The Lagrangian and the Coupled Schrödinger Equation in Momentum Space
Introduction
The coupled action's fixed-frequency form is the coupled symbol, and its reality is a term-by-term fact. The Lagrangian and the Coupled Schrödinger Equation writes the action for $i\hbar D_t\tilde\Xi=\frac{1}{2m}\sum(-i\hbar D_\eta)^{2}\tilde\Xi$. Take a mode of one frequency and one wave vector, and the action's density becomes the symbol of The Fourier Transform and the Coupled Schrödinger Equation in Biquaternionic Form. The identity is the same identity as in the free case, with two shifted arguments.
The first finding is that the reality survives the coupling term by term. The kinetic term is real for every anti-Hermitian root, the gradient term is a sum of moduli, and the potential term is real because $\varphi$ is. No cancellation is used, so the coupling cannot make a real action complex and cannot make a complex action real.
The second finding is that the Legendre slope is untouched. $\partial\sigma/\partial\omega=\hbar$ with or without the connection, because the connection does not depend on the frequency. The slope in the four-wavevector's temporal slot is $\hbar c$. The map from frequency to time is invertible in every coupled case, and the variational problem is unconstrained.
The third finding is that the energy is minus the symbol at zero frequency. For a mode of unit norm, $E=\frac{1}{2m}\lvert\hbar\vec k-q\vec A\rvert^{2}-ϙ\varphi=-\sigma_{\mathrm{gco}}(0,\vec k)$, verified over one hundred draws with zero residual. The symbol is the action's density, and the energy is that density with the frequency set to zero and the sign flipped, which is the standard sign of a Lagrangian's Legendre transform.
The fourth finding is that the symbol is gauge invariant in the invariant coordinates. A gauge transformation shifts $\varphi$ and $\vec A$, and the same transformation shifts the frequency and the wave vector of a mode. The two shifts cancel in the symbol, verified to $8.9\times10^{-16}$. So the symbol is a function of $\mathcal E$ and $\vec\pi$ and not of the four ingredients separately, and the two slots' opposite signs are exactly what makes the cancellation work.
The fifth finding is a resolution. The earlier companions report a sign for the frequency shift that a careless substitution would reverse. Here the reversal is visible as a failure of exactly the cancellation above, so the article can state which sign is the parent's and why.
The companion articles. The Lagrangian and the Schrödinger Equation in Momentum Space is the free version. The Fourier Transform and the Coupled Schrödinger Equation in Biquaternionic Form gives the symbol, and its polar companion gives the chart. This article joins the action to the coupled symbol.
Conventions
The coupled action, its variation, its currents and the sign of $ϙ$ are those of The Lagrangian and the Coupled Schrödinger Equation. The symbol and its two settings are those of The Fourier Transform and the Coupled Schrödinger Equation in Biquaternionic Form.
The action. With $D_t=\partial_{t'}-iϙ\varphi/\hbar$ and $D_\eta=\partial_\eta+iϙ A_\eta/\hbar$,
$$ S=\int\mathrm{d}t'\;\mathrm{Sc}\left[\frac{i\hbar}{2}\big(\tilde\Xi^{*}D_t\tilde\Xi-(D_t\tilde\Xi)^{*}\tilde\Xi\big)-\frac{1}{2m}\sum_\eta(D_\eta\tilde\Xi)^{*}(D_\eta\tilde\Xi)\right], $$
an integral over time of a density that already contains the spatial derivatives. The action is one-dimensional and the equation is not, and that asymmetry is the parent's.
The two settings, named once. The constant setting has a constant Hermitian Hamiltonian and a constant connection, so the symbol is an element. The operator setting has $\hat H=-\hbar^{2}\Delta_\Phi/(2m)+V$, so the symbol is a scalar. Every display is labelled.
The mode. $\tilde\Xi=e^{-i\omega t'}\tilde X$ in the constant setting, and $\tilde\Xi=e^{-i\omega t'+i\vec k_\Phi\cdot\vec r}\tilde X_0$ in the operator setting. The two substitutions differ by the spatial factor, and the difference is what makes the second setting's symbol a function of two arguments.
