Physics Library
 An open source physics library
Encyclopedia | Forums | Docs | Random  

Helmholtz Decomposition, Potentials, and Field Reconstruction

(Topic)

ED07: Helmholtz Decomposition, Potentials, and Field Reconstruction

The vector field F(r) may vary in both magnitude and direction throughout space. ED03 introduced its divergence and curl; ED04 explained their integral meanings; ED05 established the correct treatment of singular sources; and ED06 derived the differential operators in symmetry-adapted coordinates. A natural question now arises:

If we know  ∇ ⋅ F and ∇ ×  F, do we know  F?
(1)

Under suitable decay, regularity, boundary, and topological assumptions, the answer is yes. The result is called the Helmholtz decomposition theorem. In free space it splits a sufficiently well-behaved vector field into an irrotational part and a solenoidal part:

F = − ∇ Φ + ∇  × A.
(2)

The scalar potential Φ and vector potential A are computed from the field’s divergence and curl. This is a mathematical explanation for why potentials are so central to electrodynamics, not merely a convenient guessing method.

The words under suitable assumptions matter. On finite domains, boundary data are essential; in multiply connected regions, circulation around holes can carry information that neither local divergence nor local curl contains. We will derive the reconstruction, show the failure modes, and connect everything to electric and magnetic fields.

PIC

Figure 1. Divergence and curl encode complementary local information. Together with the appropriate decay, boundary, and global circulation information, they determine a vector field.

1 First distinguish two kinds of field structure

The divergence is a scalar field,

D (r) = ∇  ⋅ F(r),
(3)

while the curl is a vector field,

C (r) = ∇ × F (r).
(4)

Because a divergence of a curl is identically zero,

∇ ⋅ C = 0.
(5)

Thus the three components of C cannot be specified independently: they must satisfy this compatibility condition. Likewise, on a bounded domain the divergence source and the prescribed outward Normal field must satisfy the divergence theorem,

∫          ∮
   D d τ =     F ⋅ da.
 V          ∂V
(6)

Call a field irrotational if ∇× F = 0, and solenoidal if ∇⋅ F = 0. An irrotational field need not be constant. A solenoidal field need not circulate visibly around circular streamlines. These are differential properties, not pictures alone.

2 A vector identity provides the starting point

For a twice differentiable vector field the vector Laplacian obeys

                              2
∇  × (∇ ×  F) = ∇ (∇ ⋅ F ) − ∇ F.
(7)

Here ∇2F means applying the scalar Cartesian Laplacian to each Cartesian component of F. Rearranging gives

  2
∇  F =  ∇ (∇ ⋅ F ) − ∇ × (∇ × F).
(8)

Insert Definitions (3) and (4):

  2
∇  F = ∇D   − ∇ ×  C.
(9)

This is already close to a reconstruction equation. It says that the componentwise Laplacian of the unknown field is fixed by derivatives of its divergence and curl.

A component-level check

For the x component, the curl of the curl is

[∇× (∇× F)]x =  ∂
---
∂y( ∂Fy    ∂Fx )
  ---- − ----
   ∂x     ∂y
−-∂-
∂z(            )
  ∂Fx-   ∂Fz-
   ∂z  −  ∂x. (10)

Expand and group the mixed derivatives:

[∇× (∇× F)]x = ∂
---
∂x( ∂F     ∂F  )
  ---y+  ---z
   ∂y    ∂z
−(             )
  ∂2Fx    ∂2Fx
  ∂y2--+  ∂z2--. (11)

Adding and subtracting ∂2F x∕∂x2 yields

                   ∂D--    2
[∇ ×  (∇ × F )]x =  ∂x  − ∇  Fx.
(12)

The y and z components follow by cyclic permutation. This gives a direct derivation of Equation (7) instead of treating it as a mnemonic.

3 Review: the Green function for Poisson’s equation

The free-space Green function is

               1
G (r,r′) = ---------.
           4π|r − r′|
(13)

The coordinate r′ identifies a source point; r identifies the field or observation point. Define

R  = r − r′,    R  = |R |.
(14)

The distributional identity from ED05 is

  2     ′       (3)     ′
∇  G(r,r ) = − δ  (r − r ).
(15)

Thus the solution of

  2
∇  u(r) = − s(r)
(16)

that decays appropriately at infinity is

       ∫
                ′   ′    ′
u(r) =    G (r,r )s(r )dτ .
        ℝ3
(17)

Indeed, acting with the Laplacian and using the delta sifting property gives

∇2u(r) = ∫ (∇2G)s(r′) dτ′ (18)
= −∫ δ(3)(r − r′)s(r′) dτ′ (19)
= −s(r). (20)

This solution is particular to free space. If a boundary is present, one must impose boundary data or add an appropriate harmonic correction.

