The Lagrangian and the Klein–Gordon Equation in Momentum Space
Introduction
The variational parent states one sentence in passing and drops it. At its §The Action on the Carrier and on the Algebra it writes that for a constant $\tilde M$ the plane-wave problem becomes the eigenvalue problem $(\omega^{2}-c^{2}k_\Phi^{2})\tilde\Xi_0=\tilde M\tilde\Xi_0$, "whose four branches are the four eigenvalues". The sentence is stated, used as a remark about what a non-central coefficient would do, and not developed. This article develops it.
The finding is a count, and the count is not four. Under the transform the equation becomes $(\omega^{2}/c^{2}-k_\Phi^{2})\tilde\Xi_0=\tilde M\tilde\Xi_0$, an eigenvalue problem for one operator on $\mathbb{B}$. For a Hermitian $\tilde M=a\,e_t+i\vec v\cdot\vec e$ the spectrum of left multiplication is $a+\lvert\vec v\rvert$ and $a-\lvert\vec v\rvert$, each with multiplicity two. There are two eigenvalues and four frequencies, and the doubling is not the sector pair.
The governing thesis is one sentence. The free Klein–Gordon symbol is a central scalar; it commutes with everything, including the sector projectors; therefore the free scalar theory carries no sector content at all, and every sector statement at Klein–Gordon level is a statement about the potential. This is why the article is about $\tilde M$ and not about $\tilde\nabla_\Phi$. The Dirac family gets its sector selection from the Hermiticity of a first-order symbol; the scalar family gets it from the reality of the second-order action.
The selection rule has a direction. The potential term of the action is $\mathrm{Sc}(\tilde\Xi^{*}\tilde M\tilde\Xi)$. That number is real when $\tilde M$ is Hermitian and purely imaginary when $\tilde M$ is anti-Hermitian. So the material sector $\mathbb{B}_\Phi$ is excluded from the Klein–Gordon Lagrangian by the reality of $S$, and not by taste. The formal consequence of admitting it — a purely imaginary spectrum and a complex dispersion — is stated after the exclusion and not before it.
The potential also sets a threshold. The lower branch $a-\lvert\vec v\rvert$ is negative whenever the potential's vector part exceeds its scalar part, and then the frequency is imaginary at small wave number. One inequality, one threshold, and a measured frequency of occurrence.
The companion asks a different question of the same equation. The Fourier Transform and the Klein–Gordon Equation in Biquaternionic Form asks when the second-order symbol factors into first-order ones, and answers through the zero set rather than the spectrum. This article asks what the eigenvalues are. Neither answer depends on the other.
Conventions
The algebra and its units are those of Algebra and Element Conventions, the gradient, the d'Alembertian and the symbol rule those of Operator and Calculus 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 kernel and its $i\epsilon$ prescription are those of The Klein–Gordon Propagator and Its Green's Functions in Biquaternionic Form.
The bookkeeping is stated once. With $\tilde K_\Phi=ik_t'e_t+\vec k_\Phi$, $k_t'=\omega/c$, $k_m=mc/\hbar$ and $r_t'=ct'$,
$$ \langle\tilde K_\Phi\rangle_\natural=-k_t'^{2}+k_\Phi^{2}, \qquad \langle\tilde P_\Phi\rangle_\natural=N, \qquad \tilde\nabla_\Phi\longmapsto i\tilde K_\Phi, \qquad \tilde\nabla_\Phi(\tilde\nabla_\Phi)^{\natural}\longmapsto-\langle\tilde K_\Phi\rangle_\natural . $$
The phase carries the sign and the sign is quoted. A positive-frequency plane wave is $\tilde\Xi=\tilde\Xi_0\exp(-i\theta_{\mathrm{pwp}})$ with $\theta_{\mathrm{pwp}}=\omega r_t'-\vec k_\Phi\cdot\vec r$. On it $\partial/\partial r_t'\mapsto-i\omega$ and $\partial/\partial r_\eta\mapsto ik_\eta$, so $\Box_\Phi\mapsto k_t'^{2}-k_\Phi^{2}$, which is the parent's own computation and is used here without alteration.
