The Complex Potential of a Self-Dual Maxwell Field
Introduction
The corpus writes the vacuum Maxwell equations as a regularity condition on a biquaternion, and its field is real. There is a second, older way of writing the same equations that the corpus has not recorded, in which the potential rather than the field strength is the primary object and the duality purity of the field is a set of first-order conditions on the potential. It is the form in which the algebrodynamical programme generates its solutions, and the self-duality it imposes is also what makes those solutions carry quantised charge.
This article records that form. The potential is a complex 4-vector, so the field it produces is complex and is never a real field strength; the condition is available only because of this, and the reality obstruction of The Self-Dual and Anti-Self-Dual Split is the reason. In a gauge in which two of the four components vanish the condition becomes a pair of first-order equations of Weyl type, and with the potential written as derivatives of a single scalar one of those equations is an identity and the other is the wave equation — so the linear content of the construction is exhausted by one harmonic scalar, and everything that selects the physical class is imposed on top of it. In this construction every self-dual field is therefore generated by one complex harmonic scalar — the superpotential — and the field strengths are its second derivatives.
Three results follow, one sharp, one limiting, one clarifying. The scalar family that gives the Coulomb field satisfies the wave equation and the eikonal for every value of its parameter and the first-order constraints for exactly one value: the first two conditions are scale invariant and the constraints are not, so the elementary charge is the unique scale at which the two sides of a constraint balance, and the quantisation is a mismatch of degree rather than a dynamical accident. And the reduced system is strictly weaker than the condition: a plane-wave superpotential satisfies it and generates a field that is not duality pure, so what selects the physical class is the nonlinear constraint and not the linear system — the corpus's induced-nonlinearity theme met here from the linear side. And the complex field has no bilinear density at all — the invariant and the bilinear stress vanish identically on a duality-pure field — which does not contradict the corpus's energy–momentum tensor, that tensor being sesquilinear and belonging to the field's real projection.
Three conventions are load-bearing, and the second is the one the source's displayed equations turn on.
- The field strength is $\vec E=-\partial\vec C/\partial h_t-\vec\nabla_\Phi\Pi$, $\vec H=\vec\nabla_\Phi\times\vec C$ with complex $\Pi$ and $\vec C$, and the entries $C_u,C_v,C_w,C_{\bar w}$ below are the covariant components, as in the source's own display of the potential matrix. With these definitions the Coulomb solution has $\vec H$ purely imaginary with $\vec H=i\vec E$, so the duality condition reads $\vec E+i\vec H=0$: the residual of that combination is of order $10^{-11}$ at a generic point, while the opposite combination does not vanish: because $\vec H=i\vec E$ it equals $2\vec E$ identically, here $1.21$ at the test point, where $\lvert\vec E\rvert=0.60$.
- The sign of $i$ in the source's printed equations is the opposite one, for a reason it flags in passing: the spatial potential in its field-strength definition is the contravariant vector, $C_a=-C^a$. Flipping the sign of the spatial potential flips $\vec H$ and leaves the gradient term of $\vec E$ alone, so the two readings are one equation.
- In the corpus's naming, with $\vec H$ read as $c\vec B$, the condition $\vec E+i\vec H=0$ is the anti-self-dual sign, since the corpus calls $\vec E-ic\vec B=0$ self-dual. The source's adjectives are not uniform — its section title says complex self-dual, its text at (11) says antiself-duality — so formulas rather than adjectives are what is carried over.
Two further readings are fixed once. The invertibles are the source's, Gaussian ones in which $\mu=1$ and both $\vec B$ and $\vec H$ carry the dimension of $\vec E$, so that $\vec B$ and $\vec H$ coincide and both carry the dimension of $\vec E$; the corpus's $\vec B=\mu\vec H$ and the factor $c$ return wherever a comparison with $\vec N=\vec E+ic\vec B$ is made, and the charge $ϙ$ is a pure number in these units, rescaled to coulomb by the coupling. The field glyphs are the corpus's, and they carry no extra mark. In this article $\vec E$ and $\vec H$ are complex, as a complex potential requires; the corpus's real field appears only in cited formulas, and each citation says so. The combination $\vec E+i\vec H$ is the corpus's $\vec N$, so the duality-purity condition reads $\vec N=0$ and the pair needs no second glyph. No field glyph carries a tilde: a field is pure, and an arrowed glyph is written without one (Algebra and Element Conventions, §The Centre and the Vector Subspace).