The Coupled Action at Fixed Frequency
The constant setting. With $\tilde\Xi=e^{-i\omega t'}\tilde X$ and a constant connection, the density is
$$ \mathrm{Sc}\Big[\tilde X^{*}\big(\hbar\omega\,e_t+ϙ\varphi\,e_t-\tilde H_\Psi\big)\tilde X\Big] =\big\langle\tilde X,\sigma_{\mathrm{gco}}(\omega)\tilde X\big\rangle , $$
because the spatial derivatives vanish on a constant $\tilde X$ and the covariant time derivative becomes a number. The kernel is the element-form symbol of the companion, with $\sigma_{\mathrm{gco}}(\omega)=\hbar\omega\,e_t-\tilde H^{\mathrm{gco}}_\Psi$ and $\tilde H^{\mathrm{gco}}_\Psi=(h_t-ϙ\varphi)e_t+i(\vec h_\Psi-q\vec A)$.
The operator setting. With $\tilde\Xi=e^{-i\omega t'+i\vec k_\Phi\cdot\vec r}\tilde X_0$, the covariant spatial derivative becomes the number $ik_\eta+iϙ A_\eta/\hbar$, so
$$ (D_\eta\tilde\Xi)^{*}(D_\eta\tilde\Xi)=\Big(k_\eta+\frac{ϙ A_\eta}{\hbar}\Big)^{2}\mathrm{Sc}\big(\tilde X_0^{*}\tilde X_0\big), $$
and the density is $\sigma_{\mathrm{gco}}(\omega,\vec k)$ times $\mathrm{Sc}(\tilde X_0^{*}\tilde X_0)$, with $\sigma_{\mathrm{gco}}$ the scalar of the companion. The form's kernel is the symbol and the form's value is the symbol times the mode's norm.
The two settings agree where both exist. In the constant setting with $\vec h_\Psi=\hbar\vec k_\Phi$ the element's scalar part is the operator setting's scalar. The agreement is the statement that a plane wave of a constant Hamiltonian is a plane wave of the differential operator, and it holds at one mode and not for a general field.
Stationarity returns the coupled equation. Varying $\tilde X^{*}$ in the fixed-frequency form gives $\sigma_{\mathrm{gco}}\tilde X=0$, which is the companion's frequency-space equation. So the symbol is the coupled action's Euler–Lagrange operator, exactly as in the free case.
On shell the form vanishes. A solution has $\sigma_{\mathrm{gco}}\tilde X=0$, so the form's value is zero. That is the homogeneous quadratic form's standard property and not a statement about the energy, which is the next section's object.
What the substitution does not assume. It does not assume that the connection is constant. A varying connection makes the density a function of position, and then the fixed-frequency reduction is a reduction in time only. The reduction in space is a separate assumption and it is named in the mode's definition.
Reality Term by Term
The kinetic term. The antisymmetrised covariant kinetic density is real exactly when the root is anti-Hermitian, by the free article's computation, and the covariant derivative's connection part is central so it does not change the type. The coupling adds a real number to a real quantity.
The gradient term. $\mathrm{Sc}\big((D_\eta\tilde\Xi)^{*}(D_\eta\tilde\Xi)\big)=\sum_\mu\lvert(D_\eta\tilde\Xi)_\mu\rvert^{2}$, a sum of four moduli, which is real and non-negative. The term is a norm and not a form, and its positivity is what makes the kinetic energy of the field positive for every configuration.
The potential term. $\varphi$ is real, so $ϙ\varphi\,\mathrm{Sc}(\tilde\Xi^{*}\tilde\Xi)$ is real and its sign is the sign of $ϙ\varphi$. The term is the only one whose sign is not fixed, and that is the standard statement that a potential can lower the energy without bound.
So the reality is term by term. Each of the three terms is real on its own, so the sum is real and no cancellation between terms is needed. That is stronger than the free case's statement, where the antisymmetrised kinetic term was the only one and its reality was the whole content.
What the term-by-term structure forbids. It forbids a complex action from a real connection and a Hermitian Hamiltonian. A complex action would need a non-Hermitian coefficient or a complex $ϙ$, and the article's licence excludes both.