The potential is a constant element. Throughout, $\tilde M$ is a constant element of $\mathbb{B}$: Hermitian, $\tilde M=a\,e_t+i\vec v\cdot\vec e$ with $a$ and $\vec v$ real, or anti-Hermitian, $\tilde M=i\,a\,e_t+\vec v\cdot\vec e$. Its left multiplication $L(\tilde M)$ is complex-linear on the four-dimensional complex space $\mathbb{B}$, because $i$ is central, so it is a $4\times4$ complex matrix in the basis $e_\mu$ and its eigenvalues are ordinary complex eigenvalues.
The Momentum-Space Equation Is an Eigenvalue Problem
Finding 1. Substituting $\tilde\Xi=\tilde\Xi_0\exp(-i\theta_{\mathrm{pwp}})$ into $(\Box_\Phi-\tilde M)\tilde\Xi=0$ and using the symbol rule gives, in one line,
$$ \left(\frac{\omega^{2}}{c^{2}}-k_\Phi^{2}\right)\tilde\Xi_0=\tilde M\,\tilde\Xi_0 . $$
The right side is left multiplication, and that is the whole content. The parent's position-space equation has $\tilde M$ multiplying the field at each event; its transform is the same left multiplication at each wave number, because a constant element has no position dependence to transform. The equation is therefore an eigenvalue problem with $\tilde M$ as the operator and $\omega^{2}/c^{2}-k_\Phi^{2}$ as the eigenvalue.
The parent's display has a bookkeeping defect, and it is repaired here rather than copied. The parent prints $(\omega^{2}-c^{2}k_\Phi^{2})\tilde\Xi_0=\tilde M\tilde\Xi_0$. Take $\tilde M=k_m^{2}e_t$, which the parent's own §The Two Settings permits: the right side then has dimension $\mathrm{L}^{-2}$, because $k_m$ is an inverse length, while the left side has dimension $\mathrm{T}^{-2}$, because $\omega$ is an inverse time. The two differ by $c^{2}$. The parent's own §The Equation and Its Operator fixes $k_t'^{2}-k_\Phi^{2}=+k_m^{2}$ with $k_t'=\omega/c$, which is the corrected display. The parent's eigenvalue form is therefore its $c=1$ reading, and the article uses the corrected form throughout.
The repair is measured, not asserted. On the case $c=2$, $k_m=1.4$, $\vec k_\Phi=(0.9,-0.5,0.3)$ and a Hermitian potential $\tilde M=k_m^{2}e_t+i(0.3,-0.2,0.5)\cdot\vec e$, the two eigenvalues of $\tilde M$ are $2.576441$ and $1.343559$. On the corresponding solutions the corrected left side returns those two numbers to all printed digits; the parent's expression $\omega^{2}-c^{2}k_\Phi^{2}$ returns $10.305766$ and $5.374234$, which are the same numbers multiplied by $c^{2}=4$. The residual of the corrected form is $0.0$ and the discrepancy of the uncorrected form is exactly the factor, so the defect is a bookkeeping slip and not a physical disagreement.
What the transform buys is a diagonal basis for the coupling. The parent says that a non-central $\tilde M$ turns the four independent coefficients into four coupled channels. In momentum space the coupling is one $4\times4$ matrix acting on one four-component amplitude, and the channels are its eigenvectors. The transform does not remove the coupling; it exhibits the basis in which the coupling is diagonal, and that basis is available because $\tilde M$ is constant.
