---
title: Time-Dependent Two-Center Stark-Zeeman System
url: https://www.emergentmind.com/topics/time-dependent-two-center-stark-zeeman-system
type: topic
---

# Time-Dependent Two-Center Stark-Zeeman System

Searching arXiv for the cited papers on time-dependent Stark–Zeeman systems and related two-center regularization.
Searching arXiv: "Periodic orbits in time-dependent planar Stark-Zeeman systems" and "A variational approach to time-dependent planar two-center Stark-Zeeman systems".
A time-dependent two-center Stark–Zeeman system is a planar Hamiltonian model for a charged particle attracted by two fixed Coulomb centers and subject to both a magnetic field and a time-periodic external electric perturbation. In the formulation developed in "A variational approach to time-dependent planar two-center Stark-Zeeman systems" [2507.04904], the singular dynamics arise from the two Coulomb centers, while the explicit time dependence precludes the standard autonomous strategy of regularizing collisions on a fixed energy hypersurface. The resulting theory combines the Birkhoff regularization map with a non-local blow-up of loop space, following the Barutello–Ortega–Vernizi approach that had previously been applied to the one-center time-dependent planar Stark–Zeeman problem in "Periodic orbits in time-dependent planar Stark-Zeeman systems" [2503.09209].

## 1. Geometric and dynamical formulation

The system is defined on a punctured planar domain
\[
\mathfrak{D}=\mathfrak{D}_0\setminus\{E,M\},
\]
where \(\mathfrak{D}_0\) is an open set containing the two fixed Coulomb centers \(E\) and \(M\). The external forcing is periodic in time \(t\in S^1\), and the electric contribution is encoded by a smooth time-dependent potential \(E_t(q)\). The total potential is
\[
V_t(q)=E_t(q)-\frac{1-\mu}{|q-E|}-\frac{\mu}{|q-M|},
\]
with Hamiltonian
\[
H_t(q,p)=\frac12 |p|^2 + V_t(q).
\]
The magnetic field \(\mathfrak{B}(q)\) determines the magnetic \(2\)-form
\[
\sigma_{\mathfrak{B}}=\mathfrak{B}\,dq_1\wedge dq_2,
\]
and the phase space carries the twisted symplectic form
\[
\omega_{\mathfrak{B}}=\sum_{j=1}^2 dp_j\wedge dq_j+\pi^*\sigma_{\mathfrak{B}}.
\]
Because \(H^2(\mathfrak{D}_0)=0\), there exists a primitive \(A\) satisfying
\[
dA=\sigma_{\mathfrak{B}}.
\]
The associated twisted Hamiltonian flow is
\[
\dot x(t)=X^{\mathfrak{B}}_{H_t}(x(t)),
\]
and in configuration variables this becomes
\[
\ddot{q} = -\mathfrak{B}(q)\, i\dot q - \nabla V_t(q),
\]
that is,
\[
\ddot{q} = -\mathfrak{B}(q)\,i\dot q -\frac{(1-\mu)(q-E)}{|q-E|^3} -\frac{\mu(q-M)}{|q-M|^3} -\nabla E_t(q).
\]
After normalization \(E=-1\), \(M=+1\), the equation takes the explicit form
\[
\ddot{q} = -\mathfrak{B}(q)\,i\dot q -\frac{(1-\mu)(q+1)}{|q+1|^3} -\frac{\mu(q-1)}{|q-1|^3} -\nabla E_t(q).
\]
This formulation places the model within the Stark–Zeeman class: Coulomb attraction supplies the singular Newtonian part, the magnetic field contributes the Lorentz-type term, and the periodic electric forcing renders the system nonautonomous [2507.04904].

## 2. Time dependence, singularities, and the variational obstruction

For collision-free periodic loops, the natural action functional is
\[
\mathscr{A}(q) = \frac12\|\dot q\|^2 - \int_{S^1}q^*A - \int_0^1 V_t(q(t))\,dt.
\]
This functional is singular at the Coulomb centers \(q=\pm1\), because the potential contains terms of the form
\[
\frac{1}{|q\pm1|}.
\]
Accordingly, standard local variational methods do not apply directly to periodic orbits that may collide with the primaries [2507.04904].

