The Lagrangian and the Dirac Equation in Momentum Space

Introduction

In momentum space the Dirac action stops being a functional of neighbouring events and becomes an algebraic quadratic form at each wave number. The Lagrangian and the Dirac Equation writes the action, varies it to the chiral pair, and derives the currents. This article performs the transform on that action and reads what the transform does to the three objects the variation rests on: the real part, the integration by parts and the Hessian.

Three things happen, and each is exact. The real part becomes redundant, because the momentum-space bracket is the expectation of a Hermitian element and is therefore real for every field; the verification covered one hundred random fields with generic complex coefficients, and the largest imaginary part was $2.2\times10^{-15}$. The integration by parts becomes unnecessary, because a multiplier needs none, so the action is evaluated at one wave number at a time with no time-slicing. And the second variation stops being an operator and becomes one central scalar, $\langle i\tilde K_\Phi\rangle_\natural-k_m^{2}=k^{2}-k_m^{2}$, so that the dispersion relation is a number's vanishing rather than a matrix's determinant.

The last of the three closes the parent's third open question in one precise respect. The parent records that "the action is the extremum's object; the path integral is the sum over configurations", and that the second is not attempted there. In momentum space the two objects stand in a definite relation: the variational output is the zero of the central scalar and the path integral's free propagator carries its pole. Zero and pole are one object read twice, and the relation is not an analogy.

What the article does not do is construct the path integral. The integral is the companion articles' work. What is added here is the momentum-space reading of the action and the statement that the two objects are one function's zero and its inverse.

The denial is stated plainly. The transform acts on $\mathbb{B}$-valued fields and returns $\mathbb{B}$-valued fields. The Hermiticity statements below are statements about the symbol's coefficients. There is no new prediction and no new solution; the action, the pair and the currents are the parent's.

Conventions

The algebra and the invertibles are those of Algebra and Element Conventions, the gradient those of Operator and Calculus Conventions, the products those of Product and Form Conventions, the two forms those of Norm and Pairing Conventions, the invertibles and the material time those of Physics Units and Common Elements Conventions. The transform is the one of Derivatives, Fields and Transforms, with the symbol rule

$$ \tilde\nabla_\Phi\longmapsto i\tilde K_\Phi , \qquad (\tilde\nabla_\Phi)^{\natural}\longmapsto i\tilde K_\Phi^{\natural}, \qquad \tilde K_\Phi=ik_t'e_t+\vec k_\Phi , \qquad k_t'=\frac{\omega}{c}. $$

The parent's action is quoted once and not re-derived. It is

$$ S=\frac{\hbar c}{2}\int\mathrm{d}^4r\;\mathrm{Re}\,\mathrm{Sc}\Bigl[\tilde\Xi_L^{*}\,\tilde\nabla_\Phi^{\natural}\tilde\Xi_L+\tilde\Xi_R^{*}\,\tilde\nabla_\Phi\tilde\Xi_R-k_m\bigl(\tilde\Xi_L^{*}\tilde\Xi_R+\tilde\Xi_R^{*}\tilde\Xi_L\bigr)\Bigr], $$

with $k_m=mc/\hbar$, the two parts $\tilde\Xi_L$ and $\tilde\Xi_R$ and the Hermitian pairing $\mathrm{Sc}(\cdot^{*}\cdot)$. The two parts are not values; the parent's warning applies unchanged, and no display below splits a value.

The Momentum-Space Action

The transform of the action replaces each derivative by a multiplier and each integration by parts by nothing. The action becomes

$$ S=\frac{\hbar c}{2}\int\frac{\mathrm{d}^4k}{(2\pi)^4}\;\mathrm{Sc}\Bigl[\tilde\Xi_L^{*}\,i\tilde K_\Phi^{\natural}\tilde\Xi_L+\tilde\Xi_R^{*}\,i\tilde K_\Phi\tilde\Xi_R-k_m\bigl(\tilde\Xi_L^{*}\tilde\Xi_R+\tilde\Xi_R^{*}\tilde\Xi_L\bigr)\Bigr], $$

with $\tilde\Xi_{L}$ and $\tilde\Xi_{R}$ now the transforms of the two parts. The measure is the transform's own constant and no statement below depends on it; the four terms are the parent's four, in the same order, with the multipliers in place of the derivatives.