PIC

Figure 2. Green-function reconstruction sums contributions from source points r′ to the observation point r, weighted by G = 1∕(4πR). The integration is over three-dimensional source volume.

4 Construct a scalar potential from divergence

Given D = ∇⋅ F, define

        1  ∫   D (r′)
Φ (r) =  ---   ------′ dτ′.
        4π  ℝ3|r − r |
(21)

Equation (15) immediately gives

∇2 Φ = − D.
(22)

The associated gradient field

FL =  − ∇ Φ
(23)

is irrotational because ∇×∇Φ = 0, and it has the required divergence:

∇⋅ FL = −∇2Φ (24)
= D. (25)

The subscript L stands for longitudinal. In Fourier space the gradient points along the wave vector, giving the name its precise meaning.

5 Construct a vector potential from curl

Given C = ∇× F, define the vector potential component by component:

           ∫       ′
A (r) = 1--    C-(r-)-dτ′.
        4π  ℝ3 |r − r′|
(26)

The same Green-function identity gives

∇2A  = − C.
(27)

Now show why this particular potential is divergence-free. In the integral,

∇G  =  − ∇ ′G.
(28)

Therefore

∇⋅ A = ∫ C(r′) ⋅∇Gdτ′ (29)
= −∫ C(r′) ⋅∇′Gdτ′. (30)

Use the product rule in the source coordinates:

∇ ′ ⋅ (GC ) = C ⋅ ∇ ′G + G ∇ ′ ⋅ C.
(31)

Then

∇⋅ A = −∮ S∞GC ⋅ da′
+ ∫ G(∇′⋅ C) dτ′. (32)

For localized, sufficiently decaying sources the surface term vanishes. The remaining volume term vanishes by Equation (5). Consequently

∇  ⋅ A = 0.
(33)

This is the Coulomb gauge. It emerges automatically for the free-space integral in Equation (26) under these assumptions.

Define the transverse field

FT  = ∇ ×  A.
(34)

Its divergence is identically zero:

∇ ⋅ FT = 0.
(35)

Its curl is

∇× FT = ∇× (∇× A) (36)
= ∇(∇⋅ A) −∇2A (37)
= C. (38)

So FT reproduces exactly the prescribed curl.

6 Put the two contributions together

Define

Frec = − ∇ Φ + ∇ × A.
(39)

Take its divergence:

∇⋅ Frec = −∇2Φ + ∇⋅ (∇× A) (40)
= D + 0 = D. (41)

Take its curl:

∇× Frec = −∇×∇Φ + ∇× (∇× A) (42)
= 0 + C = C. (43)

Thus F and Frec have identical divergence and curl. Their difference

H  = F − Frec
(44)

satisfies

∇  ⋅ H = 0,    ∇  × H  = 0.
(45)

The Vector Identity then implies

∇2H  =  0.
(46)

Every Cartesian component is harmonic. In all of ℝ3, if H is smooth, bounded, and tends to zero at infinity, the maximum principle (or Liouville’s theorem for harmonic functions) gives H = 0. This proves the free-space reconstruction:

F = − ∇ Φ + ∇  × A,
(47)

with Φ and A given by Equations (21) and (26). More delicate versions of the theorem cover other decay classes and distributional sources; the formulas should not be applied indiscriminately to growing or non-decaying fields.

7 A second proof using the vector Poisson equation

We can also derive the decomposition directly from Equation (9). Formally invert its Laplacian with the free-space Green function:

         ∫        ′  2    ′   ′
F(r) = −    G (r,r)∇ r′F(r )dτ .
(48)

Substitute ∇r′2F = ∇′D −∇′× C:

F(r) = −∫ G∇′D dτ′ (49)
+ ∫ G(∇′× C) dτ′. (50)

Integrate by parts, assuming the terms at infinity vanish. For the first integral,

∫ G∇′D dτ′ = −∫ D∇′Gdτ′ (51)
= ∇∫ GD dτ′ = ∇Φ. (52)

For the second, use ∇′× (GC) = ∇′G × C + G∇′× C. The surface term again vanishes, leaving

∫ G∇′× Cdτ′ = −∫ ∇′G × Cdτ′ (53)
= ∇×∫ GCdτ′ (54)
= ∇× A. (55)

Therefore the inversion reproduces Equation (47). This proof explains structurally why both a gradient and a curl appear. The first proof established that they reproduce the prescribed local data; this one reconstructs them from the Green-function inverse of the vector Laplacian.