The coordinate $z$ below is the Cartesian $z$, in the source's notation $z=(u-v)/2$, and the null coordinates are $u,v,w,\bar w$ — the algebrodynamics article writes the same four as $\upsilon,v,w,p$, so its $\upsilon$ is this $u$ and its $p$ is this $\bar w$.
The Complex Potential and the Self-Duality Condition
Let $\Pi$ and $\vec C$ be complex functions on complexified Minkowski space, and define
$$ \vec E=-\frac{\partial \vec C}{\partial h_t}-\vec\nabla_\Phi\Pi, \qquad \vec H=\vec\nabla_\Phi\times\vec C, \qquad \vec N:=\vec E+i\vec H . $$
The homogeneous Maxwell equations for a complex field are equivalent to the single complex vector equation
$$ \boxed{\;\frac{\partial \vec C}{\partial h_t}+\vec\nabla_\Phi\Pi-i\vec\nabla_\Phi\times\vec C=0\;}, $$
which is three complex conditions on four complex unknowns, and they are equivalent again to
$$ \vec E+i\vec H=0, $$
the duality-purity condition of the complex field, the anti-self-dual sign in the corpus's naming. The system is gauge invariant, with an arbitrary smooth complex gauge function, so one further condition may be imposed; the useful one is the complex Lorentz gauge
$$ \frac{\partial \Pi}{\partial h_t}+\vec\nabla_\Phi\cdot\vec C=0 . $$
The Cartesian split of the entries. The four matrix entries are the covariant components in the null frame, and the passage to a Cartesian potential is fixed by the two components that were gauged away: $C_u=\Pi+C_z=0$ and $C_w=C_x-iC_y=0$ give
$$ \Pi=\tfrac12C_v,\qquad C_x=\tfrac12C_{\bar w},\qquad C_y=-\tfrac{i}{2}C_{\bar w},\qquad C_z=-\tfrac12C_v . $$
This is the one place where the construction is easy to misread. The split is not $C_x=\mathrm{Re}\,C_{\bar w}$, $C_y=\mathrm{Im}\,C_{\bar w}$: the null combination is $C_{\bar w}=C_x+iC_y$ with a factor one, and covariant components carry $C_w=C_x-iC_y=0$, so both Cartesian components are complex and each is half the null one. With the real split the Lorentz gauge fails by $0.41$ where the field is $0.60$; with the split above it holds to order $10^{-11}$, and the field is the Coulomb one. All the numbers quoted in this article were recomputed from the split above.
The count is the one that matters later: three conditions plus one gauge condition on four complex components. The field strength is quadratic in the derivatives of the potential and is not affected by the gauge; the charge is read from it by Gauss's theorem.
Duality invariance, and what the condition costs. Duality rotation multiplies $\vec N$ by a phase, and $\vec N=0$ is preserved by it, so the condition is duality invariant in the sense the invariants article records. It is not a condition a real field can meet: with real $\vec E$ and $\vec H$, $\vec E+i\vec H=0$ forces both to vanish, six real equations on six real components. That is the reality obstruction of the split article, and it is the reason the potential here is complex at the outset and not a real potential with a complex field strength.
Null Coordinates
The conditions become transparent in the spinor coordinates
$$ u=t+z,\qquad v=t-z,\qquad w=x-iy,\qquad \bar w=x+iy, $$
with $\partial/\partial u=\tfrac12(\partial/\partial h_t+\partial/\partial r_z)$, $\partial/\partial v=\tfrac12(\partial/\partial h_t-\partial/\partial r_z)$, $\partial/\partial w=\tfrac12(\partial/\partial r_x+i\partial/\partial r_y)$, $\partial/\partial\bar w=\tfrac12(\partial/\partial r_x-i\partial/\partial r_y)$, and $\Box_\Phi=4(\partial^2/\partial u\,\partial v-\partial^2/\partial w\,\partial\bar w)$. The four components of the potential form a $2\times2$ complex matrix, and it is the matrix, not the 4-vector, in which the structure is visible:
$$ C=\begin{pmatrix}C_u & C_w\ C_{\bar w} & C_v\end{pmatrix}, \qquad C_u=\Pi+C_z,\quad C_v=\Pi-C_z,\quad C_w=C_x-iC_y,\quad C_{\bar w}=C_x+iC_y . $$
The two rows of the matrix are the two two-spinors of the potential; the source does not distinguish primed from unprimed spinor indices, which is why one index type is used throughout. In these coordinates the Lorentz gauge is the trace of a divergence,
$$ \frac{\partial C_u}{\partial u}+\frac{\partial C_v}{\partial v}+\frac{\partial C_w}{\partial w}+\frac{\partial C_{\bar w}}{\partial \bar w}=0 , $$
which is the form in which the gauge condition joins the matrix picture.