Nothing differentiates anything. The action is a sum of terms, each of which pairs the field at one wave number with the field at the same wave number through one element of $\mathbb{B}$. There is no coupling between different wave numbers in the free case, so the action is a family of quadratic forms indexed by $\tilde K_\Phi$, and the variation is algebraic at each index.

That is what the transform buys, and it is a bookkeeping gain rather than a physical one. The physics is the same equation. The gain is that three questions — is the bracket real, does it need a boundary term, what is the second variation — become questions about one element at one wave number instead of questions about a functional.

The Real Part Is the Hermiticity of the Symbol

In position space the real part is a statement about the adjoint. The parent's derivation shows that each kinetic density is real up to a divergence, that the divergence integrates away, and that the imaginary part of the bracket is therefore a surface term. One sentence of it is worth quoting: the $i$ inside the material time is what makes the temporal kinetic term self-adjoint without disturbing the spatial ones.

In momentum space the real part is a statement about the multiplier, and it is the same statement. Take any element $\tilde A$ of $\mathbb{B}$ with complex coefficients and a wave-number element $\tilde K_\Phi$ off the shell. Then

$$ \mathrm{Sc}\bigl(\tilde A^{*}\,i\tilde K_\Phi\,\tilde A\bigr) =\overline{\mathrm{Sc}\bigl(\tilde A^{*}\,i\tilde K_\Phi\,\tilde A\bigr)} . $$

because $i\tilde K_\Phi$ is Hermitian: $\mathrm{Sc}(\tilde A^{*}H\tilde A)$ is real for every Hermitian $H$ and every $\tilde A$. The check covered one hundred random $\tilde A$ and as many random momenta, and the largest imaginary part of the bracket was $2.2\times10^{-15}$. So the $\mathrm{Re}$ of the momentum-space action removes nothing, and writing it or not writing it is the same functional.

The two readings are one mechanism seen twice. In position space the central $i$ sits in the coordinate $R_t=ir_t'$ and splits the four derivative directions into one anti-Hermitian and three Hermitian; the real part then cancels the one imaginary contribution. In momentum space the same $i$ sits in front of the symbol and makes all four directions Hermitian at once; the real part then has nothing to cancel. One element, used in one place at a time, and the real part is the trace of where it was used.

A reader should not conclude that the real part was a decoration. Without it the position-space functional is not real, and the variation is not the pair; the parent shows both. The claim here is narrower: the real part's work is done by the symbol's Hermiticity once the transform is taken, so the momentum-space action needs no boundary argument and no surface term.

The Second Variation Is One Central Scalar

Compose the two members of the pair and the operator's square is a number. In momentum space the pair is

$$ i\tilde K_\Phi\,\tilde\Xi_R=k_m\tilde\Xi_L, \qquad i\tilde K_\Phi^{\natural}\,\tilde\Xi_L=k_m\tilde\Xi_R . $$

Substituting the first into the second gives $i\tilde K_\Phi^{\natural}i\tilde K_\Phi\tilde\Xi_R=k_m^{2}\tilde\Xi_R$, and the left side is central:

$$ i\tilde K_\Phi^{\natural}\,i\tilde K_\Phi =-\tilde K_\Phi^{\natural}\tilde K_\Phi =-\tilde K_\Phi\tilde K_\Phi^{\natural} =\langle i\tilde K_\Phi\rangle_\natural =k^{2}, \qquad k^{2}=k_t'^{2}-k_\Phi^{2}, $$

where both minus signs are $i^{2}$ and the middle step is the norm's order-independence. So the second variation of the action at a mode is the central scalar

$$ \langle i\tilde K_\Phi\rangle_\natural-k_m^{2}=k^{2}-k_m^{2}. $$

The check is exact and it covered three momenta. On $k^{2}=7$ and on $k^{2}=0.4375$, $\langle\tilde K_\Phi\rangle_\natural$ came out $-k^{2}$, $\langle i\tilde K_\Phi\rangle_\natural$ came out $+k^{2}$, and $i\tilde K_\Phi^{\natural}i\tilde K_\Phi$ came out $+k^{2}$, each with zero imaginary part; on the null momentum $k^{2}=0$ all three vanished.

