🔢 mathematics

Approximating Periodic Orbits with Algebraic Curves and Related Minimal Problems

This paper introduces a method for approximating periodic orbits in the Circular Restricted Three-Body Problem using low-degree algebraic curves to construct minimal problems for spacecraft navigation, including the computation of solution counts and the development of homotopy-continuation solvers.

Original authors: Ruiqi Huang, Anton Leykin

Published 2026-04-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ruiqi Huang, Anton Leykin

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: The Cosmic Dance Card

Imagine the Earth and the Moon as two massive dancers rotating around a common center. Now imagine a tiny, weightless grain of dust (a spacecraft) trying to move around them. The gravity of the two dancers pulls on the dust grain, creating complex, repeating paths known as periodic orbits.

Scientists know these paths exist, but calculating the exact mathematics for them is incredibly difficult – as if trying to predict the exact path of a leaf swirling in a hurricane. This paper proposes a clever shortcut: Instead of solving the complex physical equations every time, we can draw simple algebraic curves (like smooth, mathematical shapes) that come very close to these cosmic paths.

The Main Idea: Drawing a "Sketch" of the Orbit

The authors take data points from known orbits (such as "Lyapunov" and "Halo" orbits) and fit them with low-degree algebraic equations.

The Analogy: Imagine a periodic orbit as a winding mountain road. Instead of memorizing every single pothole and curve (the complex physics), the authors create a smooth, simple plastic mold (the algebraic curve) that fits over the road.
The Result: They create a "family" of these shapes. By turning a knob (representing the orbit's energy), the shape changes to fit different roads. This provides them with a flexible, mathematical model that is much easier to handle than raw physics.

Why Do We Do This? The "Cosmic GPS" Problem

The actual goal is not just to draw pretty curves; it is to help spacecraft find their way without needing a huge satellite network on Earth. This is called Liaison Navigation.

The Scenario: Imagine you are in a spacecraft (Spacecraft A) and your friend is in another spacecraft (Spacecraft B). You are far from Earth. You cannot see the ground, but you can measure the following:
1. How far apart you are (Range).
2. How fast this distance is changing (Range Rate).
3. The direction in which you are looking at each other (Line of Sight).
The Puzzle: If you know the "traffic rules" (the algebraic curve model) and have these measurements, can you determine exactly where both spacecraft are located?
- This is a Minimal Problem: It is a puzzle where the number of clues (measurements) exactly matches the number of unknowns (positions and velocities). If you have too few clues, there are infinitely many answers. If you have too many, it is overdetermined. The authors want the "Goldilocks" number of clues.

The Mathematical Challenge: Counting the Answers

In mathematics, solving such a puzzle can yield one answer or many. The authors calculated exactly how many possible answers exist for these puzzles in the "generic" case (the standard situation, no special cases).

The Metaphor: Imagine a labyrinth. Sometimes there is only one exit. Sometimes there are 84 exits. The authors used powerful computer tools to count the exits for various versions of their navigation puzzle.
- For a specific puzzle with a simple fourth-degree curve model, there were 84 possible solutions.
- For a more complex sixth-degree model, the number jumped to 132.
- For 3D "Halo" orbits, the numbers became much higher (e.g., 3,024).

How They Solve It

The paper outlines a two-step strategy for using these models:

The Quick Estimate: Use the algebraic curve models to solve the "Minimal Problem" quickly. Since they know the number of solutions (the degree), they can apply a method called Homotopy Continuation (imagine tracing a path from a known solution to a new one) to find all possible answers rapidly.
The Refinement: These quick answers are not perfect, as the algebraic curves are only approximations. However, they serve as excellent starting points. You can feed these "good estimates" into a more precise, physics-based computer program (like a Kalman Filter) to refine the location until it is perfectly accurate.

Summary of Key Findings

Approximation: They successfully replaced complex orbital physics with simple polynomial curves that fit the data very well.
Navigation: They transformed these curves into navigation puzzles (Minimal Problems) that use distance and direction data between spacecraft.
Complexity: They calculated the "degree" (number of possible solutions) for these puzzles. Some are manageable (e.g., 84 solutions), while others are very complex (e.g., over 3,000 solutions).
Practicality: For the simpler puzzles, they can develop fast solvers. For the harder ones, they suggest breaking the problem into smaller parts or using the algebraic solutions as a "first estimate" for more advanced computers.

In short: The authors have built a set of simple, mathematical "shapes" for space orbits. They showed that if you know the shape of the form and measure how far apart two spacecraft are, you can mathematically determine where they are. This offers a fast way to get a "first estimate" for deep space navigation.

1. Problem Statement

The paper addresses the challenge of Liaison Navigation in the Circular Restricted Three-Body Problem (CR3BP). In this context, spacecraft must determine their states (position and velocity) using inter-spacecraft measurements (ranges, range rates, and line-of-sight angles) without continuous ground tracking.

The core difficulties are:

Complex Dynamics: The CR3BP (modeling motion under two massive primary bodies, e.g., Earth and Moon) possesses no closed-form solutions for general trajectories, although it exhibits rich families of periodic orbits (Lyapunov and Halo orbits).
Navigation Formulation: Determining spacecraft states from sparse measurements leads to systems of polynomial equations. To solve these efficiently, minimal problems are required—quadratic systems (equal numbers of equations and unknowns) with a finite number of solutions.
Computational Costs: Solving these systems requires knowledge of the algebraic degree (the number of complex solutions in the generic case) to construct efficient solvers (e.g., via homotopy continuation).