The Two-Component Gauge and the Reduced System
In these coordinates the self-duality condition plus the Lorentz gauge admits a gauge in which only two components survive, $C_{\bar w}$ and $C_v$, the other two, $C_u$ and $C_w$, being zero. The removal is the source's: the residual gauge freedom of the system is $C_\mu\mapsto C_\mu+\partial\lambda/\partial x_\mu$ with $\Box_\Phi\lambda=0$, which preserves the Lorentz gauge, and a parameter with that freedom can always be chosen to satisfy $C_u=\partial\lambda/\partial u$, $C_w=\partial\lambda/\partial\bar w$, because the compatibility condition $\partial C_u/\partial \bar w=\partial C_w/\partial u$ for those two equations holds identically on the equations of motion. Subtraction then removes both components at once.
In that gauge the system reduces to the pair
$$ \frac{\partial C_{\bar w}}{\partial \bar w}=\frac{\partial C_v}{\partial u}, \qquad \frac{\partial C_{\bar w}}{\partial v}=\frac{\partial C_v}{\partial w} , $$
which is the source's system, and which is the Weyl system of a massless two-spinor written in these coordinates. It is the content of the whole construction that this pair is solved by one function. Put
$$ C_{\bar w}=\frac{\partial G}{\partial u},\qquad C_v=\frac{\partial G}{\partial \bar w},\qquad \Box_\Phi G=0 . $$
Then the first equation holds identically, because it is the equality of the mixed partials $\partial^2G/\partial\bar w\,\partial u=\partial^2G/\partial u\,\partial\bar w$, and the second is the wave equation itself,
$$ \frac{\partial^2G}{\partial v\,\partial u}=\frac{\partial^2G}{\partial w\,\partial\bar w} \iff \Box_\Phi G=0 , $$
since $\Box_\Phi G=4(\partial^2G/\partial u\,\partial v-\partial^2G/\partial w\,\partial\bar w)$. The first is an identity on any $G$ and needs no check; the second is the wave equation. On the plane wave $G=\exp i(\vec k_\Phi\cdot\vec r-\omega h_t)$ the two sides of the first both equal $\tfrac{i}{4}(k_z-\omega)(k_y+ik_x)G$, and the two sides of the second are $\tfrac14(k_z^2-\omega^2)G$ and $-\tfrac14(k_x^2+k_y^2)G$, equal exactly when $\omega^2=\vec k_\Phi\cdot\vec k_\Phi$. So the pair holds identically for every harmonic plane wave, and the closed forms were confirmed numerically at a random point.
The pair is therefore not a differential system to be solved: one of its equations is an identity and the other is the wave equation. Every solution of the reduced system is obtained by differentiating one complex solution of the wave equation, and the source states the converse for every solution of the homogeneous Maxwell equations. The construction is a generation procedure, and the generation is differentiation.
The Superpotential, and the Field It Generates
The field strengths are the second derivatives of the same scalar, read on the components $E_{\eta}$ of $\vec E$. With the gauge above,
$$ E_{3}=-\frac{\partial C_v}{\partial u}, \qquad E_{+}:=E_{1}+iE_{2}=\frac{\partial C_{\bar w}}{\partial u}, \qquad E_{-}:=E_{1}-iE_{2}=-\frac{\partial C_v}{\partial \bar w} , $$
that is, $E_{3}=-\partial^2G/\partial u\,\partial\bar w$, $E_{+}=\partial^2G/\partial u^2$, $E_{-}=-\partial^2G/\partial\bar w^2$, and the symmetric triple $(E_{+},E_{3},E_{-})$ is the spinor of the field strength. The spinorial content is the reason the reduction works: a duality-pure two-form is a symmetric two-spinor, three complex components, and after the gauge is fixed the potential has exactly three, which the single scalar supplies through its second derivatives.