The eigenvalue is a frequency squared, so the count doubles again. Each eigenvalue $\lambda$ of $\tilde M$ gives $\omega^{2}=c^{2}(k_\Phi^{2}+\lambda)$ and therefore two frequencies $\pm c\sqrt{k_\Phi^{2}+\lambda}$, one per sign of $\omega$. The eigenvalue problem fixes the dispersion relation; it does not fix the sign of the frequency, and no choice of $\vec k_\Phi$ removes the negative root.
The licence travels with the finding: everything in this section needs $\tilde M$ constant, because the symbol exists only there, and a position-dependent $\tilde M$ replaces the matrix by a convolution with no eigenvalue problem at all.
The Spectrum of Left Multiplication
Finding 2. For a Hermitian potential $\tilde M=a\,e_t+i\vec v\cdot\vec e$ with $a$ and $\vec v$ real, the spectrum of $L(\tilde M)$ is
$$ \operatorname{eig}L(\tilde M)=\bigl\{a+\lvert\vec v\rvert,\;a+\lvert\vec v\rvert,\;a-\lvert\vec v\rvert,\;a-\lvert\vec v\rvert\bigr\}. $$
The proof is one algebraic identity. Put $q=\vec v\cdot\vec e$, the pure vector part. Then $q^{2}=-\lvert\vec v\rvert^{2}e_t$, so $L(q)^{2}=-\lvert\vec v\rvert^{2}I$, so $L(q)$ has the two eigenvalues $\pm i\lvert\vec v\rvert$, each twice. Since $i$ is central, $L(iq)=iL(q)$, and the two eigenvalues become $\mp\lvert\vec v\rvert$. Adding the central $a$ gives $a\pm\lvert\vec v\rvert$. The minimal polynomial of $L(\tilde M)$ is $(L-a)^{2}-\lvert\vec v\rvert^{2}$, of degree two, on a four-dimensional space, so each root has algebraic multiplicity exactly two.
The count is measured. Over three hundred random draws of $(a,\vec v)$ the sorted computed spectrum matched the sorted display to $3.1\times10^{-15}$, and the characteristic polynomial of the computed spectrum matched the polynomial $(x-a)^{2}-\lvert\vec v\rvert^{2}$ squared, coefficient by coefficient, to $1.6\times10^{-13}$. The doubling is therefore not a numerical accident of the examples.
Two eigenvalues, four frequencies, and the parent's "four branches" is a count of the second kind. The parent counts four eigenvalues. Two of them are distinct and each is doubled. If one counts the frequencies $\pm c\sqrt{k_\Phi^{2}+\lambda}$ instead, one gets four values, each carrying the multiplicity of its $\lambda$. Both counts reach four; neither is a count of four independent branches, and the phrase "four branches" hides which count is meant. The article writes "two eigenvalues, each doubled" and "four frequencies, each doubled" and uses no other count.
The doubling is the central complex structure, and the sectors sit inside each branch. Because $i$ is central, $L(\tilde M)(i\tilde\Xi)=iL(\tilde M)\tilde\Xi$, so every eigenspace is closed under multiplication by $i$. Because left multiplication commutes with right multiplication, the eigenspace of $\lambda$ is the right ideal $\mathbb{B}w$ generated by any one eigenvector $w$. That ideal contains $ww^{*}$, which is Hermitian, and $i\,ww^{*}$, which is anti-Hermitian. So each branch carries both sectors, and the sector split does not separate the branches. An eigenspace cannot lie inside one sector, because it is closed under multiplication by $i$ and $i$ exchanges the sectors. The smallest Hermiticity defect attainable inside an eigenspace was $0.000$ over forty branches, reached by $ww^{*}$ itself, and no eigenspace is a sector.
That last sentence is the momentum-space form of the parent's own verdict. The parent's §Does the Second-Order Structure Fit $\mathbb{B}_{\Phi}/\mathbb{B}_{\Psi}$? concludes that the split of a Klein–Gordon field into $\mathbb{B}_\Psi$ and $\mathbb{B}_\Phi$ is the real-imaginary decomposition of a complex field, two $i$-related copies of one equation. Here the same fact appears as a property of an eigenspace: the space is $i$-invariant, so it carries both sectors and is identical with neither.