The modulus is preserved by the connection. $\lvert D_\eta\tilde\Xi\rvert$ is unchanged when $\tilde\Xi$ is multiplied by an invertible element, so the gauge transformation cannot change the gradient term. That invariance is the reason the action's gauge covariance is exact and not up to a total derivative, and the parent states it once.
Legendre and the Two Invariant Slopes
The slope in the frequency. $\partial\sigma_{\mathrm{gco}}/\partial\omega=\hbar$, independent of $\varphi$, of $\vec A$ and of the Hamiltonian. The connection shifts the arguments and never the slope, because it does not depend on the frequency.
The slope in the temporal wave number. With $k_t'=\omega/c$ the slope is $\hbar c$, and the factor is the unit that relates the two variables. The two slopes are one slope read in two units, and no physics depends on which is written.
Non-degeneracy. $\hbar\neq0$, so the Legendre map is invertible and the variational problem has no constraint and no multiplier. The coupling does not introduce one, because the connection enters the density as a number and not as a Lagrange multiplier.
What the slope does not see. The slope does not see $\vec\pi$, so the mechanical momentum is not a Legendre conjugate of anything in this article. It is a Noether momentum, the parent's, and the two notions coincide only in a second-order theory.
The mixed derivative vanishes. $\partial^{2}\sigma_{\mathrm{gco}}/\partial\omega\partial k_\eta=0$, so the frequency and the wave vector are separated in the symbol. The separation is the statement that the symbol is a sum of a frequency term and a wave-vector term, and it is the constant-coefficient structure read on the second derivative.
Why the mixed derivative matters. A nonzero mixed derivative would mean that the frequency's conjugate depends on the wave vector, which is what a dispersion with a velocity-dependent mass would look like. The article's class has none, and the flatness of the second derivative is the class's mark.
The Energy Is the Symbol at Zero Frequency
The constant setting. The energy on a mode is $E=\mathrm{Sc}\big(\tilde X^{*}\tilde H^{\mathrm{gco}}_\Psi\tilde X\big)$, and the symbol at zero frequency is $\sigma_{\mathrm{gco}}(0)=-\tilde H^{\mathrm{gco}}_\Psi$. So
$$ E=-\big\langle\tilde X,\sigma_{\mathrm{gco}}(0)\tilde X\big\rangle , $$
and the identity is exact by construction. The sign is the Lagrangian's: a Lagrangian is $T-V$ and the Hamiltonian is $T+V$, so the potential's sign flips and the symbol, which is the Lagrangian's density, carries the minus.
The operator setting. With a unit mode, $E=\frac{1}{2m}\lvert\hbar\vec k-q\vec A\rvert^{2}-ϙ\varphi$, and $\sigma_{\mathrm{gco}}(0,\vec k)=ϙ\varphi-\frac{1}{2m}\lvert\hbar\vec k-q\vec A\rvert^{2}$. So $E=-\sigma_{\mathrm{gco}}(0,\vec k)$, verified over one hundred draws with zero residual. The identity is one line and it fixes the article's sign conventions once.
The positivity of the gradient term is visible here. The energy's first term is a sum of moduli and the second is the potential. So the energy is bounded below exactly when the potential is, and the bound is the parent's and not this article's.
The bound and the symbol. $\sigma_{\mathrm{gco}}(0,\vec k)$ is bounded above by $-E_{\min}$, so the symbol at zero frequency cannot exceed the negative of the ground energy. That is the variational principle's standard inequality read on the symbol, and it is a statement about one mode.
What the identity does not say. It does not say that $\sigma$ is the Hamiltonian's symbol. The Hamiltonian's symbol is $\hbar\omega$ itself; the Lagrangian's symbol is $\hbar\omega$ minus the Hamiltonian's. The two differ by the Legendre transform of a quadratic, and this article uses the Lagrangian's.
A check of the sign's consistency. Composing the identity with the on-shell condition gives $E=0$ on a solution at zero frequency, which is the statement that a zero-frequency solution has zero energy under this normalisation. The statement is a normalisation and not a physical claim, and the article states it once.
