This action might not be possible to undo. Are you sure you want to continue?

**Vector and Dyadic Analysis
**

I am greatly astonished when I consider the weakness of my mind and its proness to error. —Descartes This appendix summarizes a number of useful relationships and transformations from vector calculus and dyadic analysis that are especially relevant to electromagnetic theory.

A.1 COORDINATE SYSTEMS

A.1.1

A d`

Rectangular Coordinates: (x; y; z)

= =

z

A xx + A y y + A z z ^ ^ ^ x dx + y dy + ^ dz ^ ^ z

z = z1 plane ^ z ^ x ^ y P(x1, y1 , z 1) O

dS dV

= dy dz x + dx dz y + dx dy z ^ ^ ^ = dx dy dz

z

dsx = dy dz ds = dx dy z dz z O y y x dx x y = y1 plane dy y dsy = dx dz

x

x = x1 plane

1

2

VECTOR AND DYADIC ANALYSIS

A.1.2

A d`

Cylindrical Coordinates:

= = ^ A½ ½ + AÁ Á + Az z ^ ^ ^ ½ dÁ + z dz ½ d½ + Á ^ ^ dS dV

(½; Á; z)

^ = ½ dÁ dz ½ + d½ dz Á + ½ d½ dÁ z ^ ^ = ½ d½ dÁ dz

z

z = z1 plane

z

^ z ^ φ dsφ = dρ dz O φ y φ1 φ = φ1 plane x ρ

dsz = ρ dρ dφ dsρ = ρ dφ dz dz z

z1 O ρ = ρ1 cylinder x ρ1

^ ρ

dρ

dφ

y

A.1.3

A = d` =

Spherical Coordinates: (r; µ; Á)

^ ^ Ar ^ + Aµ µ + AÁ Á r ^ ^ r dr + µ rdµ + Á r sin µ dÁ ^ dS dV ^ ^ = r 2 sin µ dµ dÁ r + r sin µ dr dÁ µ + r dr dµ Á ^ = r 2 sin µ dr dµ dÁ

θ = θ1 cone

z z θ1 ^ r ^ φ r1 O φ1 y x φ dφ y ^ θ O dsr = r 2sin θ dθ dφ ds φ = r dr dθ θ r

dθ

dsθ = r sinθ dr dθ

x r = r1 sphere φ = φ1 plane

3 Cylindrical Ã Spherical transformation: ! ½ = r sin µ These coordinates are related by .6) A.2. Ar ) r ).2 Rectangular Ã Cylindrical transformation: ! x = ½ cos Á y = ½ sin Á z = z (A.COORDINATE TRANSFORMATIONS 3 A.1 Rectangular Ã Spherical transformation: ! x = r sin µ cos Á y = r sin µ sin Á z = r cos µ (A.1) These coordinates are related by and conversion between vectors is given by Spherical ! Rectangular Ax Ay Az Ar Aµ AÁ = Ar sin µ cos Á + Aµ cos µ cos Á ¡ AÁ sin Á = Ar sin µ sin Á + Aµ cos µ sin Á + AÁ cos Á = Ar cos µ ¡ Aµ sin µ (A.2 COORDINATE TRANSFORMATIONS Note that relations between the unit vectors in the different coordinate systems are obtained from the following by replacing the components of A with the corresponding unit vector (for example.3) Rectangular ! Spherical A.2. ^ A.2) = Ax sin µ cos Á + Ay sin µ sin Á + Az cos µ = Ax cos µ cos Á + Ay cos µ sin Á ¡ Az sin µ = ¡Ax sin Á + Ay cos Á (A.4) These coordinates are related by and conversion between vectors is given by Cylindrical ! Rectangular Ax Ay Az A½ AÁ Az = A½ cos Á ¡ AÁ sin Á = A½ sin Á + AÁ cos Á = Az = Ax cos Á + Ay sin Á = ¡Ax sin Á + Ay cos Á = Az (A.2.5) Rectangular ! Cylindrical (A.

13) The r operator takes on different forms in other coordinate systems.4 lists explicit divergence and curl operations in the three most common coordinate systems. which in rectangular coordinates is r´ @ @ @ x+ ^ y+ ^ z ^ @x @y @z (A.8) Spherical ! Cylindrical (A.9) A. The flux Ã of a vector field A through some surface S is defined as ZZ flux of A through S Ã´ A ¢ dS (A. Note that the concepts of flux/divergence are also related through the Divergence theorem (A. Section A.3.3 ELEMENTS OF VECTOR CALCULUS A.7) (the azimuthal angle Á is common to both coordinate systems).1 Flux and Circulation Maxwell’s equations are expressed in terms of two important vector field concepts: flux and circulation.10) S and the circulation of A around some path C is defined as I circulation of A around C = A ¢ d` C (A. Conversion between vectors is given by Cylindrical ! Spherical Ar Aµ AÁ A½ AÁ Az = = = = = = A½ sin µ + Az cos µ A½ cos µ ¡ Az sin µ AÁ Ar sin µ + AÁ cos µ AÁ Ar cos µ ¡ Aµ sin µ (A.4 VECTOR AND DYADIC ANALYSIS z = r cos µ (A.11) These concepts are expressed in differential form as the divergence and curl ZZ I ° A ¢ dS A ¢ d` S C Divergence: r ¢ A ´ lim ZZZ Curl: r £ A ´ lim ZZ (A.59).12) V !0 S!0 dV dS V S where we have defined the ‘del’ operator. and the concepts of circulation/curl are also related through the Stokes theorem (A.54). .

