v be a vector eld of class C 1 on D . Then,
ZZ ZZZ

v n d =
D

div v d.

Remark 3.5 If v = P i + Qj + Rk, then the right-hand side of the above formula can be written as ZZZ ZZZ Q R P + + dxdydz, div v d = D D x y z which is a volume integral.

Theorem 3.2 (Divergence theorem for the gradient) Let D R3 be a bounded smooth domain, let be the closed surface which is the boundary of D, let n be the outer unit normal (vector) eld to the surface and let be a scalar eld of class C 1 on D . Then, ZZ ZZZ n d = grad d.
D

Theorem 3.3 (Divergence theorem for the curl) Let D R3 be a bounded smooth domain, let be the closed surface which is the boundary of D, let n be the outer unit normal (vector) eld to the surface and let v be a vector eld of class C 1 on D . Then, ZZ ZZZ n v d = curl v d.
D

Theorem 3.4 (Stokes formula) Let R3 be a bounded smooth open surface, let be the closed curve which is the boundary of , let n be a unit normal (vector) eld to the surface and let v be a vector eld of class C 1 in a domain D R3 which contains the surface . Then Z ZZ v dr = curl v n d.

Special vector elds

Throughout this section, D R3 will be a domain.

4.1

Irrotational vector elds

Denition 4.1 Let v be a vector eld of class C 1 in D. We say that v is irrotational in D if curl v = 0 in D. Denition 4.2 Let v be a vector eld in D. We say that v is conservative in D if there is a scalar eld of class C 1 in D such that grad = v In this case, is called a scalar potential of v. Remark 4.1 A scalar potential of a conservative vector eld v is unique up to an additive constant, that is grad = v grad ( + c) = v and grad = grad = v c = const R s.t. = in D. Theorem 4.1 Let D R3 be a simply connected domain and let v be a vector eld of class C 1 in D. (i) If v is irrotational in D, then the circulation of v along any closed smooth curve D is zero. 7 c = const R, in D.

 (ii) If v is irrotational in D, then the circulation of v along any smooth path AB D depends only on A and B, and does not depend on the path between A and B.

(iii) The vector eld v is irrotational in D if and only if v is conservative in D. In this case, a scalar potential of v is
Z

(M ) = where A D is a xed point.

AM

v dr,

M D,

4.2

Solenoidal vector elds

Denition 4.3 Let v be a vector eld of class C 1 in D. We say that v is solenoidal in D if div v = 0 in D. Denition 4.4 Let v be a vector eld in D. We say that v is a curl eld in D if there is a vector eld w of class C 1 in D such that curl w = v In this case, w is called a vector potential of v. Remark 4.2 Since curl (grad ) = 0 , we have curl w = v curl (w + grad ) = v Theorem 4.2 Let v be a vector eld of class C 1 in D. (i) If v is solenoidal in D, then the ow of v through any closed smooth surface D is zero. (ii) If v is solenoidal in D, then the ow of v through any open smooth surface D depends only on the boundary of , and does not depend on the surface . (iii) The vector eld v is solenoidal in D if and only if v is a curl eld in D. . in D.

4.3

Vector elds represented by two scalar elds

Denition 4.5 Let v be a vector eld of class C 1 in D. We say that v is represented by two scalar elds in D if there are two scalar elds of class C 1 in D, and F of class C 2 in D such that v = grad F in D. Theorem 4.3 Let v be a vector eld of class C 1 in D. (i) If v is represented by two scalar elds in D, then v curl v = 0 in D, i.e. v curl v. 8

(ii) If v curl v = 0 in D, curl v = 0 in D, then v is represented by two scalar elds in D.

4.4

Laplacian (or harmonic) vector elds

Denition 4.6 Let v be a vector eld of class C 1 in D. We say that v is Laplacian or harmonic in D if v is both irrotational and solenoidal in D, i.e. curl v = 0 in D, div v = 0 in D. Example 4.1 The Newtonian vector eld v = k is Laplacian in R3 \ {(0, 0, 0)}. Denition 4.7 Let : D R be a function of class C 2 in D. We say that is harmonic in D if satises the Laplace equation in D, that is = 0 in D. Remark 4.3 Remember that =
def

r r3

(k = const > 0) ,

2 2 2 + 2 + 2. x2 y z

Theorem 4.4 Let D R3 be a simply connected domain and let v be a vector eld of class C 1 in D. The following statements are equivalent: (i) The vector eld v is Laplacian in D. (ii) There is a harmonic function in D such that v = grad in D.