Centrality is the point. The second variation of a biquaternionic action is normally an element or an operator, because the Hessian of a functional on a non-commutative algebra is a bilinear form that need not be central. Here it is a number. That is why the corpus can write the mass shell as a scalar equation $\tilde K_\Phi\tilde K_\Phi^{\natural}=-k_m^{2}$ rather than as the vanishing of a determinant, and it is why the dispersion relation of the Dirac equation in this algebra is one equation and not a spectral problem.

The number is the one the variational parent's action was built to produce. The parent's variation gives the pair; composing the pair gives this scalar; the scalar's vanishing is the shell. The action therefore knows the shell twice, once through each member, and the two readings agree because the norm is central.

Zero and Pole Are One Object

The variational output is the zero. The pair's compatibility condition is $(k^{2}-k_m^{2})\tilde\Xi_R=0$, so the solutions of the equation of motion are supported exactly where the central scalar vanishes. The variational principle's content in momentum space is therefore: on the shell, any amplitude solves the pair; off it, none does.

The path integral's free propagator has the pole. The path-integral companion writes the free Dirac kernel in momentum space with the denominator $\langle\tilde K_\Phi\rangle_\natural+k_m^{2}$, which is $-k^{2}+k_m^{2}$, the negative of the second variation. The kernel is large exactly where the scalar vanishes, and its residue there is the numerator the companion computes.

The two are one function, read at its zero and at its inverse. Writing $N(\tilde K_\Phi)=k^{2}-k_m^{2}$, the variational statement is $N=0$ and the propagator is proportional to $1/N$. Neither statement is more fundamental, and neither implies the other without the construction in between: the zero selects the support and the pole gives the weight.

This is the precise sense in which the parent's third open question is closed. The parent asked for the path integral and did not attempt it. The answer attempted here is not the integral; it is the observation that the two objects the parent separated — the extremum's object and the sum over configurations — meet in one momentum-space function, one at its zero and the other at its pole. The construction of the sum is the companion articles' work, and the relation is the reason the two articles describe one theory.

Limits

Constant coefficients. The momentum-space action is a family of quadratic forms only where the symbol is constant. The mass is constant, so the free action transforms exactly and every display above is exact. With a potential the symbol depends on the position, the transform of the action is a convolution, the Hessian is an operator again, and the central-scalar statement stops. The Lagrangian and the Coupled Dirac Equation and The Minimal Coupling of the Biquaternion Dirac Field to Electromagnetism are where that case is treated, and nothing here anticipates them.

Where the mass sits. The second variation is $\langle i\tilde K_\Phi\rangle_\natural-k_m^{2}$, with the mass subtracted as a central number. It is not $\langle i\tilde K_\Phi-k_m\rangle_\natural$, and the difference is not small:

$$ \langle i\tilde K_\Phi-k_m\rangle_\natural =k^{2}+2k_mk_t'+k_m^{2}, $$

checked to $3.9\times10^{-16}$ over five momenta. On $k_t'=1$, $\vec k_\Phi=(0.6,0,0)$ and $k_m=0.7$ the wrong object is $2.53$ and the right one is $0.64-0.49=0.15$. The norm is not linear in its argument, so the mass cannot sit inside it; it is a coefficient of the action, and the composition of the pair subtracts it from a number.

The measure. The momentum-space measure's constant is the transform's convention, and the action's overall constant is the parent's $\frac{\hbar c}{2}$. No display above depends on either, and no normalization is claimed.

The denial, once more. The two parts remain elements of $\mathbb{B}$, the pair remains a pair of equations on elements, and the centrality of the second variation is a statement about one coefficient's behaviour. No sector flip occurs at any step.

Open Questions

The article leaves no question of its own open. The three statements are identities and exact checks, and the zero-pole relation is a computed equality of two displays.

Two of the parent's questions are untouched, and they are not this article's. Which reading of the electric current is physical, and whether the operator setting closes for a position-dependent Hamiltonian, remain the parent's first and second open questions; nothing above bears on either.

One neighbouring question is now well posed. Whether the zero-pole relation survives when the symbol is an operator — that is, whether the path integral of the coupled theory is the inverse of the coupled action's Hessian in the same sense — is a question about the coupled articles. It is stated here because the free case makes it precise, and it is not answered here.