y. y(u. a change of variables from the coordinates (x. In three dimensions an arbitrary surface can be described by the functional relation f (r) = C (A. z) dx dy dz = g(u. Á.16) A. The gradient only operates on scalar fields. z). y. This property is useful in a geometric sense for determining tangent planes and normal directions to an arbitrary surface [2]. z) to new coordinates (u.2 The Gradient Another important operation is the gradient of a vector field. v.3.17) A. w) is given by [2] ¯ ¯ ZZZ ZZZ ¯ @(x.3 Vector Taylor Expansion The multi-dimensional Taylor series expansion of a function f (r + a) around the point r can be represented in vector form as f (r + a) = 1 X 1 n (a ¢ r) f (r) n! n=0 (A. A plane tangent to this surface at the point r0 is described by (r ¡ r0 ) ¢ rf (r0 ) = 0 (tangent plane) (A. y. y. v.3. v.3. for example.4.18) @(u. w) ¯ where ¯ ¯ ¯ ¯ ¯ @(x. v. and g(u. so a unit normal to the surface described by (A.14) at the point r 0 can be found from n= ^ rf (r0 ) jrf (r0 )j (unit normal) (A. w) = f (x(u. in rectangular coordinates. v. v. f (r) = f (x. The gradient produces a vector which points in the direction of greatest change of the scalar field. Explicit forms for the gradient operation in the three common coordinate systems is given in section A. y.ELEMENTS OF VECTOR CALCULUS 5 A.15) The gradient points in the direction normal to the surface. and is written as rÁ. which is the vector equivalent of a derivative operation. w)) .14) where C is a constant. z) ¯ ¯ ¯ du dv dw f (x. and f (r) is shorthand for a function of the three coordinate variables. w). w) ¯ = ¯ @y=@u ¯ @z=@u @x=@v @y=@v @z=@v ¯ @x=@w ¯ ¯ @y=@w ¯ ¯ @z=@w ¯ is called the Jacobian of the transformation. w) ¯ (A. v. w). z) ¯ ¯ @x=@u ¯ ¯ ¯ ¯ @(u. z(u.4 Change of Variables In three dimensions. v.

28) .25) 1 @(½A½ ) 1 @AÁ @Az + + ½ @½ ½ @Á @z · ¸ 1 @Az @AÁ r£A = ½ ^ ¡ ½ @Á @z · ¸ · ¸ ^ @A½ ¡ @Az + z 1 @(½AÁ ) ¡ @A½ +Á ^ @z @½ ½ @½ @Á µ ¶ 1 @ @© 1 @2© @2 © r2 © = ½ + 2 + 2 ½ @½ @½ ½ @Á @z 2 µ ¶ 2 @AÁ A½ r2 A = ½ r2 A½ ¡ 2 ^ ¡ 2 ½ @Á ½ µ ¶ ^ r2 AÁ + 2 @A½ ¡ AÁ + ^(r 2Az ) +Á z ½2 @Á ½2 (A.4.4 EXPLICIT DIFFERENTIAL OPERATIONS A.20) @x @y @z µ ¶ µ ¶ µ ¶ @Az @Ay @Ax @Az @Ay @Ax r£A = x ^ ¡ +y ^ ¡ +z ^ ¡ (A. Á. z) can be expressed functionally in terms of (u.6 VECTOR AND DYADIC ANALYSIS and it has been assumed that (x. A similar result applies to transformations in two-dimensions.24) (A.27) (A.21) @y @z @z @x @x @y r2 © = @2© @2© @2© + + @x 2 @y 2 @z 2 (A.26) (A.2 Cylindrical Coordinates (½. A.1 Rectangular Coordinates (x.19) @Ax @Ay @Az + + (A. w) (or vica-versa). v. z): @© ^ 1 @© + z @© +Á ^ @½ ½ @Á @z r© = ½ ^ r ¢A = (A. y.23) r2 A = xr2 Ax + y r2 Ay + z r2 Az ^ ^ ^ A.4.22) (A. z): @© @© @© +y ^ +z ^ @x @y @z r© = x ^ r ¢A = (A. y.