The assignment of the three second derivatives to the three components is the source's, and it is the contravariant reading of the spatial potential that makes it come out; with the covariant entries used here the first of the three reads $\partial C_v/\partial v$ instead of $-\partial C_v/\partial u$, and the two agree for static fields. What does not depend on the reading is the count: three components, one scalar, second derivatives.
The Coulomb case. Take
$$ G=\frac{\bar w}{z+r}, \qquad r=\sqrt{w\bar w+z^2}=\sqrt{r_x^2+r_y^2+r_z^2}, $$
which satisfies $\Box_\Phi G=0$ and the eikonal, and which is the stereographic projection $S^2\to\mathbb{C}$ from one pole. The potential is
$$ C_{\bar w}=\frac{\partial G}{\partial u}=-\frac{\bar w}{2r(r_z+r)}, \qquad C_v=\frac{\partial G}{\partial \bar w}=\frac{1}{2r}, $$
and the field is
$$ \vec E=-i\vec H=\frac{1}{4}\,\frac{\vec r}{r^3} : \qquad E_{3}=\frac{z}{4r^3},\quad E_{+}=\frac{\bar w}{4r^3},\quad E_{-}=\frac{w}{4r^3}, $$
the Coulomb field of a charge $\tfrac14$, with $\vec H$ purely imaginary and $\vec E$ purely real, so that the real part is the ordinary Coulomb field of a point charge and the imaginary part is the dual field of a magnetic monopole of the same modulus. This was recomputed here from the potential by finite differences: the self-duality condition and the Lorentz gauge both hold to order $10^{-11}$, and the field components agree with $\vec r/(4r^3)$ to the same accuracy.
A printed factor. The source writes the same solution with its charge parameter and gives $G=4ϙ\bar w/(z+r)$ together with $C_{\bar w}=-2ϙ\bar w/(z+r)$. Differentiating the printed $G$ gives
$$ \frac{\partial G}{\partial u}=-\frac{2ϙ\bar w}{r(r_z+r)} , $$
so the printed $C_{\bar w}$ is missing the factor $1/r$. The distinction is not cosmetic: with the corrected component the reduced system holds to order $10^{-11}$ at the test point, and with the printed one its two equations fail there by $2.4\times10^{-2}$ and $2.7\times10^{-1}$, where the field strength is $0.60$. The field strength, the charge and the rigidity below are unaffected, since they are computed from $C_v$ and from the correct derivative; only that one displayed component is wrong.
What the Complex Field Carries, and What It Does Not
The source remarks in passing that "complex fields do not possess any energy", adding that they acquire it only through self-division into real and imaginary parts. The remark is exact, and it is worth pinning down, because it looks like a contradiction of the corpus's energy–momentum tensor and is not one. The two statements are about two different products.
The corpus's tensor is sesquilinear, and the bilinear is a different object. The Electromagnetic Energy–Momentum Tensor builds $\Pi_{\mu\nu}=\tfrac12\mathrm{Sc}(\tilde F\mathcal{E}_\mu\tilde F^{*}\mathcal{E}_\nu)$ from the Hermitian form $\tilde F(\,\cdot\,)\tilde F^{*}$, whose scalar part is $2W$ with $W=\tfrac12(\epsilon E^2+\mu H^2)$ and whose vector part carries $\vec S=\vec E\times\vec H$. The plain bilinear is not the same object, and the difference does not wait for complex fields. Its scalar part is the invariant, $\langle\tilde F\rangle_{\natural}=-\epsilon\,(I_1+2ic\,I_2)$, with $I_1=\vec E^2-c^2\vec B^2$ and $I_2=\vec E\cdot\vec B$. The two therefore already disagree on a real field: at $\vec E=(0.3,-0.2,0.5)$ and $\vec B=(0.1,0.4,-0.2)$ the bilinear scalar is $-0.17+0.30i$ while $2W=0.59$; and on a real null field, $\vec B\perp\vec E$ with $\lvert\vec B\rvert=\lvert\vec E\rvert=0.616$, the bilinear vanishes to $8\times10^{-17}$ while $2W=0.76$. The bilinear is the invariant form; the energy is sesquilinear by construction.
