Pith. sign in

REVIEW 3 major objections 5 minor 20 references

Mass-Optimal Low-Thrust Forced Periodic Trajectories in the Earth-Moon CR3BP

T0 review · 3 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper establishes that the thrust-limited mass-optimal reachable set is a superset of the energy-limited energy-optimal reachable set in the xy-plane for low-thrust forced periodic trajectories in the Earth-Moon system.

desk verdict The method is real and useful, but the abstract's superset claim only holds against a linearized baseline; as stated, it is false for the exact reachable sets. read the letter →

arxiv 2502.05140 v1 pith:3K6ZXWIX submitted 2025-02-07 math.DS

Colby C. Merrill , Jackson Kulik , Matthew J. Bryan , Dmitry Savransky This is my paper · ORCID
classification math.DS MSC 70F0770Q0549J15
keywords forcedperiodictrajectorylow-thrustoptimizationmass-optimalcontrolreachablesetcircularrestrictedthree-bodyproblemEarth-Moonsystemparticleswarmbang-bang
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Natural periodic orbits in the Earth-Moon system occupy only a limited volume of Cislunar space, and this paper asks whether low-thrust propulsion can legitimately expand the set of periodic trajectories available to a spacecraft. The paper develops a numerical method for finding mass-optimal forced periodic trajectories in the circular restricted three-body problem, enforcing that the spacecraft returns to its starting state after one period while thrust magnitude is capped. Its central finding is that the reachable set of these nonlinear, thrust-limited, mass-optimal trajectories is a superset of the reachable set predicted by a linearized energy-optimal analysis, at least when both are projected into the xy-plane. A sympathetic reader would care because propellant-minimizing trajectories that respect thrust limits are what operational spacecraft actually fly, so a wider reachable set means more usable periodic starting states for missions in Cislunar space.

What carries the argument

The object that carries the argument is the reachable set around a fixed reference trajectory, with two constructions. The energy-limited construction is the symmetric matrix $E^*$: combining the augmented state transition matrix $\Phi$ with the linearized boundary-value map produces a quadratic cost $\frac{1}{2}\delta x_0^T E^* \delta x_0$ for returning to a perturbed initial state $\delta x_0$, and the reachable starting states are exactly the hyperellipsoid where this cost is at most $\frac{1}{2}u_{\max}^2(t_f-t_0)$. The thrust-limited construction is the nonlinear optimal control problem with cost $J_M = \int_{t_0}^{t_f} \|u(t)\|\,dt$ and constraint $\|u(t)\| \le u_{\max}$, whose boundary is sampled by an accelerated particle swarm optimizer with fitness $\psi^T \delta x$ and a stopping rule driven by $J_M/(u_{\max}(t_f-t_0)) > 0.95$. The comparison of the two sets in the xy-plane is the load-bearing comparison of the paper.

What would settle it

Recompute the energy-limited reachable set with the same nonlinear optimizer used for the mass-optimal set, and check whether every thrust-limited reachable state in the xy-plane is also energy-limited-reachable; if any mass-optimal thrust-limited state falls inside the nonlinear energy-limited set, the claimed superset fails. A cheaper check is to sample trajectories in additional azimuth directions beyond the twelve used here and see whether any sampled thrust-limited reachable state lies inside the energy ellipsoid.

Watch

Extended reading notes

Core claim

The paper's central claim, stated in the abstract and conclusions, is that the reachable set of nonlinear mass-optimal thrust-limited trajectories is a superset of the linearized energy-optimal energy-limited trajectories in the xy-plane. The energy-limited reachable set is a six-dimensional hyperellipsoid built from a linearized optimal-control analysis: states that can be returned to in one period with energy cost at most $\frac{1}{2}u_{\max}^2(t_f-t_0)$ form the ellipsoid $\{\delta x_0 : \frac{1}{2}\delta x_0^T E^* \delta x_0 \le \frac{1}{2}u_{\max}^2(t_f-t_0)\}$. The thrust-limited mass-optimal set is obtained by solving the full nonlinear problem with cost $J_M = \int_{t_0}^{t_f} \|u(t)\|\,dt$ and constraint $\|u(t)\| \le u_{\max}$, then sampling the boundary with a particle swarm search that seeks states whose mass cost exceeds 95% of the theoretical maximum. Because the two optimization methods use different assumptions and constraints, the paper argues, the energy-optimal set is not automatically the larger one; the comparison shows the mass-optimal set contains it in the xy-plane. The authors attribute the difference to the alterations that linearization makes to the dynamics, and they note the two sets look similar in that projection.