3 Spherical Coordinates (r. Á): @© ^ 1 @© + Á 1 @© ^ +µ @r r @µ r sin µ @Á (A.29) r© = r ^ r¢A = r£A = r2 © = = r2 A = 1 @(r 2 Ar ) 1 @(Aµ sin µ) 1 @AÁ + + (A.38) (A.31) r sin µ @Á @r r @r @µ µ ¶ µ ¶ 1 @ @© 1 @ @© 1 @2 © r2 + 2 sin µ + (A.4.41) (A.5 VECTOR RELATIONS A.39) (A.32) 2 r2 @r @r r sin µ @µ @µ r2 sin µ @Á2 µ ¶ 1 @2 1 @ @© 1 @2© (r©) + 2 sin µ + 2 (A.33) 2 2 r @r r sin µ @µ @µ r sin µ @Á2 · µ ¶¸ 2 @AÁ @Aµ r r2 Ar ¡ 2 Ar + Aµ cot µ + csc µ ^ + r @Á @µ · µ ¶¸ ^ r2 Aµ ¡ 1 Aµ csc2 µ ¡ 2 @Ar + 2 cot µ csc µ @AÁ +µ r2 @µ @Á · µ ¶¸ ^ r2 AÁ ¡ 1 AÁ csc2 µ ¡ 2 csc µ @Ar ¡ 2 cot µ csc µ @Aµ +Á (A.42) (A £ B) £ (C £ D) = (A £ B ¢ D)C ¡ (A £ B ¢ C )D .30) 2 r @r r sin µ @µ r sin µ @Á · ¸ r ^ @(AÁ sin µ) @Aµ ¡ r sin µ @µ @Á · ¸ ¸ ^ ^· µ 1 @Ar @(rAÁ ) Á @(rAµ ) @Ar + ¡ + ¡ (A.40) = (A ¢ C)(B ¢ D) ¡ (A ¢ D)(B ¢ C) (A. µ.1 Dot and Cross Product Identities ¤ 2 A ¢ A = jAj A¢B=B¢A A £ B = ¡B £ A A ¢ (B £ C) = B ¢ (C £ A) = C ¢ (A £ B) A £ (B £ C) = (A ¢ C)B ¡ (A ¢ B)C A £ (B £ C) + B £ (C £ A) + C £ (A £ B) = 0 £ ¤ (A £ B) ¢ (C £ D) = A ¢ B £ (C £ D) (A.35) (A.37) (A.34) r2 @Á @Á A.5.VECTOR RELATIONS 7 A.36) (A.

53) In the following.43) (A.54) combined with (A. A.46) (A. with the direction of dS taken as pointing outward from the enclosed volume.51) (A. b b a rÁ ¢ d` = a @Á d` = Á(b) ¡ Á(a) @` (A.57) .3 Integral relations Z Z From the Fundamental Theorem of Calculus.49) r £ rÁ = 0 r ¢ (r £ A) = 0 r £ r £ A = r(r ¢ A) ¡ r2 A (A.52) The last identity essentially defines the vector Laplacian r2 A.2 Vector Differential operations r ¢ rÁ = r2 Á r(ÁÃ) = ÁrÃ + ÃrÁ r ¢ (ÁA) = A ¢ rÁ + Ár ¢ A r £ (ÁA) = Ár £ A ¡ A £ rÁ r ¢ (A £ B) = B ¢ (r £ A) ¡ A ¢ (r £ B) (A.8 VECTOR AND DYADIC ANALYSIS A.5.5.55) (A.45) (A.54) (A.44) (A. by convention: ZZZ Z Z ° A ¢ dS (Divergence theorem) (r ¢ A) dV = V S (A.50) (A.47) (A. V is a volume bounded by a closed surface S. which reduces to three scalar Laplacians in rectangular coordinates only.48) r £ (A £ B) = A(r ¢ B) ¡ B(r ¢ A) + (B ¢ r)A ¡ (A ¢ r)B r(A ¢ B) = A £ (r £ B) + B £ (r £ A) + (B ¢ r)A + (A ¢ r)B (A.56) ZZZ ZZZ V V (rÁ) dV = ° Á dS S Z Z Z Z (r £ A) dV = ° (dS £ A) S Note that (A.51) gives Z Z ° S ¡ ¢ r £ A ¢ dS = 0 (A.

64) (A.56) give Z Z ° dS £ rÁ = 0 S (A.58) In the following. The direction of dS is normal to the surface following the right-hand rule with the fingers curled in the direction of C : (Stokes theorem) ZZ (r £ A) ¢ dS = (dS £ rÁ) = I I A ¢ d` Á d` (A. and also for boundary-value problems using Green’s functions. These are often useful in proving orthogonality of eigenfunctions of the scalar and vector wave equations.59) give I rÁ ¢ d` = 0 (A. The direction of d` is tangent to C. For two scalar functions Á and Ã.59) (A.50) and (A.61) C Green’s identities and theorems provide additional relations between surface and volume integrals.VECTOR RELATIONS 9 and that (A. .50) and (A.60) S C ZZ S C Note that (A. we have (Green’s first identity) ZZZ Z Z (rÁ ¢ rÃ + Ár2 Ã) dV = ° ÁrÃ ¢ dS S (A.65) V V Z Z ¢ B ¢ r £ r £ A ¡ A ¢ r £ r £ B dV = ° (A £ r £ B ¡ B £ r £ A) ¢ dS which also require that A and B are continuous through the second derivatives.62) V Interchanging Á and Ã and subtracting gives (Green’s theorem) ZZZ Z Z (Ár2 Ã ¡ Ãr2 Á) dV = ° (ÁrÃ ¡ ÃrÁ) ¢ dS S V (A. S is an open surface bounded by a contour C described by line element d`.63) The vector forms of Green’s identity and Green’s theorem are ZZZ ZZZ ¡ ¡ Z Z ¢ r £ A ¢ r £ B ¡ A ¢ r £ r £ B dV = ° (A £ r £ B) ¢ dS S S (A. which are continuous through the second derivatives in the volume V .