8 Why the names longitudinal and transverse?

For a Fourier component with nonzero wave vector k, write

       ^     ik⋅r
F(r) = F (k)e
(56)

as one mode in a Fourier superposition. On that mode, ∇ acts as ik. Define the parallel projection

            ^
F^  = k-(k ⋅-F),
  L      k2
(57)

and the perpendicular remainder

^FT =  ^F − ^FL.
(58)

These satisfy

k × ^FL =  0,     k ⋅ ^FT = 0.
(59)

The divergence and curl Fourier coefficients are

^D =  ik ⋅F^,     C^ = ik × ^F.
(60)

Solving for the two projections gives

        ik ^D             ik ×  ^C
^FL =  − --2-,    F^T  =  ---2---.
         k                k
(61)

The zero-wave-number mode is not captured by division by k2; it is precisely where constants and other global or boundary information matter.

PIC

Figure 3. Relative to a Fourier wave vector k, a vector mode separates into a longitudinal projection parallel to k and a transverse projection perpendicular to k.

9 Potentials and gauge freedom

The vector field in Equation (47) is unchanged by the transformation

A − →  A + ∇ χ,
(62)

for any sufficiently regular scalar χ, since

∇  × (∇ χ) = 0.
(63)

This is gauge freedom. The particular free-space reconstruction above chooses a divergence-free vector potential. If a different vector potential is given, it can be adjusted to Coulomb gauge by solving

∇2 χ = − ∇ ⋅ A.
(64)

The scalar potential is unchanged under an additive constant because its gradient does not depend on that constant. These familiar freedoms should not be confused with ambiguity in the physical field: gauge transformations change the potentials but not the reconstructed field.

In Electromagnetism with time-dependent fields, the physical potentials have the larger gauge transformation

                                ∂χ
A − →  A + ∇ χ,     V  −→  V −  ---.
                                ∂t
(65)

That dynamical gauge freedom will be treated in a later ED article. The scalar and vector potentials in the purely spatial Helmholtz theorem need not equal the retarded electromagnetic potentials.

10 Electrostatics: the scalar potential becomes physical

For an electrostatic electric field,

                         ρ-
∇ ×  E = 0,     ∇  ⋅ E = 𝜖 .
                          0
(66)

In all of space, assuming suitable decay and no additional boundary field, Helmholtz reconstruction gives

E  = − ∇V,
(67)

with

             ∫      ′
V (r) = --1--   -ρ(r-)-dτ ′.
        4π 𝜖0   |r − r ′|
(68)

Applying the Laplacian yields Poisson’s equation,

∇2V   = − ρ-.
          𝜖0
(69)

Thus the standard electrostatic potential is a direct specialization of the scalar part of Helmholtz decomposition. For a point charge Q at the origin, the ED05 delta distribution yields

V =  --Q---,
     4π 𝜖0r
(70)

and therefore, for r > 0,

     --Q--ˆr-
E  = 4π 𝜖 r2.
         0
(71)

This satisfies Gauss’s Law distributionally at the origin.

PIC

Figure 4. An electrostatic field is normal to equipotential contours and is locally curl-free. The scalar potential retains source and boundary information that field lines alone do not show.

11 Magnetostatics: the vector potential becomes physical

For steady, localized currents in free space,

∇  ⋅ B = 0,    ∇  × B =  μ0J.
(72)

The continuity equation for stationary sources gives

∇ ⋅ J = 0,
(73)

as required by the compatibility condition for a curl. Helmholtz reconstruction sets the scalar contribution to zero when ∇⋅ B = 0 and the field decays properly, giving

B = ∇  × A,
(74)

where

        μ0 ∫   J (r′)
A (r) = ---   ------′d τ′.
        4π    |r − r|
(75)

Taking the curl and using ∇G = −∇′G gives

B(r) =  μ
--0
4 π∫ ∇(  1)
  --
  R× J(r′) dτ′ (76)
= -μ0
4 π∫ J-(r′)-×-R-
    R3 dτ′. (77)

This is the volume-current Biot-Savart expression. Helmholtz decomposition therefore connects Ampere’s differential law to the Biot-Savart field integral, under the assumptions of magnetostatics and free space.