2. Methodology

A. Algebraic Approximation of Orbits

Instead of using the full differential equations of motion for every step, the authors propose approximating families of periodic orbits with low-degree implicit algebraic curves.

Lyapunov Orbits (Planar): Modeled as a quartic (degree 4) implicit polynomial equation $g(x, y) = 1$ , symmetric with respect to the x-axis. The model comprises 8 coefficients.
Halo Orbits (3D): Modeled by transforming coordinates into a specific basis $(u, v, w)$ and fitting two equations: a quartic curve in the $uv$-plane and a polynomial expression for the $w$ -coordinate.
Fitting Strategy:
- Single Orbit: Coefficients are determined via least-squares fitting to data points from the JPL Three-Body Periodic Orbit Catalog.
- Orbit Families: To handle the entire family of orbits parameterized by the Jacobi Constant ( $C$ ), the authors propose two approaches:
  1. Two-Stage: Fit coefficients for individual orbits, followed by fitting these coefficients as polynomials of $C$ .
  2. One-Stage: Simultaneously fit all data points to a single model $g(x, y, C)$ , where the coefficients are polynomials in $C$ .
- Result: The One-Stage method performs better for higher-degree models (sextic), while the Two-Stage method suffices for quartic models.

B. Formulation of Minimal Problems

Using the algebraic orbit models, the authors formulate navigation scenarios as systems of polynomial equations.

Measurements: Cross-link range ( $d_{AB}$ ), range rate ( $\dot{d}_{AB}$ ), and line-of-sight (LOS) directions.
Unknowns: Spacecraft positions, velocities, and Jacobi Constants ( $C$ ).
System Construction: The system combines:
1. Orbit Conditions: The algebraic curve equation $g(\mathbf{x}, C) = 1$ .
2. Measurement Conditions: Distance equations $(x_i - x_j)^2 + \dots = d^2$ and LOS direction conditions.
3. Dynamics Conditions: Relationships between velocity and $C$ via the Jacobi constant equation.

C. Degree Computation

The authors compute the algebraic degree of these minimal problems (the number of complex solutions).

Tools: Symbolic computation with Macaulay2 (Gröbner bases) and numerical methods.
Strategy: For generic parameters, the degree corresponds to the degree of the branched covering map from the solution variety to the parameter space.

3. Main Contributions

Algebraic Orbit Models: The development of low-degree implicit polynomial models approximating Lyapunov and Halo orbit families as a function of the Jacobi Constant $C$ . This bridges the gap between complex dynamics and algebraic geometry.
Minimal Problem Formulation: The systematic derivation of quadratic polynomial systems for various liaison navigation scenarios (e.g., M2S: One mothership, two spacecraft; 6S: Six independent spacecraft).
Degree Analysis: The computation of algebraic degrees for these problems, which is crucial for solver design.
- Example: The M2S problem (same orbit) has a degree of 84 for the quartic model and 132 for the sextic model.
- Example: The 6-spacecraft Halo problem (with unknown $C$ ) has a degree of 3024 for the quartic model.
Roadmap for Solver Construction: A proposal to use parameter-based homotopy continuation. By pre-computing a start system for a generic case, one can efficiently track solution paths to specific cases, thereby providing initial estimates for local nonlinear optimizers (e.g., Levenberg-Marquardt) or Kalman filters.

4. Results and Insights

Fitting Accuracy: The algebraic models successfully approximate JPL catalog data. The One-Stage fitting procedure is superior to the Two-Stage approach for higher-degree models (sextic).
Degree Variability:
- Lyapunov vs. Halo: Some problems minimal for Lyapunov orbits become over-determined or non-minimal for Halo orbits due to the 3D nature of Halo orbits (introduction of $z$ -conditions). Conversely, new minimal problems arise for Halo orbits (e.g., the 6-spacecraft problem with ranges only).
- Influence of Measurements: Adding line-of-sight (LOS) measurements significantly reduces the number of unknowns but can increase the algebraic degree of the resulting system (e.g., the 2-spacecraft LOS problem has a degree of 16 for quartic models).
- Complexity Limits: Problems involving higher derivatives of range (to avoid unknown $C$ ) lead to extremely high algebraic degrees, limiting practical usability.
Computational Feasibility: While symbolic computations for high-degree models (e.g., sextic) sometimes exceed time limits, numerical homotopy methods remain feasible for low to moderate degrees (e.g., 6, 84, 132).

5. Significance and Future Work

Practical Navigation: This work provides a mathematically rigorous foundation for autonomous navigation in deep space (e.g., lunar gateways). By converting dynamic navigation into algebraic minimal problems, spacecraft can rapidly compute initial state estimates without heavy reliance on ground stations.
Hybrid Solvers: The algebraic solutions serve as high-quality initial estimates for dynamics-aware statistical estimators (such as the Kalman filter) and ensure convergence in nonlinear optimization.
Generalizability: The methodology is independent of specific dynamics and can be extended to other periodic/quasi-periodic families or various dynamic models.
Future Directions:
- Implementation of efficient homotopy continuation solvers for the identified low-degree problems.
- Use of piecewise approximations (splines) to manage algebraic complexity for more difficult problems.
- Integration of these algebraic models into real-time onboard navigation software.

In summary, the paper successfully translates the complex, continuous dynamics of the CR3BP into discrete, solvable algebraic problems, thereby enabling the efficient computation of spacecraft states via minimal measurement sets.