67) (A.72) (from A.72) (A.77) .68) (from A.4 Distance Vector Identities Let R be the position vector defined by the two points r and r0 as shown below.74) (A. where r operates on unprimed coordinates: ^ rR = R ^ r (1=R) = ¡ R=R 2 = ¡R=R3 r ¢R = 3 ^ r ¢ R = 2=R ^ r£R = 0 r£R = 0 (from A. a is any constant vector: r ¢ (a=R) = a ¢ r(1=R) = ¡a ¢ n=R2 ^ r2 (a=R) = ar2 (1=R) = ¡4¼a±(R) £ ¤ £ ¤ r £ a £ (^ =R2 ) = 4¼a±(R) ¡ r (a ¢ n)=R2 n ^ i 1 h ^ ^ ^ (a ¢ r)R = a ¡ R(a ¢ R) R (a ¢ r)R = a (A.r' R=r r' r y x Then the following relations hold. z .71) (A.67 and A.66 and A.77) (A.5.69) (A. Also ^ ^ define R = R R.73) r2 (1=R) = ¡4¼±(R) r ¢ (R=R3 ) = 4¼±(R) In the following.78) (from A.75) (A.66 and A.67 and A.66) (from A.10 VECTOR AND DYADIC ANALYSIS A.68) (A.76) (A.70) (A.70) (A. where R is the distance between the points and R is the unit vector in the direction of R.66) (A.

79) can be written as ZZZ ZZZ r0 ¢ A(r0 ) r0 £ A(r0 ) A(r) = ¡r dV 0 + r £ dV 0 0 4¼jr ¡ r j 4¼jr ¡ r0 j (A.80a) (A.83) 0 0 V 4¼jr ¡ r j S 4¼jr ¡ r j where S is the surface enclosing the volume V .80). can always be added to » without changing (A. then r £ r £ » = r(r ¢ ») ¡ r2 » = ¡r2 » r2 ' = r ¢ A 2 and (A. then the solutions to (A. Note that in order to uniquely determine A. This essentially proves the theorem. This is the formal statement of the Helmholtz theorem.82) If the field is to be represented in a bounded region.80) become (A.79).81a) (A. rÁ. We can pick any function Á that is convenient.80).VECTOR RELATIONS 11 A.5. one which has zero divergence (the solenoidal or rotational part) and one with zero curl (the lamellar. that is. take the divergence and curl of (A. »).5 The Helmholtz theorem The Helmholtz theorem [3] states that a vector function A(r) can be expressed as the sum of two vector functions. we know these have the solution (for unbounded regions) ZZZ ZZZ r0 ¢ A(r0 ) r0 £ A(r 0) '(r) = ¡ dV 0 » (r) = dV 0 0 4¼jr ¡ r j 4¼jr ¡ r0 j and so (A.81) must be modified accordingly. if Á is chosen such that r ¢ » = 0.80b) Since A is assumed known. µ ZZZ ¶ Z Z r 0 ¢ A(r0 ) A(r0 ) ¢ dS 0 0 ° A(r) = ¡r 0 dV ¡ 0 V 4¼jr ¡ r j S 4¼jr ¡ r j µZZZ Z Z ¶ r0 £ A(r 0 ) A(r 0 ) £ dS 0 + r£ dV 0 + ° (A. and it can be shown that the representation is. since the gradient of an arbitrary scalar function. this is an alternative statement of the Helmholtz theorem. these two differential equations are uncoupled and can (in principle) be solved independently for the pair of functions ('. which gives r£r£» = r£ A r2 ' = r ¢ A (A. There are an infinite number of possible functions » which can be used to uniquely determine A. so is » + rÁ. or irrotational part). A(r) = r £ » + r' (A. more generally.81b) r » = ¡r £ A From electrostatics. if » is a solution of (A.79) To show that such a decomposition is possible. that is. . both it’s divergence and curl must be specified.

84) longitudinal rt £ At | {z } (A.85) In the following.5.88) (A.90) (A. while the normal to C is described by n.86) (A.92) (A. ^ ZZ I (2D Divergence theorem) (rt ¢ A) dS = A ¢ n d` ^ (A.94) (A.98) S C .97) @ ¡^ £ (r tAz ) + z (^ £ At) z | {z @z } + At £ B t | {z } (A.93) (A. Let the subscript t represent vector components that are transverse to z . S is an open surface bounded by a contour C described by line element d`.95) (A. so that: ^ r = rt + ^ z @ @z A = At + z Az ^ Transverse and longitudinal components of other common operations can then be similarly decomposed A£B = Az (^ £ B t ) ¡ Bz (^ £ At ) z z | {z } tr ansverse tr ansverse + longitudinal r£A = We use the earlier vector relations in three dimensions to prove the following identities: rt ¢ rt Á = r 2 Á t rt ¢ (^ £ rt Á) = 0 z rt £ rt Á = 0 z £ (^ £ rt Á) = ¡rt Á ^ z rt £ (^ £ rt Á) = z r2 Á z ^ t z £ (^ £ At ) = ¡At ^ z (^ £ At) ¢ (^ £ B t ) = At ¢ Bt z z At £ (^ £ B t ) = z (At ¢ Bt ) z ^ rt £ (^ £ At ) = z (rt ¢ At) z ^ (^ £ At ) £ (^ £ B t ) = At £ B t z z rt ¢ (^ £ At ) = ¡ z ¢ (rt £ B t ) z ^ At ¢ (^ £ B t ) = ¡ z ¢ (At £ B t ) z ^ (A.96) (A.91) (A.12 VECTOR AND DYADIC ANALYSIS A.89) (A.6 Useful Vector Relations in Two-Dimensions Situations arise where one dimension (usually taken as z ) can be factored out of the ^ analysis.87) (A. The direction of d` is tangent to C.