12 Worked example 1: a smooth, localized gradient field

Let

              (     )
                  r2
Φ(r) = Φ0 exp   − -2- ,    a >  0.
                  a
(78)

Here r2 = x2 + y2 + z2. First differentiate the exponent:

  (    2)
∇   − r-- =  − 2r.
      a2       a2
(79)

Thus

                  (     )
         2Φ0-         r2-
∇ Φ  = −  a2 rexp   − a2  .
(80)

Set

                      (     )
             2Φ0-         r2-
F =  − ∇Φ  =  a2 r exp  − a2  .
(81)

Because it is a gradient, its curl vanishes everywhere. Find its divergence with the product rule:

∇⋅ F = 2Φ0-
 a2∇⋅(  − r2∕a2)
 re (82)
= 2Φ0
--2-
 a[     2 2          2  2]
 3e− r∕a +  r ⋅ ∇e − r∕a (83)
= 2Φ0-
 a2(       2)
  3 − 2r--
       a2e−r2∕a2 . (84)

The divergence is positive near the origin and negative sufficiently far away, while the field decays rapidly. Its total divergence integrates to zero, consistent with zero net flux at infinity. Reconstructing from Equation (21) recovers the original Φ up to an additive constant, fixed here by Φ → 0 at infinity. This is a convenient smooth example without point singularities.

13 Worked example 2: a smooth, localized solenoidal field

Choose a simple vector potential in Cartesian coordinates,

             (   r2)
A0  = A0 exp   − -2- ˆz,
                 a
(85)

with constant A0. Its curl is

F  = ∇ ×  A0 =  2A0-e−r2∕a2 (− yˆx + x ˆy).
                 a2
(86)

This field loops around the z axis and decays rapidly in every spatial direction. The divergence is

∇⋅ F = 2A0-
 a2[                              ]
 -∂-(− ye−r2∕a2) +-∂-(xe− r2∕a2)
 ∂x               ∂y (87)
= 2A0-
 a2[                        ]
 2xy-e− r2∕a2 − 2xy-e−r2∕a2
  a2           a2 (88)
= 0. (89)

Let K = 2A0∕a2 and g = e−r2∕a2. The curl of Equation (86) is

∇× F = 2K--
a2g(xz ˆx + yzˆy )
+ 2Kg(            )
      x2 +-y2-
  1 −   a2z. (90)

The field is solenoidal, but its curl is nonzero. The free-space Helmholtz integral reconstructs it entirely from Equation (90) and the decay condition. The convenient trial potential A0 is not necessarily in Coulomb gauge; different potentials producing the same F may differ by a gradient.

PIC

Figure 5. A smooth solenoidal field may wrap around an axis, have nonzero curl, and decay in all directions. Divergence measures net local flux, not whether nearby field lines visually curve.

14 Worked example 3: a constant field cannot be reconstructed from local data alone on a ball

Let V be a spherical region and define

F1  = 0,     F2 = E0 ˆx.
(91)

Both satisfy

∇  ⋅ Fi = 0,    ∇ × Fi =  0
(92)

throughout the ball. They are nevertheless different physical fields. The difference is the nonzero harmonic field E0x.

The boundary normal component distinguishes them. On a sphere of radius R,

F2 ⋅ ˆn = E0 sin 𝜃cosϕ,
(93)

where 𝜃 is polar angle measured from +z and ϕ is azimuth. For F1 the normal component is zero. Thus specifying only divergence and curl does not fix this finite-domain problem; the boundary field is additional data.

PIC

Figure 6. Two fields inside the same finite ball share zero divergence and curl but differ in boundary-normal data. The missing information is not captured by the local differential operators alone.

15 Uniqueness with suitable normal boundary data

Suppose F1 and F2 have the same divergence, curl, and normal component on the boundary of a connected, simply connected region V with sufficiently smooth boundary. Their difference H satisfies

∇ ⋅ H = 0,     ∇ ×  H =  0,    H  ⋅ ˆn = 0 on ∂V.
(94)

In a simply connected region, curl-free H can be represented by a single-valued scalar function:

H  = ∇ ψ.
(95)

The divergence condition becomes Laplace’s equation,

∇2 ψ = 0.
(96)

The boundary condition is

∂ψ- = 0.
∂n
(97)

Now integrate the divergence identity

                  2      2
∇ ⋅ (ψ ∇ ψ) = |∇ ψ| + ψ ∇ ψ
(98)

over V . The divergence theorem gives

∫ V |∇ψ|2 dτ = ∮ ∂V ψ∂ψ-
∂nda −∫ V ψ∇2ψ dτ (99)
= 0. (100)

The integrand |∇ψ|2 is nonnegative, so it must vanish everywhere for smooth ψ. Consequently ∇ψ = 0, hence F1 = F2. This proves uniqueness for the specified simply connected finite-domain setting. Existence additionally requires compatible source and boundary data; uniqueness by itself does not guarantee a solution exists.