The time dependence creates a second obstruction. In the one-center time-dependent planar Stark–Zeeman setting, the explicit \(t\)-dependence of the Hamiltonian implies that
\[
\frac{d}{dt}H_t(q(t),p(t)) \neq 0 \quad \text{in general},
\]
so there is no preserved energy and no invariant energy hypersurface available for the classical blow-up procedure [2503.09209]. The two-center paper states the same structural difficulty in the time-dependent case: energy is not conserved, so one cannot regularize by blowing up a fixed energy hypersurface as in autonomous classical approaches [2507.04904].

A common misconception is that collision regularization in Stark–Zeeman dynamics can always be reduced to a local change of coordinates on phase space. The time-dependent two-center theory does not support that view. In the nonautonomous setting, the regularization is formulated on loop space rather than on an energy level, and the resulting Euler–Lagrange equation is non-local rather than an ordinary differential equation [2507.04904].

## 3. Birkhoff regularization and the blown-up loop space

The central regularizing device is the Birkhoff map
\[
B(z)=\frac12\left(z+\frac1z\right), \qquad z\in \mathbb{C}\setminus\{0\}.
\]
This is a branched double cover of the physical \(q\)-plane, branched at \(z=\pm1\), which correspond to the two collision points in the physical plane. The associated weight function is
\[
w(z)=\frac{|z-1|^2|z+1|^2}{4|z|^2},
\]
and it is invariant under inversion \(z\mapsto 1/z\) [2507.04904].

Following the Barutello–Ortega–Vernizi philosophy, the regularization does not act on a fixed energy hypersurface. Instead, it blows up the loop space by means of a loop-dependent time reparametrization. For a loop \(z(\tau)\),
\[
\hat z=\int_0^1 w(z(\tau))\,d\tau,
\qquad
t_z(\tau)=\frac1{\hat z}\int_0^\tau w(z(s))\,ds,
\]
so that
\[
t_z'(\tau)=\frac{w(z(\tau))}{\hat z}.
\]
The critical points of \(t_z\) are exactly the collision points \(z=\pm1\). Writing \(\tau_z=t_z^{-1}\), the reconstructed physical loop is
\[
q_z(t)=B\bigl(z(\tau_z(t))\bigr)=\frac12\left(z(\tau_z(t))+\frac1{z(\tau_z(t))}\right).
\]
Thus, the regularized object is a loop in the \(z\)-plane together with a non-local time change, while the physical periodic orbit is obtained only after composing the Birkhoff map with the inverse reparametrization [2507.04904].

The one-center precursor uses the analogous Levi–Civita squaring map
\[
s(z)=z^2
\]
and a loop-dependent time change
\[
t_z(\tau)=\frac{\int_0^\tau |z(\sigma)|^2\,d\sigma}{\|z\|_{L^2}^2},
\qquad
q_z(t)=z^2(T_z(t)).
\]
That earlier construction is conceptually significant because it demonstrates, in the time-dependent Stark–Zeeman setting, why loop-space blow-up replaces energy-surface blow-up and why the regularized equation becomes a delay equation [2503.09209]. This suggests a structural continuity between the one-center and two-center time-dependent theories.