centered at the origin of a spherical coordinate system. In doing so.100) @n @n A.60) and Green’s theorem (A. The solid-angle is a unitless concept. centered at the origin. In the figure above.= 4¼a2 which is interpreted as meaning that the entire closed surface of the sphere subtends a total solid angle of 4¼. which from the . and integrating over the surface of the sphere gives Z Z ZZ ° dA = a2 ° d. It is sometimes convenient to view this element of surface area as subtending a “solid angle”.5.7 Solid Angle An element of surface area for a sphere of radius a. we write dA = a2 d-. That is. z dA dS a r dΩ ˆ r ˆ n S x y This concept can be extended to any arbitrary surface S by forming the projection of each surface element dS onto a sphere. so that the angular integration in µ and Á is replaced by an integration over the range of “solid angles” subtended by the surface. both dA and dS subtend the same solid angle d-. dA is the projection of the surface element dS along the radial direction onto a sphere of radius a.99) S C (2D Green’s theorem) S I · C ¸ @Ã @Á Á ¡Ã d` (A. but it is conventionally given the dimensionless units of steradians. is given by dA = a2 sin µ dµ dÁ.61) generalize to two dimensions as follows: (2D Green’s identity) ZZ ZZ (rt Á ¢ rtÃ + Ár2 Ã) dS = t (Ár2 Ã t ¡ Ãr2 Á) dS t = I Á @Ã d` @n (A.VECTOR RELATIONS 13 Green’s identity (A. d-.

this result takes the more general form ZZ ½ ^ R ¢ dS 4¼ ° = 2 0 R S r0 inside S r0 outside S (A. then the surface no longer subtends a total solid angle of 4¼. The projection dA is found by taking the dot product of dS = ndS with the radial unit vector.67) note that d. a mathematical description of the charge density must have the properties ZZZ ½(r) = 0 for r 6= 0 and ½(r) dV = q . ½ and J . From (A. In the context of Maxwell’s equations. the actual size of the electron is neglible. consider the charge density of a single electron—how do we represent such a thing? From a macroscopic point of view. For an electron located at r = 0 with charge q. such difficulties can arise from our description of charge and current distributions as density functions.= ¡r(1=r) ¢ dS From the divergence theorem.72).= ¡ r (1=r) dV = 4¼ ±(r) dV where the last equality follows from (A. where r is the distance to the surface element. This is essentially Gauss’ law.101) H It is important to note that the result d. A. respectively. Mathematically this can be seen as follows. Shifting the coordinate system by r 0 .14 VECTOR AND DYADIC ANALYSIS discussion above is defined as d. The last integral is zero unless the volume bounded by S contains the point r = 0. and acounting for it would unnecessarily complicate the mathematics. For example.= ° a2 r2 Z Z r ¢ dS ^ = 4¼ r2 (A. Clearly if the origin were outside of the surface S.= dA=a2 . and scaling the result by a factor of ^ a2 =r2 . ZZ ZZZ ZZZ 2 ° d.102) where R = r ¡ r 0 . dA = ^ ¢ dS r and therefore d.6 DIRAC DELTA FUNCTIONS Dirac delta functions are a convenient mathematical shorthand that are used to help us out of difficult situations.= r ^ ¢ dS r2 and Z Z ° d.= 4¼ is critically dependent on having chosen the surface enclose the origin of the coordinate system.

z) = I0 ±(x)±(y)^ z Although the current density so defined is singular at x = y = 0. consider a current I0 flowing along a thin wire colinear with the z-axis. the integral over the cross section of the wire will remain finite and provide the correct answer ZZ ZZ I = J ¢ dS = I0 ±(x)±(y) dx dy = I0 These examples also illustrate that the delta function must have units. written as ±(x).104) 0 otherwise a where the range of integration is taken over all values of x. We define ±(r ¡ r0 ) by the properties ZZZ n 0 0 0 ±(r ¡ r ) = 0 for r 6= r and ±(r ¡ r0 ) dV = 1 if r in V (A. Using the delta function. f (x). is defined to have similar properties. and therefore ±(r ¡ r0 ) can be represented as a product of three one dimensional delta functions rectangular: ±(r ¡ r0 ) = ±(x ¡ x 0 )±(y ¡ y 0)±(z ¡ z 0 ) (A. Examining the . we can represent the corresponding current density as J (x. Z b n 1 a · x0 · b ±(x ¡ x0 ) = 0 for x 6= x 0 and ±(x ¡ x0 ) dx = (A. then the only contribution to the integral will come from this point. we find that the charge density function associated with a point charge at r0 can now be represented concisely as ½(r) = q±(r ¡ r0 ) As another example. The extension to three dimensions is straightforward. This is called the “sifting” property of the delta function.106) In rectangular coordinates.107) Returning to our original example. ie.105) 0 otherwise V which in turn lead to the sifting property ZZZ n f(r0 ) if r0 in V f (r)±(r ¡ r0 ) dV = 0 otherwise V (A. A delta function in one dimension. y. then ±(x) has the units of inverse length.103) 0 otherwise a The most important property of the delta function follows from the above definition and involves its appearance in an integrand with another ordinary function. If x represents a physical length dimension. dV = dx dy dz. and we get (in one dimension) Z b n f (x)±(x ¡ x 0 )dx = f (x0 ) a · x 0 · b (A.DIRAC DELTA FUNCTIONS 15 where the integral is taken over the region contining the charge. at least in rectangular coordinates. As long as f is continuous at the location of the delta function singularity.