The source's remark is about the bilinear. Take the same combinations without conjugation and impose duality purity, $\vec H=i\vec E$. Every bilinear combination vanishes, the invariant and the stress alike:
$$ E_{j}E_{j}+H_{j}H_{j}=E_{j}E_{j}-E_{j}E_{j}=0, \qquad \vec E\times\vec H=i\,\vec E\times\vec E=0, $$
and the same cancellation kills every entry of the bilinear analogue of the stress, $E_{j}E_{\eta}+H_{j}H_{\eta}=0$ — so it vanishes component by component and not merely in its trace. Checked here at the Coulomb solution and at a generic complex self-dual field: the scalar part is $5.5\times10^{-17}$ at the first and exactly zero at the second, and the Poynting vector and all nine entries $E_{j}E_{\eta}+H_{j}H_{\eta}$ are exactly zero at both.
In the corpus's own element the same fact is one line, and the corpus has already recorded it. Null Electromagnetic Fields and the Shear-Free Congruence states in its caution on the carrier that $\tilde F=i\sqrt{\epsilon}\,\vec E-\sqrt{\mu}\,\vec H=i\sqrt{\epsilon}\,(\vec E+ic\vec B)$ vanishes identically on a duality-pure field of the $+i$ sign, so that on that sign the field is carried by the pair $(\vec E,\vec B)$ and by no element of $\mathbb{B}$ at all. That is the same statement, and with $\mu_r=1$ and the magnetic field quoted in units of $\vec E/c$ the complexified field here obeys it: written with the complexified pair, $\tilde F=i\sqrt{\epsilon_0}\,(\vec E+i\vec H)=0$ exactly when the field is duality pure. Every density built from it, $\tilde\Sigma_{\Psi}=\tfrac12\tilde F\tilde F^{*}$ included, vanishes with it — not as a small residual but as an identity, because the object is zero. Checked here: $\lvert\tilde F\rvert=0$ and $\mathrm{Sc}(\tilde F\tilde F^{*})=0$ at both test fields, as they must be.
What survives, and what it is. Energy returns when the complex field is split into its real and imaginary parts, and the split is the corpus's material–informational division. Writing $\vec E=\vec E_r+i\vec E_i$ and $\vec H=\vec H_r+i\vec H_i$, the sesquilinear density of the complex field is exactly the sum of the densities of the two parts,
$$ \tfrac12\bigl(\epsilon\lvert\vec E\rvert^2+\mu\lvert\vec H\rvert^2\bigr) =\tfrac12\bigl(\epsilon\lvert\vec E_r\rvert^2+\mu\lvert\vec H_r\rvert^2\bigr) +\tfrac12\bigl(\epsilon\lvert\vec E_i\rvert^2+\mu\lvert\vec H_i\rvert^2\bigr), $$
and each part is a real Maxwell field obeying the ordinary conservation laws. At the Coulomb solution this reads $0.182+0.182=0.364$ in the units of this article, the real part being the electric Coulomb field of the charge and the imaginary part the dual monopole field of the same modulus.
So there is no contradiction, and the reason is worth keeping. The corpus's $T_{\mu\nu}$ is the energy of a real field; the source's vanishing bilinear is the statement that a duality-pure complex field has no energy of its own. Energy is not a property the complex field carries through the self-duality condition — it is what the condition's two real projections carry separately. That is the technical content of the source's jest about fields that "acquire energy only through self-division", and it is a sharper statement than it sounds: the division is forced, because the undivided object has no energy tensor at all.
Why the Elementary Charge Is the Quantum
The Coulomb scalar is the first member of a one-parameter family, $G_{\mathrm{ϙ}}=\mathrm{ϙ}\,\bar w/(z+r)$, and the charge is $\mathrm{ϙ}/4$. What fixes $\mathrm{ϙ}$ is not the wave equation. Two exact statements show it.
The wave equation and the eikonal are scale invariant. $\Box_\Phi G=0$ is linear and the eikonal is quadratic in the derivatives, so $G_{\mathrm{ϙ}}$ satisfies both for every $\mathrm{ϙ}$; the Coulomb-like field therefore exists for every charge. This is why the family is a family.
