{"id":"a72ae6d7-e179-411c-a3c7-110e95b73412","arxiv_id":"2508.01769","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A grid search plus continuation in the planar Earth-Moon three-body problem reveals twelve families of two-impulse transfers from 167 km LEO to a 1:1 distant prograde lunar orbit.","lead":"This paper maps the many possible routes a spacecraft could take from a low Earth orbit to a distant prograde orbit around the Moon, using a simplified Earth-Moon gravity model. It identifies twelve distinct transfer families, with different flight times and fuel costs, which mission designers can compare.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim of twelve distinct transfer families rests on manual extraction and is explicitly conceded by the authors to be possibly over-split; the count and Table 4 ranges would change if same-revolution families F6-F9 or F11-F12 are connected.","rationale":"The reader's weakest_assumption identifies exactly the load-bearing issue: whether the twelve families are genuinely distinct components of the solution space. My reading of the manuscript confirms this is the most consequential vulnerability. The paper's own text in Section 4.2 admits that same-revolution families may belong to the same family and that no bridge was found, which means the manual extraction in Section 4.1 is under-tested. The novelty claim ('most of which are new or underexplored'), the family count, and the mission-selection guidance in Table 4 all depend on this distinctness. No code, data, or quantitative clustering is provided, so an independent reader cannot verify the separation. A secondary issue is the typo in Eq. (17), where A23 is written with ∂y_i/∂β_f instead of ∂y_i/∂TOF; if the implementation followed the printed formula, the continuation direction would be corrupted, but this is likely a transcription error and less central than the family-connectivity problem. The comparison with Mingotti et al. is appropriately contextualized as model-dependent, and the absence of a Δv_f = 0 solution is explained by the PCR3BP model choice, so that comparison is not a flaw. The numerical machinery appears sound in outline: backward propagation, constraint satisfaction to 5e-8, and continuation with a linear predictor are standard. The verdict should remain CONDITIONAL because the solution catalogue is potentially valuable but the twelve-family structure is not yet rigorously substantiated.","tokens_in":16907,"tokens_out":2248,"duration_ms":31078,"concrete_test":"Take one seed from each of F6-F9 and F11-F12 and run predictor-corrector continuation with pseudo-arclength parameterization and step sizes Δs = 1e-6 and 1e-7 in both directions, terminating only on constraint-tolerance failure or a singular Jacobian. If any two previously distinct labels are joined by a continuous curve of feasible solutions, the family count should be reduced and Table 4 ranges merged; if all remain disconnected at the finer step, the manual family labels are supported. Alternatively, if the raw solution data are released, perform a density-based connected-component clustering in (τf, βf, TOF) and compare the resulting component count with twelve.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that the LEO-to-DPO solution space contains at least twelve distinct transfer families with the ranges in Table 4—depends on the grid resolution and continuation step size actually separating solution components. Section 4.1 states the twelve families are 'manually extracted,' and Section 4.2 concedes: 'Families with the same revolution number possibly belong to the same family. However, with the grid search and trajectory continuation performed in this paper, the bridge between these families has not been found.' This is not a minor caveat: F6, F7, F8, and F9 all have revolution number eight, and F11 and F12 both have revolution number eleven, so the novelty claim and the per-family Δv, Δv_f, and TOF ranges used for mission selection would be materially wrong if those pairs are connected. The continuation step Δs = 1e-5 and grid spacings Δτf = π/5000 and Δβf = 1e-4 could jump over narrow connecting corridors, and no connected-component analysis is performed. The solutions themselves may all be valid, but their organization into twelve families is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper constructs bi-impulsive transfers from a 167 km circular low Earth orbit to a 1:1 distant prograde orbit around the Moon in the planar Earth-Moon circular restricted three-body problem. The transfers are parameterized by the DPO phase, velocity ratio, and time of flight; backward propagation