108) Using this we find. and ±(r ¡ r0 ) = (A. we use the change of variable theorem of the previous section and note that the volume element in the new coordinate system is given by jJ jdu dv dw. points on the z-axis (corresponding to µ0 = 0 or µ 0 = ¼) are represented by ±(r ¡ r0 ) = 1 ±(r ¡ r 0)±(µ ¡ µ 0) 2¼r2 sin µ ±(r) 4¼r 2 (A. with this association of units.112) For points at the origin.114) jaj . and so the delta function must be represented somewhat differently in each case. we see that it has the units of inverse volume. For example.111) One can always check the validity of a delta function representation using the integral properties defined above. w). we see that the correct units of £ ¤ £ ¤ C=m3 and A=m2 are obtained. corresponding to the singularities of the Jacobian. in cylindrical coordinates if the delta function is located on the z-axis. From the sifting property of the three dimensional delta function. for cylindrical coordinates cylindrical: and for spherical coordinates spherical: ±(r ¡ r0 ) = (A. v. v. the differential element of volume takes different forms in different coordinate systems. however.110) ±(r ¡ r0 ) = (A.109) There are situations where this approach breaks down. where jJj is the Jacobian of the transformation. Similarly in spherical coordinates. respectively.113) Having shown how the three-dimensional delta function can be represented by products of one-dimensional delta functions. both µ and Á are irrelevant. In three dimensions. Therefore a representation for the delta function is ±(r ¡ r0 ) = 1 ±(u ¡ u0 )±(v ¡ v 0)±(w ¡ w 0 ) jJ j 1 ±(½ ¡ ½0)±(Á ¡ Á0 )±(z ¡ z0 ) ½ ±(r ¡ r0 )±(µ ¡ µ 0 )±(Á ¡ Á0 ) r0 2 sin µ0 (A.107) to some other set of coordinates (u. and the representation is instead [4] ±(r ¡ r0 ) = 1 ±(½)±(z ¡ z0 ) 2¼½ (A. we now list some additional properties of the latter that are useful in electromagnetic analysis: 1 ±(ax ¡ b) = ±(x ¡ b=a) (A.16 VECTOR AND DYADIC ANALYSIS expressions for the charge and current density above. [1=dV ]. w) is irrelevant in the transformation. the azimuthal angle Á does not appear in the transformation. This occurs when the delta function peak is located such that one of the variables (u. To transform from the representation in rectangular coordinates (A.

A. ¾xz describes the x ^ -component of current flow due to the z -component of the electric field. A more general linear transformation would allow each component of E to influence each component of J . We could write this in matrix form as 2 3 2 32 3 Jx ¾xx ¾ xy ¾xz Ex 4 Jy 5 = 4 ¾ yx ¾yy ¾yz 5 4 E y 5 Jz ¾ zx ¾z y ¾ zz Ez The matrix [¾] is referred to as a second-rank tensor. and the summation is over all the possible zeroes. In matrix form.118) where xi are the zeroes of f (x).115) (A.7 DYADIC ANALYSIS In elementary vector analysis we frequently encounter scalar relationships between two vectors. which is why tensor relationships are rarely covered in elementary electromagnetics courses! . 2 3 2 3 Jx Ex 4 J y 5 = ¾ 4 Ey 5 Jz Ez This is a very simple relationship which takes the vector quantity E and scales each component by the number ¾ to give a new vector.117) ±(x ¡ a)±(x ¡ b) dx = ±(a ¡ b) f (x)±0 (x ¡ a) dx = ¡f 0 (a) where in the last relation the prime denotes a derivative with respect to the argument. Each component of the tensor describes the influence of one field quantity on another. or wave propagation in a plasma.116) (A.DYADIC ANALYSIS 17 ±(x 2 ¡ a2 ) = Z Z 1 [±(x ¡ a) + ±(x + a)] 2a (A. such as current flow in an anisotropic crystal. such as in Ohm’s law. which consequently retains the original direction of E. so that the transformation changes the direction as well as the magnitude (ie. ie. Another useful transformation is given by ± (f (x)) = X ±(x ¡ xi ) jdf (x i)=dxj i (A. where ¾ is a scalar quantity. Naturally the mathematics becomes more complicated. Such tensor ^ relationships arise in many physical contexts. J. for example. J = ¾E . involves a rotation in addition to a scaling). A more exhaustive collection of delta function properties relevant to electromagnetic theory is found in [4]. f (x i) = 0.

