Halo orbits — linearized motion near Lagrange points

WHY do we linearize near a Lagrange point?

The Circular Restricted Three-Body Problem (CR3BP) equations are nonlinear and unsolvable in closed form. But a spacecraft parked near $L_1$ (e.g. Sun–Earth $L_1$ where SOHO lives, or Earth–Moon $L_2$ for Gateway) stays close to the point. So we Taylor-expand the forces about the equilibrium and keep only the first (linear) term. Linear systems we CAN solve — we get exponentials and sinusoids, and we read off the oscillation frequencies directly.

Setup: the rotating frame and the effective potential

We work in the frame co-rotating with the two primaries at angular rate $\omega$ (take $\omega=1$ in normalized units). Let $\mu$ = mass ratio of the smaller primary. In this frame the equations of motion for the small third body are:

\ddot x - 2\dot y = \frac{\partial \Omega}{\partial x},\qquad \ddot y + 2\dot x = \frac{\partial \Omega}{\partial y},\qquad \ddot z = \frac{\partial \Omega}{\partial z}

where the effective potential (also called the pseudo-potential) is

\Omega(x,y,z) = \frac{1}{2}(x^2+y^2) + \frac{1-\mu}{r_1} + \frac{\mu}{r_2}

A Lagrange point is where all forces balance: $\nabla\Omega = 0$ and velocities are zero.

HOW to linearize: Taylor expand $\Omega$ about $L_i$

Let the displacement from the collinear point be $(\xi,\eta,\zeta)$ where $x = x_{L}+\xi$ , etc. Taylor-expand each force component. Since $\nabla\Omega=0$ at the point, the constant term vanishes and the leading term is linear:

\frac{\partial \Omega}{\partial x} \approx \Omega_{xx}\,\xi + \Omega_{xy}\,\eta + \Omega_{xz}\,\zeta

For a collinear point (on the $x$ -axis) by symmetry the cross terms $\Omega_{xy}=\Omega_{xz}=\Omega_{yz}=0$ , and the diagonal second derivatives take a clean form. Define

\boxed{U_{xx} = 1 + 2c,\quad U_{yy} = 1 - c,\quad U_{zz} = -c,\qquad c \equiv \frac{1-\mu}{r_1^3}+\frac{\mu}{r_2^3} > 0}

Formula Derivation of the in-plane characteristic equation — from scratch

The in-plane $(\xi,\eta)$ system is linear with constant coefficients. Try $\xi=A e^{\lambda t}$ , $\eta=B e^{\lambda t}$ : $\lambda^2 A - 2\lambda B = (1+2c)A$ $\lambda^2 B + 2\lambda A = (1-c)B$ Rearrange into matrix form $M\,(A,B)^T=0$ : $\begin{pmatrix} \lambda^2-(1+2c) & -2\lambda \\ 2\lambda & \lambda^2-(1-c)\end{pmatrix}\begin{pmatrix}A\\B\end{pmatrix}=0.$ Nontrivial solution ⇒ determinant $=0$ : $\big(\lambda^2-(1+2c)\big)\big(\lambda^2-(1-c)\big) + 4\lambda^2 = 0$ Expand: $\lambda^4 - (1+2c)\lambda^2 - (1-c)\lambda^2 + (1+2c)(1-c) + 4\lambda^2 = 0$ $\boxed{\lambda^4 + (2-c)\lambda^2 + (1+2c)(1-c) = 0}$ Let $\Lambda=\lambda^2$ . Solve the quadratic: $\Lambda = \frac{-(2-c)\pm\sqrt{(2-c)^2 - 4(1+2c)(1-c)}}{2}.$ For collinear points $c>1$ , so the discriminant is positive and the two roots have opposite sign: one positive $\Lambda_1>0$ (⇒ real $\lambda=\pm\gamma$ , unstable saddle) and one negative $\Lambda_2<0$ (⇒ imaginary $\lambda=\pm i\omega_{xy}$ , oscillation).