generates initial guesses, which are corrected to satisfy the LEO radius and tangentiality constraints. A linear predictor derived from the variational equations is used in a predictor-corrector continuation to extend the solution set. The authors report 5,663,373 solutions and identify twelve transfer families, with total Δv from 3.319 to 3.758 km/s and TOF from 4 to 112 days (Table 4). They compare their minimum-Δv solution with prior PBCR4BP solutions and attribute the differences to the absence of solar gravity perturbation.","tokens_in":17183,"tokens_out":5048,"duration_ms":59915,"significance":"If the family identification is robust, this is a useful global map of the LEO-to-DPO transfer solution space and a practical catalog for mission designers. The numerical pipeline is carefully executed: the integrator tolerance is 1e-13, the constraint residual threshold is 5e-8, no fitted parameters are used to define the families, and the linear predictor follows from the variational equations rather than from the target data. The explicit comparison with Mingotti et al. is also a strength. The main caveat is that the twelve-family taxonomy is manually extracted and the paper itself concedes that same-revolution families may be connected, so the central claim of twelve distinct families is not yet fully established.","major_comments":[{"comment":"The central claim that there are twelve distinct transfer families is based on manual extraction, and the text explicitly concedes in §4.2 that families with the same revolution number may belong to the same family and that the bridge between them has not been found. Since F6, F7, F8, and F9 all have eight revolutions, and F11 and F12 both have eleven revolutions, the family count and the per-family Δv, Δv_i, Δv_f, and TOF ranges in Table 4 are not established as distinct unless the solution components are shown to be separated. Please provide quantitative evidence of separation, for example by analyzing connected components in the (τ_f, β_f, TOF) parameter space with finer grid/continuation steps, or explicitly restate the claim as 'twelve branches observed under the current grid and continuation resolution' and temper the corresponding novelty statements.","section":"§4.1, §4.2, Table 4"},{"comment":"The scatter in the (TOF, Δv_i) distribution of family F2 is attributed to the constraints ψ_i not being satisfied rigorously, but the acceptance criterion in §3.2.1 is ||ψ_i|| < 5e-8. This explanation is inconsistent with the stated tolerance unless the scatter is within the numerical error implied by that threshold. The paper should quantify how the 5e-8 constraint residual translates into uncertainty in Δv_i and determine whether the observed scatter exceeds that uncertainty. Since the Δv_i ranges in Table 4 are presented as guidance for launch-vehicle selection, this issue is load-bearing for the practical conclusions drawn from the family analysis.","section":"§4.2, Fig. 8"}],"minor_comments":[{"comment":"The phrase 'Gird search' should read 'Grid search'.","section":"Highlights"},{"comment":"In the expression for A23, the terms involving (v_i + x_i + μ) and (∂v_i/∂TOF + ∂x_i/∂TOF) should contain ∂y_i/∂TOF and ∂v_i/∂TOF; as printed they repeat ∂y_i/∂β_f.","section":"Eq. (17)"},{"comment":"The identity matrix in the state transition matrix initial condition should be 4×4 to match the four-dimensional state vector, not 6×6.","section":"Eq. (30)"},{"comment":"There are several typographical errors, including 'thethe', 'preform', and other repeated or misspelled words; a careful proofreading pass is needed.","section":"§1, §4.2"},{"comment":"The explanation of the blank region in the (TOF, Δv) map is vague; please specify the Δv < 3.45 km/s continuation threshold and clarify whether the blank region is a search artifact or an actual absence of solutions.","section":"§4.1, Fig. 6"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the journal's scope and the numerical exploration appears carefully executed. My main concern is that the twelve-family taxonomy is not rigorously justified, which affects the novelty claim and the practical ranges in Table 4; this is fixable either by adding a connected-component analysis or by softening the claim. I also recommend a careful proofreading pass before resubmission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper is a serious numerical survey of bi-impulsive transfers from 167 km LEO to a 1:1 distant prograde lunar orbit in the planar circular restricted three-body problem. The genuinely new thing is the breadth: the authors grid-search the insertion parameters and then continue the solutions, ending with about 5.6 million numerically converged transfers, from which they extract twelve families. The per-family cost ranges in Table 4 are the kind of thing mission designers actually want to browse, and the comparison against Mingotti et al. is fair.