## 4. The non-local regularized action functional

Let
\[
\Sigma:\mathfrak{L}\mathfrak{Z}\to \mathfrak{L}\mathfrak{D},\qquad \Sigma(z)=q_z,
\]
where \(\mathfrak{Z}=B^{-1}(\mathfrak{D})\). Pulling back the original action \(\mathscr{A}\) by \(\Sigma\) yields the regularized functional
\[
\mathscr{B}=\Sigma^*\mathscr{A},
\]
given by
\[
\mathscr{B}(z) = \mathcal{F}(z)\mathcal{G}(z) + \frac{(1-\mu)\mathcal{H}_1(z)+\mu\mathcal{H}_2(z)}{\mathcal{F}(z)} -\mathcal{M}(z)-\mathcal{E}(z),
\]
with
\[
\mathcal{F}(z)=\hat z=\int_0^1 w(z(\tau))\,d\tau,
\]
\[
\mathcal{G}(z)=\frac12\int_0^1 \frac{|z'(\tau)|^2}{|z(\tau)|^2}\,d\tau,
\]
\[
\mathcal{H}_1(z)=\frac12\int_0^1 \frac{|z(\tau)-1|^2}{|z(\tau)|}\,d\tau,
\qquad
\mathcal{H}_2(z)=\frac12\int_0^1 \frac{|z(\tau)+1|^2}{|z(\tau)|}\,d\tau,
\]
\[
\mathcal{M}(z)=\int_{S^1} z^*B^*A,
\]
and
\[
\mathcal{E}(z) = \frac{1}{\hat z} \int_0^1 E_{t_z(\tau)}(B(z(\tau))) \frac{|z(\tau)-1|^2|z(\tau)+1|^2}{4|z(\tau)|^2} \,d\tau.
\]
The dependence on \(t_z\) makes \(\mathscr{B}\) a non-local action functional: the value of the integrand at time \(\tau\) depends on the entire loop through the normalization factor \(\hat z\) and the time reparametrization itself [2507.04904].

The functional extends naturally to
\[
\mathfrak{L}^*\mathfrak{Z}_0 = \{z\in \mathfrak{L}\mathfrak{Z}_0\mid \hat z\neq 0\},
\]
where collisions may occur. This extension is the key variational device that allows collisional periodic solutions to be captured rather than excluded. In the one-center time-dependent theory, the regularized functional likewise lives on a blown-up loop space and decomposes as
\[
\mathcal F = K-A+C-E,
\]
with kinetic, magnetic, Coulomb, and time-dependent electric parts; there too the non-locality arises from the loop-dependent time change [2503.09209].

## 5. Delay equation and reconstruction of generalized periodic solutions

A critical point \(z\) of the regularized functional \(\mathscr{B}\) satisfies a second-order delay differential equation. In the notation of the two-center paper,
\[
\begin{aligned}
z''(\tau) &= \frac{1}{2\hat{z}^2}\frac{ z(\tau)(z(\tau)^2-1)(\bar{z}(\tau)^2+1) }{\lvert z(\tau)\rvert^2}C + \frac{\bar{z}(\tau)}{\lvert z(\tau)\rvert^2}z'(\tau)^2 \\
&\quad + \frac{1}{2\hat{z}^2}\Bigg[ (1-\mu) \frac{z(z-1)(\bar{z}+1)}{ \lvert z \rvert } + \mu   \frac{z(z+1)(\bar{z}-1)}{ \lvert z \rvert } \Bigg] \\
&\quad - \frac{w(z(\tau))}{\hat{z}} \mathfrak{B}(B(z(\tau)))i z'(\tau)
- \frac{\lvert z(\tau)\rvert^2}{\hat{z}^2}( \varepsilon_1(z)(\tau)+\varepsilon_2(z)(\tau)+\varepsilon_3(z)(\tau)).
\end{aligned}
\]
The constant \(C\) is
\[
C = \frac{\hat z}{2}\int_0^1 \frac{|z'(\tau)|^2}{|z(\tau)|^2}\,d\tau - \frac{1-\mu}{2\hat z\int_0^1 \frac{|z(\tau)-1|^2}{|z(\tau)|}\,d\tau} - \frac{\mu}{2\hat z\int_0^1 \frac{|z(\tau)+1|^2}{|z(\tau)|}\,d\tau} +\mathcal{E}(z)+\mathcal{E}^1(z).
\]
The paper notes that the displayed typesetting is somewhat corrupted in places, but this is the stated structure [2507.04904].

The physical orbit is recovered by
\[
q(t)=B(z(\tau_z(t))).
\]
If \(z\) has no collisions, then \(q\) is a smooth periodic solution of the original system. If \(z\) has collisions, then \(q\) is a generalized periodic solution. The reconstruction identity is
\[
\ddot{q}(t)+\mathfrak{B}(q(t))\,i\dot q(t)+\nabla E_t(q(t)) = \frac{2\bar q(t)}{\bar q(t)^2-1}\Phi(t)-\nabla U(q(t)),
\]
where
\[
U(q)=-\frac{1-\mu}{|q+1|}-\frac{\mu}{|q-1|}
\]
and
\[
\Phi(t) = C-\frac12|\dot q(t)|^2-U(q(t))-\int_t^1 \dot E_s(q(s))\,ds - E_t(q(t)).
\]
A crucial lemma proves that \(\Phi\equiv 0\). Hence the extra term vanishes, and \(q\) satisfies exactly
\[
\ddot q = -\mathfrak{B}(q)i\dot q - \nabla U(q)-\nabla E_t(q)
\]
on the nonsingular part of the loop [2507.04904].