Load-bearing premise

The superset conclusion depends on the linearized hyperellipsoid being a faithful stand-in for the true energy-limited reachable set and on the particle-swarm collection of thrust-limited trajectories being a faithful stand-in for the full thrust-limited reachable set; the paper itself notes that the full thrust-limited set is incomplete because the sampling is imperfect.

Editorial extensions

If this is right

  • Forced periodic trajectories can be generated directly as mass-optimal solutions by feeding an energy-optimal solution into the mass-optimal problem, without smoothing or homotopy.
  • Mass-optimal forced periodic trajectories have bang-bang thrust profiles, typically an extended burn at the start and a short burn near perilune.
  • The xy-plane projection of the thrust-limited reachable set is at least as large as the energy-limited reachable set, so linearized energy analysis may understate which periodic starting states are reachable.
  • Many energy-optimal trajectories already have thrust magnitudes near the mass-optimal maximum even though they do not enforce a thrust bound, so the two solution families are close in thrust usage.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if the superset holds in other projections, the linearized ellipsoid is a conservative bound on reachable volume, and mission designers could search a wider starting-state region for forced periodic operations than the energy analysis alone would suggest.
  • Editorial inference: the consistent bang-bang pattern of a long initial burn and a short perilune burn suggests that a reduced control model with two thrust arcs might approximate the mass-optimal family closely enough for fast preliminary design; that is a testable simplification, not a claim of the paper.
  • Editorial inference: the twelve-direction PSO sampling is a coarse boundary estimate; a denser angular sweep would reveal whether the apparent superset saturates or whether unsampled directions contain thrust-limited reachable states inside the energy ellipsoid.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper presents a methodology for computing mass-optimal low-thrust forced periodic trajectories in the circular restricted three-body problem (CR3BP) using the ASSET/PSIOPT direct transcription framework, and uses a particle swarm optimization (PSO) scheme to sample the thrust-limited reachable set around a reference orbit. The authors compare this nonlinearly computed mass-optimal reachable set with the linearized energy-limited reachable set obtained from their prior work, and the abstract claims that the thrust-limited mass-optimal reachable set is a superset of the energy-limited energy-optimal reachable set in the xy-plane. The paper reports that the two sets are similar in projection and notes in the Results that the comparison is limited by the different assumptions and constraints of the two methods.

Significance. If the claim is suitably rephrased, the paper makes a useful methodological contribution: it provides a reproducible pipeline for generating mass-optimal forced periodic trajectories in the CR3BP, with explicit PSO hyperparameters, mesh refinement, and reintegration verification. The linearized energy-limited ellipsoid from prior work is a parameter-free construction, and the observation that nonlinear mass-optimal solutions can extend outside this linearized ellipsoid is a concrete, falsifiable finding. However, the headline claim as written is not supported by the paper's own definitions and is in fact contradicted by a simple inequality: every thrust-limited control is also energy-feasible, so the exact energy-limited reachable set must contain the exact thrust-limited reachable set. The paper's significance therefore hinges on reframing the comparison as one between the nonlinear mass-optimal set and the linearized energy ellipsoid, not the exact energy-limited set.