we can often ascribe an independent physical significance to dyads such as ¾. we can see that interchanging the order of the dot product in (A. ¾ ¢ E 6= E ¢ ¾. For example. but ^^ x interchanging the order of the unit vectors clearly gives a different result.36)-(A.18 VECTOR AND DYADIC ANALYSIS The tensor relationship can be written in a different way using vector notation as J =¾¢E where ¾ is defined as ¾ = ¾xx x x + ¾xy ^ y + ¾ xz x ^ + ^^ x^ ^z ¾yx y x + ¾yy y y + ¾yz y z + ^^ ^^ ^^ ¾z x ^x + ¾ zy ^y + ¾z z zz z^ z^ ^^ The only new feature is the appearance of products of unit vectors. in this case the “conductivity” dyadic. Clearly there is a close relationship between dyads and second-rank tensors. x y ¢ E = ^ E y .37) where ordering of the vectors is not as significant. For example. ie. P =X Y (A. a dyad only has meaning when it acts upon another vector. As noted above.36)-(A. provided we strictly obey the order of the unit vectors and the dot product. . ¾ T = ¾ xxx x + ¾yx y x + ¾z xz x + ^^ ^^ ^^ ¾ xy x ^ + ¾yy ^y + ¾z y z y + ^y y^ ^^ ¾ xz xz + ¾ yz y z + ¾z z z ^ ^^ ^^ ^z Consequently. Since the components of ¾ are characterized by pairs of unit vectors. or a dyadic quantity (the word “dyad” means pair).119) also markedly affects the result. dyad-vector multiplications do not obey the familiar vector commutation rules (A. which are derived assuming the simpler vector commutation laws (A.119) To the extent that the vector fields represent (or can be related to) physically meaningful quantities.37). y x ¢ E = y Ex .120) (A. A dyad or dyadic operator is expressable as the algebraic product of two vectors or vector operators. This is not surprising given the obvious similarity between this new quantity ¾ and an ordinary matrix. but obey instead the matrix-like commutative laws P ¢ A = A ¢PT ³ ´T P £A = ¡ A£PT where the superscript T suggests a matrix-like transpose operation.5. it is called a dyad. This definition gives the correct result using the normal rules for the vector dot product. much like a matrix can be formed from the product of two vectors. ^^ ^ Similarly. one must resist the temptation to use dyads in place of vectors in the vector identities of section A. However.

7. where rr is a dyadic operator.122) (A. defined such that A¢I =I¢A =A In rectangular coordinates. the function J ¡ (^ ¢ J)^ which appears in the radiation integrals can be represented by (I ¡ ^r) ¢ J r r r^ A.129) (A.133) (A.130) (A. In electromagnetic theory.125) (A. Ordinarily this expression is understood to mean r(r ¢ A).135) (A.1 Dyadic Dot and Cross Product Identities A ¢ P = PT ¢ A ³ ´T A£P = ¡P T £ A ³ ´ ³ ´ A ¢P ¢B = A ¢P ¢B = A ¢ P ¢B (A. Once the reader becomes familiar with the notation.124) (A.134) (A. I.131) (A. Similarly.136) A ¢P ¢B = B ¢PT ¢ A ³ ´ ³ ´ ¡ ¢ A £ B ¢ P = A ¢ B £ P = ¡B ¢ A £ P ³ ´ ³ ´ ¡ ¢ P ¢ A£B = ¡ P £B ¢A = P £A ¢B ³ ´ ³ ´ ¡ ¢ A£ B £P = B A¢P ¡ A¢B P ³ ´ ³ ´ A£P ¢B = A£ P ¢B = A£P ¢B ³ ´ ³ ´ A ¢P £B = A ¢ P £ B = A ¢P £B ³ ´ ³ ´ A£P £ B = A£ P £B = A£P £B ³ ´T P ¢Q = QT ¢ P T ³ ´ ³ ´ A ¢P ¢Q = A¢ P ¢ Q = A ¢ P ¢ Q ³ ´ ³ ´ P ¢ Q ¢ A = P ¢ Q ¢ A = P ¢ Q ¢A ³ ´ ³ ´ P ¢ A £Q = P £ A ¢Q . but it can also be represented as (rr) ¢ A.123) (A.126) (A.128) (A. dyadic notation is frequently used for brevity.127) (A.DYADIC ANALYSIS 19 It is frequently useful to employ a unit dyad.132) (A. I = xx + yy + z z ^ ^ ^^ ^^ (A.121) This is analogous to the identity matrix in linear algebra. A simple example given in the text is the function rr ¢ A. we find it can be employed in many situations formerly handled by vector manipulations.