16 Topology: a field around a hole can escape the local tests

Consider the annular cylindrical region

V  = {(s,ϕ, z) : a < s < b, 0 < z < L},    a > 0.
(101)

Define

H =  K-ϕˆ,
     s
(102)

where K is constant. Throughout this region, the cylindrical divergence is

         1∂H-ϕ-
∇ ⋅ H =  s ∂ϕ  =  0,
(103)

and the z component of its curl is

(∇× H)z = 1-
s-∂-
∂s(sHϕ) (104)
= 1
--
s∂K
----
 ∂s = 0. (105)

The other curl components also vanish. The field is tangent to the inner cylinder, outer cylinder, and flat end caps, so

H ⋅nˆ=  0    on  all of ∂V.
(106)

Nevertheless the circulation around a loop of fixed s and z is

∮ CH ⋅ dl = ∫ 02πK--
 s(sdϕ) (107)
= 2πK. (108)

This is nonzero when K≠0. The same local divergence, curl, and normal boundary data are also satisfied by 0. Here the domain has a hole, so the previous simply connected uniqueness proof does not apply. The circulation around a noncontractible loop supplies additional information.

PIC

Figure 7. In an annular region, Kϕ∕s has zero divergence, zero ordinary curl, and zero normal boundary component, yet nonzero circulation around the excluded axis. A global circulation period distinguishes it from the zero field.

If one extends Equation (102) to all space including the axis, it becomes singular there. ED05 showed the corresponding distributional curl is concentrated on the axis, so the local zero-curl claim must be restricted to s > 0. This example connects the topology of a hole with the singular-source mathematics of the previous lessons.

17 Finite-domain Green representations and boundary information

Why did the free-space integral omit boundary information? The simplest explanation comes from Green’s second identity applied to each Cartesian component Fi. With

G (r,r′) = ----1----,
           4π|r − r′|
(109)

and r inside a bounded volume V , the component Green representation is

Fi(r) = −∫ V G∇r′2F i dτ′
+ ∮ ∂V (   ∂F       ∂G )
  G --i′-− Fi---′
    ∂n      ∂nda′. (110)

The volume term contains ∇′D −∇′× C by Equation (9). The surface term contains information about the field and its derivatives on the boundary. This does not mean both quantities must be specified independently: boundary-value problems prescribe suitable data, and the equations determine the rest. It means that the interior source data alone are generally insufficient for reconstruction on a finite region.

PIC

Figure 8. For a finite region, source information in the interior must be accompanied by suitable boundary information. Green’s representation separates volume and boundary contributions explicitly.

18 A caution about time-dependent electrodynamics

At a single instant, the Helmholtz theorem may be applied mathematically to E(r,t) or B(r,t) as a spatial field, under the same spatial assumptions. But the resulting instantaneous potentials are not by themselves a causal solution of Maxwell’s dynamical equations. In particular,

            ∂B--
∇  × E =  − ∂t
(111)

means the electric field need not be purely longitudinal. Physical retarded potentials, gauge conditions, and initial radiation data must be handled in the full time-dependent problem. Avoid interpreting the free-space Helmholtz integrals at each time as propagation rules with infinite signal speed.

19 Practical field-reconstruction workflow

When given divergence, curl, or measured field data:

  1. State the spatial domain. Is it all space, a finite simply connected volume, or a region with holes?
  2. Check compatibility: ∇⋅ C = 0 and any boundary flux constraints.
  3. Specify decay behavior at infinity or appropriate boundary conditions.
  4. For full space, construct Φ from D and A from C using the Green integrals.
  5. Reconstruct F = −∇Φ + ∇× A.
  6. Verify both divergence and curl; then verify the boundary conditions or decay.
  7. In multiply connected domains, check circulation around independent noncontractible loops.
  8. For singular fields, interpret the differential identities distributionally and inspect excluded points, lines, or surfaces.

This workflow turns a formal theorem into a practical method for electromagnetics and computational field modeling.