The anti-Hermitian case has the same shape with $i$ in front. For $\tilde M=i\,a\,e_t+\vec v\cdot\vec e$ the same identity gives $\operatorname{eig}L(\tilde M)=i(a\pm\lvert\vec v\rvert)$, each twice. Over three hundred draws the real part of every eigenvalue was below $8.9\times10^{-16}$ and the sorted spectrum matched $i(a\pm\lvert\vec v\rvert)$ to $3.6\times10^{-15}$. The spectrum is purely imaginary, and that is exactly what disqualifies the case in the next section.
Reality of the Action and the Selection Rule
Finding 3. The potential term of the action is $\mathrm{Sc}(\tilde\Xi^{*}\tilde M\tilde\Xi)$, and its reality is equivalent to $\tilde M$ being Hermitian. The parent's kinetic term is already real for every field, so the potential term is the only place where a sector question can arise, and this is what it decides.
The proof is the Hermiticity of the pairing. The form $\langle A,B\rangle=\mathrm{Sc}(A^{*}B)$ is a positive-definite Hermitian form on the complex vector space $\mathbb{B}$. For a Hermitian $\tilde M$ one has $\langle\tilde\Xi,\tilde M\tilde\Xi\rangle=\langle\tilde M\tilde\Xi,\tilde\Xi\rangle$, so the number is real. For an anti-Hermitian $\tilde M$ the same computation gives $\langle\tilde\Xi,\tilde M\tilde\Xi\rangle=-\langle\tilde M\tilde\Xi,\tilde\Xi\rangle$, so the number is purely imaginary. There is no third case: the generic element splits uniquely into a Hermitian part and an anti-Hermitian part, $\tilde Q=\tilde Q_\Psi+\tilde Q_\Phi$ with $\tilde Q_\Psi=\tfrac12(\tilde Q+\tilde Q^{*})$ and $\tilde Q_\Phi=\tfrac12(\tilde Q-\tilde Q^{*})$, and the quadratic form splits with it into a real and an imaginary piece. The potential's own glyph is not doubled to say the same thing: it is Hermitian or anti-Hermitian as it stands, and when it is neither, the two parts are written $\tfrac12(\tilde M+\tilde M^{*})$ and $\tfrac12(\tilde M-\tilde M^{*})$.
The count confirms both halves. Over three hundred random pairs of a field and a Hermitian potential, the quantity $\mathrm{Sc}(\tilde\Xi^{*}\tilde M\tilde\Xi)$ had an imaginary part below $3.6\times10^{-15}$. Over three hundred random pairs with an anti-Hermitian potential, its real part was below $7.1\times10^{-15}$. The two numbers are the two halves of the selection rule, and they are structural rather than numerical: no field can defeat them.
So the material sector is excluded, and the exclusion is a selection rule. A Klein–Gordon Lagrangian whose potential is anti-Hermitian, $\tilde M=\tfrac12(\tilde M-\tilde M^{*})\in\mathbb{B}_\Phi$, has an action whose imaginary part is $\int\mathrm{Sc}(\tilde\Xi^{*}\tilde M\tilde\Xi)$, which is not a boundary term and does not vanish. A complex action is not a variational principle with real stationary points; it is a saddle-point problem, and its stationarity conditions are not the equation the parent derives. The exclusion is therefore not a restriction of convenience. It is the condition under which the action of §The Action in Element Form exists at all.