137) (A.7.147) (A.152) (A.156) (A.153) (A.154) (A.138) A.145) (A.151) (A.141) (A.148) (A.157) (A.146) (A.143) (A.158) .149) A.155) (A.144) (A.142) (A.20 VECTOR AND DYADIC ANALYSIS ³ ´ ³ ´ A £P ¢ Q = A £ P ¢ Q = A £P ¢Q ³ ´ ³ ´ P ¢ Q £A = P ¢ Q£ A = P ¢Q £A (A.3 Properties of the Unit Dyad (A.7.139) (A.150) (A.140) (A.2 Differential Operations Involving Dyads ¡ ¢ r ÁA = (rÁ) A + ÁrA ³ ´ r ¢ ÁP = (rÁ) ¢ P + Ár ¢ P ³ ´ r £ ÁP = (rÁ) £ P + Ár £ P ¡ ¢ ¡ ¢ ¡ ¢ r ¢ A B = r ¢ A B + A ¢ rB ¡ ¢ ¡ ¢ r ¢ AB ¡B A = r £ B £ A ¡ ¢ ¡ ¢ r £ A B = r £ A B ¡ A £ rB ¡ ¢ ¡ ¢ ¡ ¢ r A £ B = rA £ B ¡ rB £ A ³ ´ ¡ ¢ r ¢ A £P = r £A ¢ P ¡ A ¢r £ P ¡ ¢ r £ rA = 0 ³ ´ r¢ r£P = 0 ³ ´ ³ ´ r £ r £ P = r r ¢ P ¡ r2 P I = IT A¢ I = I ¢A = A I£A=A£I ³ ´ ³ ´ A£ I ¢ B = A¢ I £ B = A£ B ¡ ¢ I £ A £ B = BA ¡ AB ³ ´ A£ I ¢ P = A £ P ³ ´ r ¢ ÁI = rÁ ³ ´ r¢ I £A = r£A ³ ´ r £ ÁI = rÁ £ I (A.

164) S ZZ (dS £ rA) = I C d` A C The vector-dyadic form of Green’s identity (A. by convention: ZZZ Z Z r ¢ P dV = ° dS ¢ P (A.163) (A.159) A.160) ZZZ ZZZ V V V rA dV = ° dS A Z Z S Z Z S (A. with the direction of dS taken as pointing outward from the enclosed volume.167) . In the following.162) r £ P dV = ° dS £ P S In the following.64) and Green’s theorem (A.165) i ¢ r £ r £ A ¢ P ¡ A ¢ r £ r £ P dV Z Z h i ¡ ¢ = ° ^ ¢ A £ r £ P + r £ A £ P dS n (A. The direction of d` is tangent to C .161) (A. The dyadic-dyadic forms of the above are: n ¸ ZZZ ·³ ´T ³ ´T r £ Q ¢ r £ P ¡ r £ r £ Q ¢ P dV V S Z ³ Z ´T ³ ´ = ° r £ Q ¢ n £ P dS ^ (A. The ^ direction of dS is normal to the surface following the right-hand rule with the fingers curled in the direction of C: ZZ I ³ ´ dS ¢ r £ P = d` ¢ P S (A.166) where dS = ^ dS. V is a volume bounded by a closed surface S. while the normal to C is described by n.65) are (see [5] for a derivation) ZZZ h Z Z i ¡ ¢ r £ A ¢ r £ P ¡ A ¢ r £ r £ P dV = ° n ¢ (A £ r £ P ) dS ^ ZZZ h ¡ V S (A.DYADIC ANALYSIS 21 (A.4 Integral relations We can generalize the earlier vector integral theorems in a straightforward manner to accomodate dyadic functions.7. S is an open surface bounded by a contour C described by line element d`.

-T. Grad. C. . 5. W.W. An excellent introductory text written especially for electrical engineers.E. 1973. and all that: an informal text on vector calculus. Schey.E. NJ. W.22 VECTOR AND DYADIC ANALYSIS ZZZ · ¸ ³ ´T Q T ¢ r £ r £ P ¡ r £ r £ Q ¢ P dV Z ·³ Z ´T ³ ´ ³ ´¸ = ° r £ Q ¢ n £ P + QT ¢ n £ r £ P ^ ^ dS (A. Marsden and A. IEEE Press: Piscataway. Freeman & Company: New York. 4. J. Field Theory of Guided Waves.168) REFERENCES 1. J. Singular Electromagnetic Fields and Sources.M. Clarendon Press: Oxford. Norton & Company: New York. Curl. 2. 1981.H. 1991. Vector Calculus. Tai. Collin. 1991. Div. Van Bladel. 3..J. R. Tromba. Dyadic Green Functions in Electromagnetic Theory. 1994. H. NJ. 2nd ed. IEEE Press: Piscataway.

Sign up to vote on this title

UsefulNot useful- Fast Calculations of Dyadic Green’s Functions
- Vectors
- Dyadic Green Functions in EM Theroy
- Crash Course on Vectors
- Lecture Notes for Applied Mathematics
- Vector Calculus
- Lembretes Matemática
- CN W02D1 Vectors
- Lec_2.pdf
- ee1 notes
- 35675966 EMT Electromagnetic Theory MODULE I (1)
- Vector Review
- Jackson Sol
- EM Lecture Notes Chapter 1 Griffiths
- Coordinates
- 302-1ba-chapter1.ppt
- VECTRS3D
- Spherical Coords Note
- cyli
- Student Handout 02 2014
- FisDas - Week 01 - Unit, Dimension & Vector
- 12
- 2.1 Scalars and Vector Quantities
- SM Chapter3
- Engineering Mechanics Notes2222
- Physics Sample
- Physics - Ch5 Vectors
- 1 Solutions x
- 고미 2013 여름계절 렉쳐노트
- Estática 1
- Dyadics Identities