Building the periodic (halo) motion

Ignore the unstable saddle direction (set its amplitude to zero — you sit on the center manifold). The bounded motion is:

\xi(t) = -A_x\cos(\omega_{xy}t+\phi),\quad \eta(t) = \kappa A_x\sin(\omega_{xy}t+\phi),\quad \zeta(t) = A_z\sin(\omega_z t+\psi)

where $\kappa$ (the in-plane amplitude ratio) comes from the eigenvector:

\kappa = \frac{\omega_{xy}^2 + (1+2c)}{2\,\omega_{xy}}.

Figure — Halo orbits — linearized motion near Lagrange points

Worked Example 1 — compute $c$ and the frequencies for Earth–Moon $L_2$

Given: $\mu \approx 0.01215$ . The $L_2$ point sits at distance $\gamma_L\approx 0.1678$ beyond the Moon. Take a representative $c \approx 3.19$ (computed from geometry).

Step 1 — Out-of-plane frequency. $\omega_z=\sqrt{c}=\sqrt{3.19}\approx 1.786$ . Why this step? The $\zeta$ equation $\ddot\zeta=-c\zeta$ is a pure SHM with $\omega^2=c$ .

Step 2 — In-plane roots. $\Lambda=\frac{-(2-c)\pm\sqrt{(2-c)^2-4(1+2c)(1-c)}}{2}$ . With $c=3.19$ : $2-c=-1.19$ , $(1+2c)=7.38$ , $(1-c)=-2.19$ . Discriminant $=(-1.19)^2 - 4(7.38)(-2.19)=1.416+64.65=66.07$ , $\sqrt{}=8.13$ . $\Lambda=\frac{1.19\pm 8.13}{2}$ → $\Lambda_1=+4.66$ , $\Lambda_2=-3.47$ . Why this step? One positive, one negative root ⇒ confirms saddle × center structure.

Step 3 — Read the physics. $\gamma=\sqrt{4.66}\approx 2.16$ (instability rate), $\omega_{xy}=\sqrt{3.47}\approx 1.86$ . Why this step? The unstable exponential grows like $e^{2.16\,t}$ — that's why unmanaged spacecraft leave $L_2$ fast; the oscillation $\omega_{xy}\approx1.86$ sets the in-plane period $T\approx 2\pi/1.86\approx 3.4$ (dimensionless time units).

Worked Example 2 — the in-plane amplitude ratio $\kappa$

Given: $\omega_{xy}=1.86$ , $c=3.19$ .

Step 1: $\kappa=\dfrac{\omega_{xy}^2+(1+2c)}{2\omega_{xy}}=\dfrac{3.47+7.38}{3.72}=\dfrac{10.85}{3.72}\approx 2.92$ . Why this step? It's the eigenvector component: for a unit $\xi$ amplitude, the $\eta$ amplitude is $\kappa$ times bigger. So the planar ellipse is elongated in the $y$ -direction by factor $\approx 2.9$ .

Step 2 — interpret. The in-plane motion is not a circle but an ellipse squashed along $x$ . Real halo orbits inherit this stretched shape.

Common mistakes (Steel-manned)

Active recall

Recall Feynman: explain to a 12-year-old

Imagine a giant merry-go-round with the Sun in the middle and Earth on the edge, both spinning together. There are special "quiet spots" where if you sit still, you neither fall toward the Sun nor fly outward — the pushes cancel. But these quiet spots are like sitting on top of a saddle: perfectly balanced left-right (you rock back and forth like a swing), but slippery front-back (nudge and you slide off). A spacecraft can use that gentle rocking to trace a big loop around the quiet spot — that loop is a halo orbit. To ride it, you must gently correct the slippery direction now and then, or you'll slide away.

Flashcards

What defines a Lagrange point mathematically?

\nabla\Omega=0

(gradient of the effective/pseudo-potential vanishes) with zero velocity in the rotating frame.

Write the linearized out-of-plane equation near a collinear point.