The first-order conditions are not. The programme's generating function is not an arbitrary harmonic scalar: it is subject to the two constraints
$$ \frac{\partial G}{\partial w}=G\,\frac{\partial G}{\partial u}, \qquad \frac{\partial G}{\partial v}=G\,\frac{\partial G}{\partial \bar w} , $$
which are what makes it come from the twistor, i.e. from the Kerr equation. They are linear in $G$ on the left and quadratic on the right, so a rescaling $G\mapsto\mathrm{ϙ} G$ multiplies the two sides by $\mathrm{ϙ}$ and $\mathrm{ϙ}^2$. For the family above the derivatives are
$$ \frac{\partial G_{\mathrm{ϙ}}}{\partial w}=-\frac{\mathrm{ϙ}\bar w^2}{2r(r_z+r)^2}, \qquad G_{\mathrm{ϙ}}\,\frac{\partial G_{\mathrm{ϙ}}}{\partial u}=-\frac{\mathrm{ϙ}^2\bar w^2}{2r(r_z+r)^2}, $$
so the first constraint is $\mathrm{ϙ}=\mathrm{ϙ}^2$, and the second gives the same, with $\partial G_{\mathrm{ϙ}}/\partial v=\mathrm{ϙ}\bar w/(2r(r_z+r))$ and $G_{\mathrm{ϙ}}\partial G_{\mathrm{ϙ}}/\partial \bar w=\mathrm{ϙ}^2\bar w/(2r(r_z+r))$. The only nonzero solution is $\mathrm{ϙ}=1$:
$$ \mathrm{ϙ}=1, \qquad |ϙ|=\frac14 . $$
So the elementary charge is the unique normalisation at which the two sides of a first-order condition have the same degree. The eikonal and the wave equation, both scale invariant, allow a continuum; the condition that is linear on one side and quadratic on the other allows one point of it. That is the exact form of the statement, recorded in the algebrodynamics article, that the primary system is not invariant under the scaling that Maxwell's equations admit — and it is why the fundamental solution carries the quantum and not a multiple of it. At $\mathrm{ϙ}=1$ the scalar is $\bar w/(z+r)$, the stereographic chart from the south pole. Its companion $\bar w/(z-r)$ satisfies the same two constraints, with the same constant $1/2$ in the algebra above and a singularity on the opposite half of the axis, so the constraints admit both signs. The companion is not the second stereographic chart: the programme's $f^-=w/(r-z)$ satisfies the eikonal — which is quadratic and so blind to conjugation — but fails the constraints by residuals $4.0$ and $9.3$ at a point where $\lvert f^-\rvert=2.3$, checked here. The constraints, unlike the eikonal, distinguish $G$ from $\bar G$, and that is the sharpest illustration of what the nonlinear condition does that the linear ones do not.
What the Reduced System Does Not Give
The reduced pair is implied by self-duality in the two-component gauge, and it is strictly weaker. A plane-wave superpotential
$$ G=\exp\,i(\vec k_\Phi\cdot\vec r-\omega h_t), \qquad \omega^2=\vec k_\Phi\cdot\vec k_\Phi , $$
satisfies the pair exactly, since it is $\Box_\Phi G=0$ together with an identity; the potential built from it, however, does not satisfy the duality-purity condition, and the failure is of the order of the field itself: checked here at five wave vectors, $\lvert\vec E+i\vec H\rvert$ is between $1.1$ and $1.3$ times $\lvert\vec E\rvert$, and $\lvert\vec E-i\vec H\rvert$ between $0.86$ and $0.96$ times it, so neither combination vanishes. The failure is not a small deformation to be removed by a better gauge, and the field is not a null field either: $\lvert\vec E\rvert/\lvert\vec H\rvert$ is between $2.0$ and $3.6$ across the same five wave vectors, where nullness would put it at $1$. It is a plain non-duality-pure complex field, and the ratio drifts without bound as $\vec k_\Phi$ approaches the third axis, where the potential degenerates and both fields vanish. This is the source's own remark that the corresponding Maxwell field is charge-free: it belongs to the other solution class, not to a deformation of this one. The wave equation is therefore not sufficient; the self-duality of the field is an extra condition on the scalar, and in the programme's construction that extra condition is the pair above together with the twistor, which is nonlinear.
This is the linear face of induced nonlinearity. The linear equations of the potential — the wave equation — admit far more solutions than the self-dual fields, and what restricts them is a nonlinear condition on the generating function, imposed not for convenience but as the definition of the admissible class. The corpus records the same phenomenon from the other side in The Algebrodynamical Programme: there the primary equation is nonlinear and the Maxwell equations are its consequence; here the Maxwell equation is linear and the nonlinearity is what selects its self-dual sector.