major comments (3)
  1. [Reachable Set Definitions and Abstract] The stated superset claim is false for the exact sets defined in Eqs. (20) and (21). If a mass-optimal control satisfies ||u_M(t)|| ≤ u_max for all t, then J_E(u_M) = (1/2)∫||u_M||² dt ≤ (1/2)u_max²(t_f − t_0), so the same δx0 lies in the exact energy-limited set of Eq. (20). Hence the exact thrust-limited set is a subset, not a superset, of the exact energy-limited set. The result plotted in Fig. 4 can only mean that the linearized hyperellipsoid of Eq. (34) is smaller than the exact energy-limited set in the sampled directions. The abstract omits the qualifier 'linearized' and therefore makes a claim that is mathematically impossible; the Conclusions include the qualifier, but the abstract and the 'superset' wording throughout the paper need to be revised to state that the nonlinear mass-optimal set can extend beyond the linearized energy-limited ellipsoid.
  2. [Particle Swarm Optimization and Fig. 4] The thrust-limited reachable set boundary is approximated by the heuristic JM/(umax(tf−t0)) > 0.95 and is sampled in only 12 azimuth directions, with ψ increased in increments of π/6. The paper itself states that the full set is incomplete due to imperfect sampling. A 95% thrust fraction does not guarantee proximity to the true boundary of the thrust-limited reachable set, and unsampled azimuth directions may contain states that would alter the comparison. As presented, the 'superset' statement is only established for the sampled directions and the chosen proxy threshold, not for the entire xy-plane.
  3. [Energy-Limited Reachable Set and Results] The comparison in Fig. 4 is asymmetric: the energy-limited reachable set is the first-order STM hyperellipsoid of Eq. (34), while the mass-limited set is obtained from full nonlinear optimization. The paper concedes in the Results that if the same solution method were used for both problems, the energy-optimal reachable set would be a superset of the mass-optimal reachable set. This concession indicates that the reported 'superset' is likely an artifact of comparing a linearized object with a nonlinear one. To support the paper's central claim, the authors should compute an energy-optimal reachable set with the same nonlinear tool (for example, the same ASSET/PSO pipeline) or provide a quantitative bound on the linearization error in Eq. (34); without this, the observed set difference cannot be attributed to the physical difference between mass and energy optimality.
minor comments (5)
  1. [Equation (40)] The symbol ψ is used both for the weight vector in Eq. (36) and for the scalar azimuth angle in Eq. (40), which is confusing; one of these should be renamed.
  2. [Introduction and References] The Introduction contains a typo: 'equilibirum' should be 'equilibrium'; reference [16] contains 'Reahcable' instead of 'Reachable'.
  3. [Figure 4 caption] The caption should state that the 'Energy Optimal Set' is the projection of the linearized ellipsoid from Eq. (34), not a nonlinear reachable set, to avoid misleading readers.
  4. [Equations (35) and (38)] The notation N is used both for the normal distribution and for the random vector in Eq. (37); clarifying the distinction between the distribution and the sampled vector would improve readability.
  5. [Particle Swarm Optimization stopping criterion] The stopping criterion JM/(umax(tf−t0)) > 0.95 should be justified; a trajectory can be near the reachable boundary with a lower thrust fraction if the control is not fully bang-bang, and the threshold is not derived from any error bound.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the superset claim in this paper is an empirical comparison of a nonlinearly optimized thrust-limited set against a linearized energy-limited set, not a result forced by the defining equations or by self-citation.