The same section identifies the conserved generalized energy:
\[
\frac12|\dot q(t)|^2+U(q(t))+\int_t^1\dot E_s(q(s))\,ds+E_t(q(t)).
\]
This quantity is not a standard autonomous first integral; rather, it is the expression that appears in the reconstruction argument and extends continuously across collisions for generalized solutions [2507.04904]. A plausible implication is that the regularized variational problem restores an effective energy balance at the level of reconstructed trajectories, even though the original time-dependent Hamiltonian has no preserved energy in the ordinary autonomous sense.

## 6. Collisions, symmetry, winding classes, and relation to the one-center theory

The notion of generalized solution used in the two-center theory requires that the collision set
\[
\mathfrak{C}_q=\{t\mid q(t)\in\{E,M\}\}
\]
is finite, that \(q\) is smooth away from collisions and solves the ODE there, and that the generalized energy extends continuously across collisions. The paper proves finiteness of the collision set for critical points of the regularized functional: the collision set is discrete and therefore finite on \(S^1\) [2507.04904]. This directly addresses the singularity problem without removing collisional orbits from the variational framework.

A major structural feature is the involution
\[
I(z)=\frac1z.
\]
Because
\[
w(z)=w(1/z),
\]
the regularized functional is invariant:
\[
\mathscr{B}\circ I=\mathscr{B}.
\]
Consequently, if \(z\) is a critical point, then so is \(I(z)\), and both yield the same physical periodic solution:
\[
\Sigma(z)=\Sigma(I(z))=q_z.
\]
This is the two-center analogue of the symmetry induced by the double covering nature of the regularization map [2507.04904].

The theory also distinguishes even and odd total winding number. For even winding number, \(B^{-1}(q)\) has two connected components; for odd winding number, \(B^{-1}(q)\) is connected. To treat the odd case, the paper introduces the twisted loop space
\[
\mathfrak{L}_I\mathfrak{Z} = \{z\in C^\infty(\mathbb{R},\mathfrak{Z})\mid z(t+1)=I(z(t))\},
\]
and then doubles time via
\[
\tilde z(t)=z(2t).
\]
Using the combined map
\[
\Sigma_0=\Sigma\oplus \Sigma_I,
\]
the theory yields a one-to-one correspondence, modulo the involution \(I\), between critical points of the extended regularized functional and periodic orbits of the original system [2507.04904].

The main theorem states that there is a one-to-one correspondence between critical points of the extended regularized functional \(\mathscr{B}_0\) modulo the involution \(I\) and generalized periodic solutions of the time-dependent Stark–Zeeman system [2507.04904]. This should be read together with the one-center precursor, where the analogous correspondence is formulated modulo the sign symmetry \(z\mapsto -z\), with untwisted loops corresponding to even winding number solutions and twisted loops corresponding to odd winding number solutions [2503.09209]. The comparison clarifies what is specific to the two-center problem: Levi–Civita squaring is replaced by the Birkhoff map, sign symmetry is replaced by inversion symmetry, and the collision structure is adapted from one singularity to two.

The bicircular restricted four-body problem provides the motivating example for both papers. In rotating coordinates, the earth and moon are fixed, the sun moves periodically, and the resulting Hamiltonian has two fixed Coulomb-like singularities, a time-dependent perturbation from the sun, and a magnetic/Coriolis-type term. This fits precisely into the time-dependent two-center Stark–Zeeman framework [2507.04904]. The broader significance is therefore not merely formal: the variational theory is designed for periodic orbit analysis in a singular, nonautonomous Hamiltonian setting in which both collisions and explicit time dependence are essential features rather than perturbative complications.

Source: https://www.emergentmind.com/topics/time-dependent-two-center-stark-zeeman-system