Summary
The complex potential of a self-dual Maxwell field is a $2\times2$ complex matrix whose entries are the four null-coordinate components $C_u=\Pi+C_z$, $C_w=C_x-iC_y$, $C_{\bar w}=C_x+iC_y$, $C_v=\Pi-C_z$. Duality purity, $\vec E+i\vec H=0$ with $\vec E=-\partial\vec C/\partial h_t-\vec\nabla_\Phi\Pi$ and $\vec H=\vec\nabla_\Phi\times\vec C$, together with the complex Lorentz gauge, is four complex first-order conditions on four complex unknowns; no real field can satisfy it, since the condition forces a real field to vanish. In the gauge $C_u=C_w=0$ the system reduces to $\partial C_{\bar w}/\partial \bar w=\partial C_v/\partial u$, $\partial C_{\bar w}/\partial v=\partial C_v/\partial w$, and with $C_{\bar w}=\partial G/\partial u$, $C_v=\partial G/\partial \bar w$ the first equation is the equality of mixed partials and the second is $\Box_\Phi G=0$: the reduced system is solved by one complex harmonic scalar, and the field strengths are its second derivatives.
The Coulomb member of the family is $G=\bar w/(z+r)$, giving $\vec E=-i\vec H=\vec r/(4r^3)$, the Coulomb field of a charge $\tfrac14$ with a magnetic partner of the same modulus. The family $G_{\mathrm{ϙ}}=\mathrm{ϙ}\bar w/(z+r)$ satisfies the wave equation and the eikonal for every $\mathrm{ϙ}$ and the programme's two first-order constraints only at $\mathrm{ϙ}=1$, because those constraints are linear on one side and quadratic on the other; that mismatch is the mechanism of charge quantisation, and it fixes $|ϙ|=\tfrac14$. The constraints admit $\bar w/(z+r)$ and $\bar w/(z-r)$ but not the programme's second eikonal solution $f^-=w/(r-z)$, which the quadratic eikonal cannot distinguish from its conjugate. The reduced system is strictly weaker than duality purity — a plane-wave scalar satisfies it and produces a field that is not duality pure — so the linear construction does not by itself characterise the physical class; the nonlinear condition does. One printed component of the source, $C_{\bar w}=-2ϙ\bar w/(z+r)$, is missing a factor $1/r$.
Energy. The bilinear density of a duality-pure complex field vanishes identically — invariant, Poynting vector and bilinear stress: $E_{j}E_{j}+H_{j}H_{j}=0$, $\vec E\times\vec H=0$, $E_{j}E_{\eta}+H_{j}H_{\eta}=0$, checked to $5.5\times10^{-17}$. The bilinear is the invariant form and is not the energy even for real fields, where its scalar part is $-\epsilon(I_1+2icI_2)$ while $2W=\epsilon E^2+\mu H^2$. In the corpus's element the same fact is $\tilde F=i\sqrt{\epsilon}(\vec E+i\vec H)=0$, so duality purity is the vanishing of the field-strength biquaternion and the corpus's carrier measures departure from purity. The sesquilinear density does not vanish; it is the sum of the densities of the real and imaginary parts, $0.182+0.182=0.364$ at the Coulomb solution, and each part is a real Maxwell field with ordinary conservation laws. The corpus's $T_{\mu\nu}$ is the tensor of such a part.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $\Pi$, $\vec C$ | Complex scalar and vector potentials, four complex components |
| $\vec E=-\partial\vec C/\partial h_t-\vec\nabla_\Phi\Pi$ | Complex electric field |
| $\vec H=\vec\nabla_\Phi\times\vec C$ | Complex magnetic field; opposite in sign to the source's printed definition, see the Introduction |
| $\vec N=\vec E+i\vec H$, $\vec N=0$ | The complex field as one vector and its duality purity; the corpus's anti-self-dual sign |
| $\partial\Pi/\partial h_t+\vec\nabla_\Phi\cdot\vec C=0$ | Complex Lorentz gauge |
| $u,v,w,\bar w$ | $u=t+z$, $v=t-z$, $w=x-iy$, $\bar w=x+iy$ |
| $C_u,C_v,C_w,C_{\bar w}$ | $\Pi+C_z$, $\Pi-C_z$, $C_x-iC_y$, $C_x+iC_y$; entries of the potential matrix |
| $\Box_\Phi=4(\partial^2/\partial u\,\partial v-\partial^2/\partial w\,\partial\bar w)$ | Wave operator in these coordinates |