20 Practice exercises

  1. Starting from the component definition of curl, verify Equation (7) for the y component.
  2. Given D(r) and C(r), write the two free-space potential integrals and list their decay assumptions.
  3. Starting from Equation (26), show carefully that ∇⋅ A = 0 when ∇⋅ C = 0 and all surface terms vanish.
  4. Show that ∇⋅ (∇× A) = 0 and ∇×∇Φ = 0 wherever the potentials are sufficiently smooth.
  5. Verify Equation (84) independently in spherical coordinates.
  6. Verify ∇⋅ F = 0 for the localized vortex in Equation (86) using Cartesian partial derivatives.
  7. In Equation (85), compute ∇⋅ A0 and explain why the trial potential is not generally in Coulomb gauge.
  8. Let F = F0z inside a finite ball. Find D, C, and F ⋅n on its surface.
  9. Show directly that the circulation in Equation (108) does not depend on the loop radius s.
  10. Derive the free-space electrostatic scalar potential by setting D = ρ∕𝜖0 and C = 0.
  11. Starting with the steady-current vector potential, take its curl to obtain the Biot-Savart integral.
  12. For k = (k, 0, 0), show explicitly which Cartesian components of a Fourier vector mode are longitudinal and which are transverse.

21 Summary

The vector identity

∇2F  = ∇ (∇ ⋅ F ) − ∇ × (∇ × F )
(112)

reveals why divergence and curl are the source data for the vector Poisson problem. Under appropriate free-space assumptions,

F = − ∇ Φ + ∇  × A,
(113)

with

           ∫    ′    ′
        -1-   ∇--⋅ F-(r-)  ′
Φ (r) = 4π       R     d τ,
(114)

and

            ∫
         1    ∇ ′ × F(r′)   ′
A (r) = 4-π   -----R-----dτ .
(115)

The first field part is longitudinal and curl-free; the second is transverse and divergence-free. Electrostatic and magnetostatic potentials arise as the corresponding special cases. On finite or multiply connected domains, boundary data and global circulation information are essential to a correct uniqueness statement.

References

[1]   D. J. Griffiths, Introduction to Electrodynamics, 5th ed., Cambridge University Press, 2023.

[2]   H. M. Schey, Div, Grad, Curl, and All That, 4th ed., W. W. Norton, 2005.

[3]   G. B. Arfken, H. J. Weber, and F. E. Harris, Mathematical Methods for Physicists, 7th ed., Academic Press, 2013.

[4]   J. D. Jackson, Classical Electrodynamics, 3rd ed., Wiley, 1998.

[5]   P. M. Morse and H. Feshbach, Methods of Theoretical Physics, McGraw-Hill, 1953.


"Helmholtz Decomposition, Potentials, and Field Reconstruction" is owned by bloftin.
(view preamble)
View style:
Other names:  ED07
Also defines:  Helmholtz decomposition, longitudinal field, transverse field, irrotational field, solenoidal field, harmonic field, scalar potential, vector potential, Coulomb gauge, Green function, Helmholtz reconstruction, boundary uniqueness, circulation period, topological ambiguity, field reconstruction
Keywords:  Helmholtz decomposition, vector fields, divergence, curl, scalar potential, vector potential, Green functions, Coulomb gauge, boundary conditions, harmonic fields, topology, electrostatics, magnetostatics, field reconstruction

Cross-references: spherical coordinates, speed, radiation, representation, distributional curl, Cartesian coordinates, flux, magnetostatics, continuity equation, equipotential, electrostatic field, Gauss's Law, charge, electric field, Electromagnetism, regular, Fourier coefficients, superposition, formulas, distributional sources, functions, Vector Identity, wave vector, gradient, volume, harmonic, sifting property, identity, Laplacian, vector, divergence theorem, Normal, scalar, magnetic fields, circulation, boundary data, domains, solenoidal, theorem, topological, boundary, operators, ED06, ED05, ED04, curl, divergence, ED03, magnitude, vector field
There are 4 references to this object.

This is version 1 of Helmholtz Decomposition, Potentials, and Field Reconstruction, born on 2026-11-21.
Object id is 1590, canonical name is HelmholtzDecompositionPotentialsAndFieldReconstruction.
Accessed 6 times total.

Classification:
Physics Classification: 02.30.Em (Potential theory)
 02.30.Jr (Partial differential equations)
 41.20.-q (Applied classical electromagnetism)

Pending Errata and Addenda

None.

Discussion

No messages.

Interact