\ddot\zeta=-c\zeta

, giving SHM at

\omega_z=\sqrt{c}

What is

c

in the linearization?

c=\dfrac{1-\mu}{r_1^3}+\dfrac{\mu}{r_2^3}

, the sum of scaled inverse-cube gravity terms;

c>1

at collinear points.

Where do the

-2\dot\eta,\,+2\dot\xi

terms come from?

The Coriolis force in the rotating frame.

Why are collinear Lagrange points unstable?

The in-plane characteristic equation has a real positive root

+\gamma

(saddle), so displacements along that eigenvector grow exponentially.

What is the in-plane characteristic equation?

\lambda^4+(2-c)\lambda^2+(1+2c)(1-c)=0

Difference between a Lissajous and a halo orbit?

Lissajous:

\omega_{xy}\neq\omega_z

, path never closes (quasi-periodic). Halo: nonlinear amplitude forces

\omega_{xy}=\omega_z

(1:1 resonance) → closed periodic 3-D loop.

Why can't pure linear theory produce a true halo?

Because in-plane and out-of-plane frequencies generally differ; closing requires nonlinear frequency-amplitude corrections (Richardson).

What is

\kappa

and what does it tell you?

\kappa=\dfrac{\omega_{xy}^2+(1+2c)}{2\omega_{xy}}

, the eigenvector ratio giving how elongated the in-plane ellipse is along

y

Connections

Circular Restricted Three-Body Problem — the parent model.
Lagrange Points L1–L5 — where these live; triangular points differ (Routh criterion).
Jacobi Integral and Zero-Velocity Curves — the conserved energy that constrains motion.
Invariant Manifolds and Low-Energy Transfers — stable/unstable manifolds off the saddle enable interplanetary superhighways.
Station-keeping and Orbit Maintenance — why the saddle instability demands $\Delta v$ .
Simple Harmonic Motion — the $\ddot\zeta=-c\zeta$ analogy.

Concept Map

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Dekho, Lagrange point ke paas rotating frame mein gravity aur centrifugal force lagbhag cancel ho jaate hain. Jo thoda sa bacha hua force hota hai, use hum Taylor expand karte hain aur sirf linear term rakhte hain — kyunki spacecraft point ke bahut paas hi ghoomta hai. Isse equations solve-able ban jaate hain: exponentials aur sinusoids nikalte hain, aur seedha oscillation frequencies mil jaati hain.

Collinear points ( $L_1,L_2,L_3$ ) par ek important baat: in-plane motion ka characteristic equation ke roots mein ek real positive hota hai (yeh saddle wala unstable direction hai — isliye spacecraft ko station-keeping chahiye) aur ek imaginary hota hai (yeh planar wobble deta hai frequency $\omega_{xy}$ ). Aur out-of-plane $z$ equation alag ho jaata hai — simple spring jaisa, frequency $\omega_z=\sqrt{c}$ . Yahi decoupling halo analysis ka sabse pyara trick hai.

Ab yaad rakho: pure linear theory sirf Lissajous deta hai, kyunki generally $\omega_{xy}\neq\omega_z$ , isliye 3-D loop band nahi hota. Asli halo orbit tab banta hai jab amplitude bada ho aur nonlinear frequency-corrections dono frequency ko barabar (1:1 resonance) kar dete hain. Matlab linear theory se ingredients milte hain, par closing condition nonlinear hoti hai (Richardson third-order). Ye samajhna interview aur real missions (SOHO, JWST, Gateway) dono ke liye ekdum core hai.

Halo orbits — linearized motion near Lagrange points

WHY do we linearize near a Lagrange point?

Setup: the rotating frame and the effective potential

HOW to linearize: Taylor expand $\Omega$ about $L_i$

Building the periodic (halo) motion

Worked Example 1 — compute $c$ and the frequencies for Earth–Moon $L_2$

Worked Example 2 — the in-plane amplitude ratio $\kappa$

Common mistakes (Steel-manned)

Active recall

Flashcards

Connections

Concept Map

Hinglish (regional understanding)

Go deeper — visual, from zero

Test yourself — Orbital Mechanics & Astrodynamics

Connections