| $G$ | Superpotential: complex scalar with $\Box_\Phi G=0$ generating the potential |
| $\vec E,\vec H$ | Complex field strengths of the construction, the source's tilde; $\mathbb C$-valued, and the corpus's real field is cited only where the text says so; vacuum, $\epsilon_r=\mu_r=1$, with the magnetic field in units of $\vec E/c$ |
| $C_{\bar w}=\partial G/\partial u$, $C_v=\partial G/\partial \bar w$ | The two surviving components in the gauge $C_u=C_w=0$ |
| $\partial G/\partial w=G\,\partial G/\partial u$, $\partial G/\partial v=G\,\partial G/\partial\bar w$ | First-order constraints; they make $G$ come from the twistor |
| $E_{+},E_{3},E_{-}$ | $E_{1}+iE_{2}$, $E_{3}$, $E_{1}-iE_{2}$; the spinor of the field strength |
| $ϙ$ | Electric charge; the Coulomb family has $\mathrm{ϙ}=4ϙ$ |
| $G_{\mathrm{ϙ}}=\mathrm{ϙ}\bar w/(z+r)$ | The Coulomb family; $\mathrm{ϙ}=1$ gives the elementary charge $\lvertϙ\rvert=\tfrac14$ |
| $r=\sqrt{r_x^2+r_y^2+r_z^2}$ | Radial coordinate |
| $\vec E_r,\vec E_i$ | Real and imaginary parts of $\vec E$; each is a real Maxwell field carrying its own energy |
| $\vec E\cdot\vec E+\vec H\cdot\vec H$ | The bilinear density, without conjugation; it is the invariant combination and vanishes on a duality-pure field |
| $I_1,I_2$ | The two invariants, $I_1=\vec E^2-c^2\vec B^2$ and $I_2=\vec E\cdot\vec B$; the bilinear's scalar part is $-\epsilon(I_1+2icI_2)$ |
Further Reading
- The Electromagnetic Energy–Momentum Tensor, for the sesquilinear tensor whose scalar part is $2W$ and whose time row is $\tilde\Sigma_{\Psi}$; the comparison in What the Complex Field Carries is with that construction.
- The Self-Dual and Anti-Self-Dual Split, for the reality obstruction that forces the potential here to be complex.
- The Algebrodynamical Programme, for the same construction in the programme's own terms, with the twistor and the nonlinear primary equation.
- V. V. Kassandrov, "Singular Sources of Maxwell Fields with Self-Quantized Electric Charge", in Has the Last Word Been Said on Classical Electrodynamics? (Rinton Press, 2004), 42–66, arXiv:physics/0308045, for the complex self-dual electrodynamics, the complex potential and the superpotential construction recorded here, and the charge-quantisation theorem that the construction inherits.
- V. V. Kassandrov, "Biquaternion Electrodynamics and the Weyl–Cartan Geometry of Space-Time", Gravitation & Cosmology 1 (1995) 216, arXiv:gr-qc/0007027, for the complex potential in the quaternionic setting and the closed-form static solution $f^\pm=(r_x\pm ir_y)/(r\pm r_z)$ that the Coulomb scalar reproduces.
- E. T. Newman, "Maxwell's Equations and Complex Minkowski Space", Journal of Mathematical Physics 14 (1973) 102–107, for the complexified potential and the generation of fields from a complex world-line, the complex-space counterpart of the superpotential here.
- Roger Penrose and Wolfgang Rindler, Spinors and Space-Time (Cambridge, 1984), for the symmetric two-spinor of a self-dual field strength and the Weyl equation that it obeys, which is the invariant content of the reduced system.
- Iwo Białynicki-Birula and Zofia Białynicka-Birula, "The role of the Riemann–Silberstein vector in classical and quantum theories of electromagnetism", Journal of Physics A 46 (2013) 053001, for the complex-vector description and the self-dual and anti-self-dual splitting.
- R. P. Kerr, "Gravitational field of a spinning mass as an example of algebraically special metrics", Physical Review Letters 11 (1963) 237–238, and G. C. Debney, R. P. Kerr and A. Schild, "Solutions of the Einstein and Einstein–Maxwell Equations", Journal of Mathematical Physics 10 (1969) 1842–1854, for the Kerr equation and the shear-free null congruence structure that lies behind the two first-order constraints.