full rationale

The paper's derivation chain is self-contained. The energy-limited reachable set is defined in Eq. (20) and computed through the linearized hyperellipsoid of Eq. (34), with the construction reproduced in Eqs. (22)-(33) and reference [11] used only as a detailed derivation source rather than as an unverified load-bearing theorem. The thrust-limited set is defined by Eq. (21) and populated by independent particle-swarm optimization; the JM/(umax(tf-t0))>0.95 stopping rule is a sampling criterion, not a fitted parameter later renamed as a prediction. No equation in the paper is equal to another by construction, and the central superset claim is not derived from Eqs. (20)-(21): indeed, the exact set definitions imply the opposite inclusion, because any control satisfying ||u||≤umax also satisfies J_E≤(1/2)umax^2(tf-t0). The paper's own Results section concedes this methodological asymmetry ('if the same solution method is used to find an energy-optimal reachable set and then a mass-optimal reachable set, the energy-optimal reachable set would be a superset of the mass-optimal reachable set'). The claimed superset is therefore an artifact of comparing a nonlinear optimizer against a linearized energy ellipse, which is a correctness and approximation concern rather than a circular argument. The only self-citations ([11], [16]) provide parameter-free linearizations and prior related comparisons, and they do not smuggle in the target superset conclusion.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim leans on approximation and search heuristics rather than on fitted physical constants. The main free parameters are PSO exploration scales, convergence coefficients, azimuth sampling, and the 95% thrust-fraction boundary proxy. The core axioms are the CR3BP model, the validity of the linearized STM energy set, local optimality of the NLP solutions, and the representativeness of the PSO boundary search. No new physical entities are introduced.

free parameters (5)
  • PSO state-component standard deviations = sigma = [8e-3, 8e-4, 6e-3, 8e-3, 1e-2, 4e-3] in canonical units
    Hand-selected exploration scales informed by the prior linear analysis; directly control which initial-state deviations the swarm can discover.
  • PSO update coefficients beta and alpha = beta = 0.7, alpha = 0.5k
    Hand-tuned to balance swarm convergence and exploration; affect whether boundary points are reached.
  • Thrust fraction threshold for boundary proxy = JM / (umax (tf - t0)) > 0.95
    Ad hoc surrogate for the pointwise thrust limit ||u(t)|| <= umax; changes which trajectories count as reachable-set boundary samples.
  • Azimuth sampling grid for fitness weights = psi in {0, pi/6, ..., 11pi/6}, 12 directions
    Only the xy-plane is sampled at 30 degree increments; the claimed xy-plane superset is conditioned on this grid.
  • PSO stopping and fitness-switch thresholds = 20 iterations without gain; 80-iteration switch; 0.85 cost threshold
    Hand-tuned termination and switching criteria; affect completeness of the boundary search and cost of the run.
assumptions (5)
  • domain assumption CR3BP models Earth-Moon cislunar motion for this study
    The dynamics, periodic reference, and reachable sets are all defined in the circular restricted three-body model; unmodeled perturbations such as solar gravity and lunar eccentricity are not addressed, so the claim is conditional on this model.
  • domain assumption The linearized STM analysis yields a faithful energy-limited reachable set
    Equations 24-33 produce a hyperellipsoid; the central comparison uses this linearized set as the baseline against the nonlinear thrust-limited set.
  • domain assumption PSIOPT/ASSET solutions are sufficiently accurate and the continuation from energy-optimal to mass-optimal finds true optimal controls
    No global optimality certificate is given; local optima would shrink the thrust-limited set and could invalidate the superset claim.
  • ad hoc to paper The PSO search with the stated hyperparameters adequately samples the thrust-limited reachable set boundary
    The paper states the full set is incomplete due to imperfect sampling; the superset finding depends on this heuristic boundary search being representative.
  • ad hoc to paper Trajectories with thrust fraction above 0.95 lie on the boundary of the thrust-limited reachable set
    The paper uses JM/(umax(tf-t0)) > 0.95 as a proxy for the hard constraint ||u(t)|| <= umax, without an error bound connecting the threshold to the true boundary.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mass-Optimal Low-Thrust Forced Periodic Trajectories in the Earth-Moon CR3BP." pith.science (2026). https://pith.science/paper/3K6ZXWIX

@misc{pith2026250205140,
  author       = {Pith},
  title        = {Pith review of: Mass-Optimal Low-Thrust Forced Periodic Trajectories in the Earth-Moon CR3BP},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3K6ZXWIX}},
  note         = {Machine review of arXiv:2502.05140}
}
read the original abstract