Summary

In momentum space the Dirac action of the variational parent becomes one algebraic quadratic form per wave number, with each derivative replaced by a multiplier and no integration by parts anywhere. The real part of the position-space action becomes redundant, because the bracket is $\mathrm{Sc}(\tilde A^{*}i\tilde K_\Phi\tilde A)$ for a Hermitian multiplier, which is real for every field: one hundred random fields gave a largest imaginary part of $2.2\times10^{-15}$.

The second variation is a single central scalar. Composing the pair gives $i\tilde K_\Phi^{\natural}i\tilde K_\Phi=\langle i\tilde K_\Phi\rangle_\natural=k^{2}$, so the second variation at a mode is $k^{2}-k_m^{2}=\langle i\tilde K_\Phi\rangle_\natural-k_m^{2}$, with no imaginary part and no matrix left to diagonalise. That centrality is why the mass shell is a scalar equation in this algebra and not a spectral condition.

The variational output is the zero of that scalar and the free propagator is proportional to its inverse: the path-integral companion's denominator $\langle\tilde K_\Phi\rangle_\natural+k_m^{2}$ is its negative. Zero and pole are one object, and that relation is the precise sense in which the variational parent's unattempted path integral is accounted for, without the integral being constructed here. The mass is subtracted as a coefficient and does not enter the norm: $\langle i\tilde K_\Phi-k_m\rangle_\natural=k^{2}+2k_mk_t'+k_m^{2}$, which is not the dispersion relation.

Summary of Notation

Symbol Meaning
$S$ The parent's action, four terms, coefficient $\frac{\hbar c}{2}$
$i\tilde K_\Phi$, $i\tilde K_\Phi^{\natural}$ The symbols of the gradient and its conjugate; Hermitian
$\mathrm{Sc}(\tilde A^{*}i\tilde K_\Phi\tilde A)$ The momentum-space kinetic bracket; real for every $\tilde A$
$\langle i\tilde K_\Phi\rangle_\natural=k^{2}$ The symbol's norm; central, real, equal to the interval
$k^{2}=k_t'^{2}-k_\Phi^{2}$ The interval in wave numbers
$\langle i\tilde K_\Phi\rangle_\natural-k_m^{2}$ The second variation at a mode; one central scalar
$\langle i\tilde K_\Phi-k_m\rangle_\natural=k^{2}+2k_mk_t'+k_m^{2}$ The wrong object; the mass inside the norm
$\langle\tilde K_\Phi\rangle_\natural+k_m^{2}=-(k^{2}-k_m^{2})$ The propagator's denominator; the pole
$N(\tilde K_\Phi)=k^{2}-k_m^{2}$ One function, read at its zero and at its pole

Further Reading

  • The Lagrangian and the Dirac Equation, for the action, the variation, the currents and the three open questions.
  • The Fourier Transform and the Dirac Equation in Biquaternionic Form, for the symbol, its Hermiticity and the light cone as the idempotent variety.
  • The Dirac Equation in Biquaternionic Form, for the chiral pair and the mass coefficient.
  • The Path Integral and the Dirac Equation in Biquaternionic Form, for the free propagator whose denominator is read here as a pole.
  • The Path Integral and the Dirac Equation in Biquaternionic Polar Variables, for the same kernel in the polar word.
  • The Lagrangian and the Coupled Dirac Equation, for the case in which the symbol is not constant.
  • Derivatives, Fields and Transforms, for the transform and its symbol rule.
  • Norm and Pairing Conventions, for the two forms, the norm's centrality and the Hermitian pairing.
  • P. A. M. Dirac, "The Lagrangian in quantum mechanics," Physikalische Zeitschrift der Sowjetunion 3 (1933) 64–72, for the action's place in the integral and its relation to the classical extremum.
  • R. P. Feynman, "Space-time approach to non-relativistic quantum mechanics," Reviews of Modern Physics 20 (1948) 367–387, for the sum over configurations the relation is about.
  • J. H. Van Vleck, "The correspondence principle in the statistical interpretation of quantum mechanics," Proceedings of the National Academy of Sciences 14 (1928) 178–188, for the fluctuation determinant that a second variation produces in the position-space form.
  • C. Itzykson and J.-B. Zuber, Quantum Field Theory (McGraw-Hill, 1980), Chapter 2, for the free Dirac propagator's denominator and its residue in the standard language.