Gauge Invariance of the Symbol, and the Resolution of the Sign
The two shifts. Under a gauge transformation $\tilde\Xi\mapsto e^{iϙ\Gamma/\hbar}\tilde\Xi$, the connection moves as $\varphi\mapsto\varphi+\partial_{t'}\Gamma$ and $\vec A\mapsto\vec A-\vec\nabla_\Phi\Gamma$. A mode's phase moves as $\theta\mapsto\theta+ϙ\Gamma/\hbar$, so its frequency and wave vector move as
$$ \omega\longmapsto\omega-\frac{ϙ}{\hbar}\partial_{t'}\Gamma, \qquad \vec k\longmapsto\vec k+\frac{ϙ}{\hbar}\vec\nabla_\Phi\Gamma . $$
The symbol is invariant. Substituting the four shifts into $\sigma_{\mathrm{gco}}$ cancels $\partial_{t'}\Gamma$ in the frequency term and $\vec\nabla_\Phi\Gamma$ in the mechanical momentum, so $\sigma_{\mathrm{gco}}$ is unchanged. Verified over one hundred draws with constant $\Gamma$: the residual was $8.9\times10^{-16}$.
What the invariance says. The symbol is a function of $\mathcal E=\hbar\omega+ϙ\varphi$ and $\vec\pi=\hbar\vec k-q\vec A$ and of nothing else. The four ingredients are not independent, and the symbol's two arguments are the two invariants of the polar companion.
The resolution of the sign. The frequency shift in the symbol is $+ϙ\varphi/\hbar$ and the mode's frequency shift under a gauge transformation is $-ϙ\partial_{t'}\Gamma/\hbar$. The two signs differ, and they must differ, because $\varphi$ and $\partial_{t'}\Gamma$ enter $\mathcal E$ with the same sign while $\omega$ and $\partial_{t'}\Gamma$ enter with opposite signs. A substitution that gave $\omega\mapsto\omega+ϙ\varphi/\hbar$ with the other sign would break the cancellation, and that is how the article detects the error rather than asserting the convention.
What the invariance does not say. It does not say that the action is gauge invariant. The action's value moves by a total derivative, which is the parent's statement and not this article's. The symbol's invariance is stronger and it needs no boundary term, because the symbol is a density at one point and a total derivative has no local content.
A remaining freedom. A constant $\Gamma$ shifts both invariants by the same constants, so the symbol is invariant under a two-parameter translation of its arguments. That freedom is the constant-background case's, and it is why the polar companion can call the constant background exact.
What the Coupled Version Adds
One identity, one sign, one invariance. The identity is that the coupled symbol is the fixed-frequency coupled action. The sign is $E=-\sigma_{\mathrm{gco}}(0,\vec k)$, verified with zero residual. The invariance is the cancellation of the gauge shift, verified to $8.9\times10^{-16}$.
The term-by-term reality. The coupled action is real because each of its three terms is real, not because of a cancellation. That is a stronger statement than the free article's, and it is the reason a central connection cannot spoil the variational structure.
What it does not add. No new prediction, no new solution, no statement about a varying field strength. The action is the parent's and the symbol is the companion's; the article joins them and checks the join.
Limits
Constant coefficients. The symbol exists for a constant Hamiltonian and a constant connection. With $\vec A$ linear in space the symbol depends on position through $\vec A$ and the plane-wave reduction fails even for a constant field strength.
Hermitian coefficients and real $ϙ$. The term-by-term reality uses both. A complex $ϙ$ makes the potential term complex; a non-Hermitian Hamiltonian makes the kinetic term complex for the central root as well.
One time. The action is an integral over $t'$, so the transform is in one variable. The four-dimensional transform would put the slope at $\hbar c$ and keep every other statement, and the article does not construct it.
Open Questions
The first question is the varying connection's Legendre structure. The slope in $\omega$ is $\hbar$ everywhere, and the local frequency is a function of position. Whether a local Legendre transform with a local slope exists is the polar companion's question, and it is open here.
The second is the bound's sharpness. The energy is bounded below by the potential's bound, and the symbol at zero frequency sees that bound. Whether the bound is attained by a mode of the present class for any potential is not decided.