In Cislunar space, spacecraft are able to exploit naturally periodic orbits, which provide operational reliability. However, these periodic orbits only exist in a limited volume. Enabled by low-thrust propulsion, spacecraft can produce a greater number of periodic trajectories in Cislunar space. We describe a methodology for producing mass-optimal trajectories that enforce periodic structure in the circular-restricted three body problem and study the thrust-limited reachable set around a reference trajectory. In this study, we find that the thrust-limited mass-optimal reachable set is a superset of the energy-limited energy-optimal reachable set in the xy-plane.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 15 canonical work pages

  1. [1]

    Solar sail parking in restricted three-body systems,

    C. R. McInnes, A. J. C. McDonald, J. F. L. Simmons, and E. W. MacDonald, “Solar sail parking in restricted three-body systems,” Journal of Guidance, Control, and Dynamics , V ol. 17, No. 2, 1994, pp. 399–406, 10.2514/3.21211

    doi:10.2514/3.21211 1994

  2. [2]

    Artificial Equilibrium Points in the Low-Thrust Re- stricted Three-Body Problem,

    M. Y . Morimoto, H. Yamakawa, and K. Uesugi, “Artificial Equilibrium Points in the Low-Thrust Re- stricted Three-Body Problem,” Journal of Guidance, Control, and Dynamics , V ol. 30, No. 5, 2007, pp. 1563–1568, 10.2514/1.26771

    work page doi:10.2514/1.26771 2007

  3. [3]

    Artificial Three-Body Equilibria for Hybrid Low-Thrust Propulsion,

    S. Baig and C. R. McInnes, “Artificial Three-Body Equilibria for Hybrid Low-Thrust Propulsion,”Jour- nal of Guidance, Control, and Dynamics, V ol. 31, No. 6, 2008, pp. 1644–1655, 10.2514/1.36125

    work page doi:10.2514/1.36125 2008

  4. [4]

    Dynamical structures in a low-thrust, multi-body model with ap- plications to trajectory design,

    A. Cox, K. C. Howell, and D. C. Folta, “Dynamical structures in a low-thrust, multi-body model with ap- plications to trajectory design,”Celestial Mechanics and Dynamical Astronomy, 2019, 10.1007/s10569- 019-9891-7

    doi:10.1007/s10569- 2019

  5. [5]

    Trajectory Design Leveraging Low-Thrust, Multi-Body Equi- libria and their Manifolds,

    A. Cox, K. C. Howell, and D. C. Folta, “Trajectory Design Leveraging Low-Thrust, Multi-Body Equi- libria and their Manifolds,” Journal of Astronautical Sciences, 2020, 10.1007/s40295-020-00211-6

    work page doi:10.1007/s40295-020-00211-6 2020

  6. [6]

    Periodic Orbits with Low-Thrust Propulsion in the Re- stricted Three-body Problem,

    M. Morimoto, H. Yamakawa, and K. Uesugi, “Periodic Orbits with Low-Thrust Propulsion in the Re- stricted Three-body Problem,” Journal of Guidance, Control, and Dynamics , V ol. 29, No. 5, 2006, pp. 1131–1139, 10.2514/1.19079

    work page doi:10.2514/1.19079 2006

  7. [7]

    Low-Thrust Orbit Dynamics and Periodic Trajectories in the Earth–Moon System,

    L. DeLeo and M. Pontani, “Low-Thrust Orbit Dynamics and Periodic Trajectories in the Earth–Moon System,” Aerotecnica Missili and Spazio, 2022, 10.1007/s42496-022-00122-9

    work page doi:10.1007/s42496-022-00122-9 2022

  8. [8]

    Artificial halo orbits for low-thrust propulsion spacecraft,

    S. Baig and C. R. McInnes, “Artificial halo orbits for low-thrust propulsion spacecraft,” Celestial Me- chanics and Dynamical Astronomy, 2009, pp. 321–335, 10.1007/s10569-009-9215-4

    work page doi:10.1007/s10569-009-9215-4 2009

Show all 20 references
  1. [9]