The formal consequence, stated after the exclusion and not before. If one keeps the anti-Hermitian potential anyway, the equation still has plane-wave solutions, because the algebra does not care about the reality of an integral. Their eigenvalues are purely imaginary, by §The Spectrum of Left Multiplication, so
$$ \frac{\omega^{2}}{c^{2}}=k_\Phi^{2}+i\,\lambda, \qquad \lambda\in\mathbb{R}, $$
The result is not a wave. The frequency is complex for every wave number, and a complex frequency is growth or decay in time with no propagation threshold, so the mode is not a wave in any sense the parent's dispersion relation recognises. That is the formal content of the excluded case; it is recorded so that a reader who meets it elsewhere can identify it, not as a candidate theory.
The comparison with the Dirac family is exact and it is instructive. There the symbol is first order and the object that must be Hermitian is the symbol itself, because a first-order equation is a flow and a flow wants a unitary generator. Here the equation is second order, its action contains $\tilde M$ quadratically, and the object that must be real is the action's density. Same selection rule, different quantity selected. The scalar family selects on $S$ and the spinor family selects on $H$, and neither selection is available in the other family. The free case is the boundary of the rule: with $\tilde M=k_m^{2}e_t$ the potential is central and Hermitian at once, the anti-Hermitian part vanishes, and both sectors pass the test, which is why the free theory cannot tell them apart.
The Tachyon Threshold
Finding 4. The lower branch $a-\lvert\vec v\rvert$ is the threshold, and the condition for a non-negative spectrum is one inequality:
$$ a\geq\lvert\vec v\rvert . $$
The mechanism is one substitution. With $\lambda=a-\lvert\vec v\rvert$ the dispersion is $\omega^{2}=c^{2}(k_\Phi^{2}+a-\lvert\vec v\rvert)$. At small wave number the bracket is dominated by $a-\lvert\vec v\rvert$; when that is negative, $\omega^{2}$ is negative for every $k_\Phi$ below $\sqrt{\lvert\vec v\rvert-a}$, and $\omega$ is imaginary there. The mode is evanescent in time: it does not oscillate, it grows or decays.
The threshold itself is the massless case. At $a=\lvert\vec v\rvert$ the lower branch vanishes and its dispersion is $\omega^{2}=c^{2}k_\Phi^{2}$, the cone. The upper branch has $\lambda=2\lvert\vec v\rvert$ and its dispersion is the massive one with $k_m^{2}=2\lvert\vec v\rvert$. So the two branches are one massless and one massive cone, and the potential's two parts decide which is which.
The occurrence is common, and the count says so. Over three hundred draws with $a$ standard normal and $\vec v$ standard normal in three dimensions, the lower branch was negative in $276$ of them. The vector part of a random Hermitian potential has three components and its length is typically larger than one component, so a positive scalar part does not by itself keep the spectrum non-negative. A potential whose scalar part is not larger than the length of its vector part has a tachyon branch, and that is the generic case for unconstrained coefficients. What the threshold is not: it is not a bound on the energy of a state, and it is not a statement about the sign of the charge. It is a statement about the sign of one eigenvalue of one operator, and its only consequence is the sign of $\omega^{2}$ at small $k_\Phi$. The parent's two defects — indefinite charge and negative energies — are separate, and neither is decided by this inequality.
Why the Free Case Says Nothing
Finding 5. With $\tilde M=k_m^{2}e_t$ the four eigenvalues coincide at $k_m^{2}$, the symbol commutes with every element of $\mathbb{B}$, and no sector statement can be made. The eigenvalue problem degenerates from a matrix problem to a multiple of the identity. The count is exact and it is the point: the computed spectrum was $k_m^{2}$ four times with a residual below $10^{-15}$, and the commutator of the symbol with fifty random elements of $\mathbb{B}$ was $0.0$ in every component. A central symbol commutes with the sector projectors, so $\Box_\Phi-k_m^{2}$ acts identically on $\mathbb{B}_\Psi$ and on $\mathbb{B}_\Phi$. The equation cannot tell the sectors apart because its operator does not contain them.