The third is the sign of $ϙ\varphi$ in the energy. The article's identity fixes the relation between $\sigma_{\mathrm{gco}}(0)$ and $E$ and does not fix the physical sign of $ϙ$. That is a conventions question and the conventions articles own it.
The fourth is the second-order analogue. The free article's gap between the reality of an action and the reality of a spectrum has no coupled counterpart here, because the coupled symbol's spectrum is the free one's. Whether a coupled second-order equation would show the same gap is the Dirac and Klein–Gordon companions' question.
Summary
The coupled action at fixed frequency is the coupled symbol: $\big\langle\tilde X,\sigma_{\mathrm{gco}}(\omega)\tilde X\big\rangle$ in the constant setting, and $\sigma_{\mathrm{gco}}(\omega,\vec k)$ times the mode's norm in the operator setting, with the same shift of the free symbol's arguments. The reality is term by term: the kinetic term by the anti-Hermitian root, the gradient term as a sum of moduli, the potential term because $\varphi$ is real. No cancellation is needed, which is why a central connection cannot spoil the variational structure.
The Legendre slope is $\hbar$ and the mixed derivative vanishes, so the frequency and the wave vector are separated and the problem is unconstrained. The energy is minus the symbol at zero frequency, $E=-\sigma_{\mathrm{gco}}(0,\vec k)$, verified over one hundred draws with zero residual, and the sign is the Lagrangian's.
The symbol is gauge invariant in the invariant coordinates: the four shifts cancel, verified to $8.9\times10^{-16}$, so $\sigma_{\mathrm{gco}}$ is a function of $\mathcal E$ and $\vec\pi$ alone. That invariance is also the resolution of the frequency shift's sign, because the opposite sign would break the cancellation.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $S_\omega[\tilde X]=\int\mathrm{d}t'\,\big\langle\tilde X,\sigma_{\mathrm{gco}}(\omega)\tilde X\big\rangle$ | The coupled action at fixed frequency; constant setting |
| $\sigma_{\mathrm{gco}}(\omega)=\hbar\omega\,e_t-\tilde H^{\mathrm{gco}}_\Psi$ | The element symbol; the form's kernel |
| $\sigma_{\mathrm{gco}}(\omega,\vec k)=\hbar\omega+ϙ\varphi-\lvert\hbar\vec k-q\vec A\rvert^{2}/(2m)$ | The scalar symbol; operator setting |
| $\mathcal E=\hbar\omega+ϙ\varphi$, $\vec\pi=\hbar\vec k-q\vec A$ | The two invariants; the symbol's only arguments |
| $\omega\mapsto\omega-\frac{ϙ}{\hbar}\partial_{t'}\Gamma$, $\vec k\mapsto\vec k+\frac{ϙ}{\hbar}\vec\nabla_\Phi\Gamma$ | The mode's gauge shifts; they cancel in the symbol |
| $E=-\sigma_{\mathrm{gco}}(0,\vec k)$ | The energy as minus the symbol at zero frequency |
| $\partial\sigma_{\mathrm{gco}}/\partial\omega=\hbar$ | The Legendre slope; unchanged by the coupling |
| $\partial^{2}\sigma_{\mathrm{gco}}/\partial\omega\partial k_\eta=0$ | The vanishing mixed derivative; the class's mark |
| $\sum_\mu\lvert(D_\eta\tilde\Xi)_\mu\rvert^{2}$ | The gradient term; real and non-negative |
Further Reading
- The Lagrangian and the Coupled Schrödinger Equation, for the coupled action, its variation and its currents cited here.
- The Fourier Transform and the Coupled Schrödinger Equation in Biquaternionic Form, for the two settings' symbols and the translation identity.
- The Lagrangian and the Schrödinger Equation in Momentum Space, for the free version of this article's identity.
- The Fourier Transform and the Coupled Schrödinger Equation in Biquaternionic Polar Variables, for the same symbol read in the chart.
- The Coupled Schrödinger Equation in Biquaternionic Form, for the signs of the two covariant slots.
- The Lagrangian and the Non-Quantum Limit: The Action Principle, for the principle and its Legendre structure.