    New Equilibria and Dynamic Structures Under Continuous Optimal Feedback Control,

    A. Tsuruta, M. Bando, S. Hokamoto, and D. J. Scheeres, “New Equilibria and Dynamic Structures Under Continuous Optimal Feedback Control,” Journal of Guidance, Control, and Dynamics, V ol. 47, No. 10, 2024, pp. 2029–2040, 10.2514/1.G008270

    2024 doi

  2. [10]

    Natural and Forced Spacecraft Loitering in a Near Rectilinear Halo Orbit,

    C. Sandel and R. Sood, “Natural and Forced Spacecraft Loitering in a Near Rectilinear Halo Orbit,” Journal of Astronautical Sciences, V ol. 71, 2024, 10.1007/s40295-024-00446-7

    2024 doi

  3. [11]

    Generation of energy-optimal low-thrust forced periodic trajectories in the cr3bp,

    C. C. Merrill, J. Kulik, and D. Savransky, “Generation of energy-optimal low-thrust forced periodic trajectories in the cr3bp,” AAS/AIAA Astrodynamics Specialist Conference, Broomfield, CO , 2024, pp. AAS 24–206

    2024

  4. [12]

    A. E. Bryson, Applied optimal control: optimization, estimation and control. Routledge, 2018

    2018

  5. [13]

    ASSET: Astrodynamics Software and Science Enabling Toolkit,

    J. B. Pezent, J. Sikes, W. Ledbetter, R. Sood, K. C. Howell, and J. R. Stuart, “ASSET: Astrodynamics Software and Science Enabling Toolkit,” AIAA SCITECH 2022 Forum , 2022, pp. AIAA 2022–1131, 10.2514/6.2022-1131

    2022 doi

  6. [14]

    Enhancements to the Astrodynamics Software and Science Enabling Toolkit (ASSET),

    J. B. Pezent, J. Sikes, R. Sood, K. C. Howell, and J. R. Stuart, “Enhancements to the Astrodynamics Software and Science Enabling Toolkit (ASSET),” AAS/AIAA Astrodynamics Specialist Conference , 2023, pp. AAS 23–326

    2023

  7. [15]

    Pezent, On the Implementation and Applications of a High Performance Trajectory Optimization Toolkit

    J. Pezent, On the Implementation and Applications of a High Performance Trajectory Optimization Toolkit. PhD thesis, University of Alabama, 2024

    2024

  8. [16]

    Comparing Relative Reahcable Sets About Nearly Circular Orbits,

    J. Kulik, M. Zweig, and D. Savransky, “Comparing Relative Reahcable Sets About Nearly Circular Orbits,” AAS/AIAA Astrodynamics Specialist Conference, 2024, pp. AAS 24–164

    2024

  9. [17]

    Reachable set computation for spacecraft relative motion with energy-limited low-thrust,

    S. Lee and I. Hwang, “Reachable set computation for spacecraft relative motion with energy-limited low-thrust,” Aerospace Science and Technology, V ol. 77, 2018, pp. 180–188

    2018

  10. [18]

    Analysis on reachable set for spacecraft relative motion under low-thrust,

    H. Sun and J. Li, “Analysis on reachable set for spacecraft relative motion under low-thrust,” Automat- ica, V ol. 115, 2020, p. 108864

    2020

  11. [19]

    Particle swarm optimization,

    J. Kennedy and R. Eberhart, “Particle swarm optimization,” Proceedings of ICNN’95 - International Conference on Neural Networks, V ol. 4, 1995, pp. 1942–1948 vol.4, 10.1109/ICNN.1995.488968

    1995

  12. [20]

    Chaos-enhanced accelerated particle swarm optimization,

    A. H. Gandomi, G. J. Yun, X.-S. Yang, and S. Talatahari, “Chaos-enhanced accelerated particle swarm optimization,” Communications in Nonlinear Science and Numerical Simulation, V ol. 18, No. 2, 2013, pp. 327–340, https://doi.org/10.1016/j.cnsns.2012.07.017. 12

    2013 doi

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.