This is the parent's conclusion reached in one line. The parent's §Does the Second-Order Structure Fit $\mathbb{B}_{\Phi}/\mathbb{B}_{\Psi}$? argues in position space that the sector split gives two $i$-related copies of the same equation and is therefore not the doubling the equation needs. In momentum space the same statement is a property of a scalar multiple of the identity: an operator proportional to $I$ has one eigenspace, which is the whole space, and the whole space is where the sectors live.
The thesis of the article is this finding read backwards. Because the free symbol is central and carries no sector content, every sector statement at Klein–Gordon level must come from the potential. The article's four other findings are the four ways in which it does: the spectrum, the multiplicity, the reality, and the threshold. Remove the potential and all four collapse to one number.
The Denial That Survives
The informational sector is not the Fourier dual of the material sector. Every transform in this article acts on a $\mathbb{B}$-valued function and returns a $\mathbb{B}$-valued function. No step carries a field out of its sector, and no display above identifies a coefficient's sector with a function's sector. The sector statements are statements about one constant element that multiplies, at one wave number or at one event.
A second denial belongs here because the eigenvalue problem invites the error: the eigenvalues of $L(\tilde M)$ are complex numbers, and their real or imaginary character is a statement about $\tilde M$'s sector and not about the field's. A field in $\mathbb{B}_\Psi$ and a field in $\mathbb{B}_\Phi$ feel the same matrix, because the matrix is left multiplication on the whole algebra; what differs is only which sector their values occupy, and that is not an eigenvalue.
Limits
Constant coefficients are the licence. The eigenvalue problem needs $\tilde M$ constant, because the symbol exists only there. A position-dependent Hermitian $\tilde M$ keeps the equation and loses the spectrum: the momentum-space operator becomes a convolution, its kernel is the transform of $\tilde M$, and there is no matrix and no eigenvalue. The parent's constant setting is therefore not a simplification of convenience but the only setting in which the four branches exist as branches.
The abelian component form matters for the same reason. The parent separates the covariant derivative into components because the coefficients are central scalars, and that is what makes $L(\tilde M)$ a single constant matrix rather than a connection. A non-abelian coefficient would add a commutator inside the potential term and the eigenvalue problem would become a matrix-valued one with no scalar spectrum.
The companion's question is the zero set, and it is not this article's. The Fourier Transform and the Klein–Gordon Equation in Biquaternionic Form asks when the second-order symbol factors into first-order ones and answers through the reducibility of its quadric. A spectrum and a factorization are different properties of the same symbol: an operator can have a perfectly good spectrum and no square root, which is exactly the Klein–Gordon case, and the scalar Klein–Gordon operator has both a known spectrum and a known obstruction.
The third article of the family reads the polar word, and this one does not. The Fourier Transform and the Klein–Gordon Equation in Biquaternionic Polar Variables asks where the transform's phase comes from and what removes the symbol; its answer is about the amplitude, and it is not a restatement of anything here. Nothing in the article is a transform between sectors: every object above is an element of $\mathbb{B}$ or a coefficient of one, and the transform returns the same algebra.
Open Questions
One question is left open, and it is about a potential the selection rule might admit. The exclusion of §Reality of the Action and the Selection Rule is the exclusion of an anti-Hermitian part, because that part makes the action's density imaginary. An anti-Hermitian part that is a total derivative — that is, one whose contribution to the action is a boundary term — would make the action real up to a surface term, and the variation would be unaffected. Whether such a potential exists in this algebra, whether it can be written as $\tilde\nabla_\Phi\tilde\Gamma$ for some $\tilde\Gamma$, and whether its spectrum is then admissible, is not decided here. No claim is made in either direction.
A second question follows from the doubling, and it is narrower. The eigenspace of a branch is a right ideal $\mathbb{B}w$, and its Hermitian representative is $ww^{*}$. Whether the physical restriction to a Peirce carrier — the question the parent's §Open Questions raises for a non-central coefficient — can be imposed inside one eigenspace without breaking the $i$-invariance that produces the doubling, is open as well.
Summary
The transform of the variational parent's coupled four-channel equation is one eigenvalue problem, $(\omega^{2}/c^{2}-k_\Phi^{2})\tilde\Xi_0=\tilde M\tilde\Xi_0$, with the parent's $c^{2}$ restored: the parent prints $\omega^{2}-c^{2}k_\Phi^{2}$, whose dimension is $\mathrm{T}^{-2}$ against the potential's $\mathrm{L}^{-2}$, and at $c=2$ the two forms differ by the factor $4$.
For a Hermitian $\tilde M=a\,e_t+i\vec v\cdot\vec e$ the spectrum is $a\pm\lvert\vec v\rvert$, each doubled, because $L(\vec v\cdot\vec e)^{2}=-\lvert\vec v\rvert^{2}I$ and the minimal polynomial is quadratic on a four-dimensional space. Measured at $3.1\times10^{-15}$ over three hundred draws. There are two eigenvalues and four frequencies, each doubled; "four branches" names the second count and hides the first.
The reality of the action selects the potential's sector and not its own: $\mathrm{Sc}(\tilde\Xi^{*}\tilde M\tilde\Xi)$ is real for Hermitian $\tilde M$ and purely imaginary for anti-Hermitian, verified at $3.6\times10^{-15}$ and $7.1\times10^{-15}$. So the material sector is excluded by the reality of $S$, and the excluded case has a purely imaginary spectrum and a complex dispersion.
The lower branch is negative when $\lvert\vec v\rvert>a$, which happened in $276$ of $300$ random draws: the tachyon threshold is generic for unconstrained coefficients. The free case has all four eigenvalues at $k_m^{2}$ and a symbol that commutes with everything, so the free scalar symbol carries no sector content and every sector statement belongs to the potential.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $\tilde K_\Phi=ik_t'e_t+\vec k_\Phi$ | The four-wavevector; $k_t'=\omega/c$ |
| $\langle\tilde K_\Phi\rangle_\natural=-k_t'^{2}+k_\Phi^{2}$ | Its quaternionic norm |
| $\tilde M$ | The constant potential; Hermitian $a\,e_t+i\vec v\cdot\vec e$ or anti-Hermitian $i\,a\,e_t+\vec v\cdot\vec e$ |
| $L(\tilde M)$ | Left multiplication by $\tilde M$; complex-linear on $\mathbb{B}$ |
| $\operatorname{eig}L(\tilde M)=a\pm\lvert\vec v\rvert$ | The spectrum, each value twice |
| $\omega^{2}/c^{2}-k_\Phi^{2}$ | The eigenvalue of $\Box_\Phi$ in wave-number units |
| $\mathrm{Sc}(\tilde\Xi^{*}\tilde M\tilde\Xi)$ | The potential density; real iff $\tilde M$ Hermitian |
| $a\geq\lvert\vec v\rvert$ | The non-negativity condition; its failure is the tachyon threshold |
| $ww^{*}$ | The Hermitian representative of the eigenspace $\mathbb{B}w$ |
Further Reading
- The Lagrangian and the Klein–Gordon Equation, for the action, the four scalar actions, the Hermitian pairing and the sentence this article completes.
- The Fourier Transform and the Klein–Gordon Equation in Biquaternionic Form, for the same symbol read through its zero set and the factorization question.
- The Fourier Transform and the Klein–Gordon Equation in Biquaternionic Polar Variables, for the transform's phase as the action and the role of the amplitude.
- The Klein–Gordon Equation in Biquaternionic Form, for the plane waves, the dispersion relation and the verdict on the sector split.
- Derivatives, Fields and Transforms, for the transform, its kernel and the symbol rule.
- Norm and Pairing Conventions, for the Hermitian form and the centrality of the quaternionic norm.
- The First-Order Root of the Biquaternion Klein–Gordon Operator, for the spinor square root that the scalar operator does not have.