\n\nWhat the paper does well: the numerics are tight (integration tolerance 1e-13, constraint residual below 5e-8), the linear predictor is a reasonable, scenario-specific application of standard continuation theory, and the authors are honest about the model limitation—their PCR3BP solutions cannot reproduce Mingotti's Δv_f = 0 insertions because solar gravity is absent.\n\nThe soft spots are real but not fatal. The family taxonomy is manually extracted, and the authors concede in Section 4.2 that families with the same revolution number 'possibly belong to the same family.' That is candid, but it means the headline claim of twelve distinct families is a hypothesis, not a proven partition. If F6–F9 or F11–F12 are connected by a corridor the grid or continuation jumped over, the family count and the Table 4 ranges would shift. The authors could settle this with a connected-component analysis on the accumulated solutions; they didn't. Also, continuation is only run for Δv < 3.45 km/s, so the upper end of the solution space is less complete—they acknowledge this, but the map is therefore not the full census the abstract might imply.\n\nMinor issues: Eq. (17) has a typo in A23 (the third term uses ∂y_i/∂β_f instead of ∂y_i/∂TOF). No code or data are released, which makes it harder to verify the family extraction independently. The scatter noted in F2's (TOF, Δv_i) plot is attributed to constraints 'not satisfied rigorously' even though the stated residual tolerance is 5e-8—that explanation doesn't quite add up and deserves a clearer look.\n\nWho should read it: anyone working on cislunar transfer design, or on solution-space methods for three-body problems. It deserves a proper peer review; the underlying survey is valuable. The revision should either strengthen the family claim with a connected-component check or soften it to 'heuristic clusters.' I would accept it in review with that request.","headline":"A useful and honest numerical census of LEO-to-DPO transfers, but the twelve-family taxonomy is a manual interpretation rather than a proven partition.","tokens_in":17685,"tokens_out":3058,"would_cite":true,"duration_ms":33531,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["70F07","70F15","70M20"],"pacs":[],"model":"deepseek-v4-flash","headline":"A grid-search and continuation study finds 5,663,373 valid bi-impulsive transfers from a 167 km low Earth orbit to the Moon's 1:1 distant prograde orbit, organized into twelve families.","keywords":["Planar circular restricted three-body problem","Distant prograde orbit","Earth-Moon transfer","Transfer families","Grid search","Trajectory continuation","Bi-impulsive transfer","Lunar mission design"],"falsifier":"A finer grid search (e.g., $\\tau_f$ step below $\\pi/5000$ and $\\beta_f$ step below $0.0001$) that finds a continuous solution curve bridging any pair among F6-F9 or F11-F12 would collapse the corresponding family labels and change the reported ranges.","tokens_in":16715,"feed_emoji":"🌙","tokens_out":9132,"duration_ms":83762,"temperature":0.7,"pith_summary":"The paper tries to establish that the bi-impulsive transfer problem from a 167 km circular low Earth orbit to the Moon's 1:1 distant prograde orbit has a large and structured solution space, not just a few hand-built trajectories. By seeding a grid search with backward-propagated states from the target orbit and then extending the solutions with a linear predictor-corrector continuation, the authors obtain 5,663,373 valid solutions and sort them into twelve transfer families, most of which they argue are new or underexplored. The families span total $\\Delta v$ from about 3.32 to 3.76 km/s, Earth-injection impulses of 3.12-3.15 km/s, Moon-insertion impulses of 0.20-0.61 km/s, and flight times of 4 to 112 days. A sympathetic reader would care because this turns a mission-design question into a catalog: fast transfers for crewed missions (F1, F2) and low-fuel transfers for cargo or surveillance missions (F4, F6-F12) can be picked directly from the mapped families.","feed_headline":"5.66 million trajectories map 12 Earth-Moon transfer families","feed_subtitle":"A full catalog of Moon-bound transfers now spans 3.32-3.76 km/s fuel and 4-112 day flight times.","key_machinery":"The construction rests on two linked objects. First, a backward-propagation grid search: initial guesses are generated by choosing a time phase $\\tau_f$ along the DPO and a velocity ratio $\\beta_f$ at the insertion point, propagating backward from the DPO for up to $12\\pi$ time units, and keeping states that nearly satisfy the low-Earth-orbit constraints $\\boldsymbol{\\psi}_i$. Second, a predictor-corrector continuation whose predictor is a linear map derived from the state transition matrix $\\mathbf{\\Phi}(t_i, t_f)$: the feasible direction $\\delta\\boldsymbol{y}$ is the right singular vector $\\boldsymbol{V}_3$ from the singular value decomposition of the $2 \\times 3$ constraint Jacobian $\\boldsymbol{A} = \\partial\\boldsymbol{\\psi}_i/\\partial\\boldsymbol{y}$, so each new solution is predicted by $\\tilde{\\boldsymbol{y}}_1 = \\boldsymbol{y}_0 + \\delta\\boldsymbol{y}\\,\\Delta s$ and then corrected with a least-squares solver. This mechanism lets the authors extend sparse initial guesses into long continuous families.","core_discovery":"In the planar circular restricted three-body model of the Earth-Moon system, transfers from a 167 km circular low Earth orbit to a 1:1 distant prograde orbit are not scarce: a dense grid of insertion-phase and velocity-ratio parameters, together with continuation, yields 5,663,373 corrected solutions. These solutions organize into twelve families, labeled F1-F12, distinguished by the number of Earth revolutions, the shape of the solution curves in the (TOF, $\\Delta v$) plane, and their construction-parameter ranges. The central discovery is the structure of this solution space: all twelve families are interior transfers (their apogees remain within a few Earth-Moon distances), the minimum-$\\Delta v$ solution (3.319 km/s, in F7) exploits a high-altitude lunar flyby, and comparison with earlier four-body solutions [24] shows that the absence of solar gravity raises the Moon-insertion impulse, so zero-$\\Delta v_f$ transfers do not appear in this model.","pith_inferences":["If the same grid-search-plus-continuation pipeline were applied to other resonant DPOs (2:1, 3:1), analogous family catalogs might emerge, giving a broader classification of lunar distant-orbit transfers.","The paper's own caveat that families with equal revolution numbers may be connected suggests the true solution space could be organized into a few continuous manifolds indexed by revolution number, with the twelve labels as a coarse partition.","The linear predictor derived here is not specific to the DPO target; it could be reused to map transfer families to other prescribed orbits in the PCR3BP, such as distant retrograde orbits or low lunar orbits.","The reported $\\Delta v$ ranges are PCR3BP values; a higher-fidelity ephemeris or four-body model would likely shift them, so the catalog should be treated as a first-cut mission-design reference."],"forward_implications":["Mission designers can use Table 4 as a reference: F1 and F2 give fast transfers (4-16 days) at higher fuel cost, while F4 and F6-F12 give lower fuel consumption at 67-112 days.","The minimum-$\\Delta v$ trajectory in F7 (3.319 km/s total) demonstrates that a high-altitude lunar flyby is the mechanism that lowers insertion impulse in this model.","The absence of any exterior or zero-insertion-impulse transfer in the PCR3BP solution space indicates that solar gravity is required to obtain single-impulse DPO transfers.","The 5,663,373-solution database provides a comprehensive baseline against which higher-fidelity four-body or ephemeris models can be compared."],"supporting_citations":[{"why":"Prior construction of transfers to the same 1:1 DPO in the Sun-Earth/Moon four-body model; its zero-insertion-impulse solutions are the direct baseline for the comparison in Table 5.","marker":"[24]"},{"why":"Grid-search and continuation framework for LEO-to-low-Moon-orbit transfers that this paper adapts; supplies the interior-transfer definition and the approach to identify families.","marker":"[10]"},{"why":"Extends the grid-search methodology to low-energy, long-TOF lunar transfers, informing the continuation strategy and the long-TOF families.","marker":"[12]"},{"why":"Introduces the backward-propagation strategy for generating initial guesses from target orbits, which the paper adopts for the DPO.","marker":"[21]"},{"why":"Original discovery of the g and g' families in the Hill problem, the source of the DPO target orbit.","marker":"[25]"},{"why":"Shows the two-branch structure of g/g' families in the PCR3BP, used to describe the selected 1:1 DPO.","marker":"[38]"},{"why":"Provides the nonlinear predictor for trajectory continuation, the context against which the paper's simpler linear predictor is chosen.","marker":"[43]"}],"fun_headline_variants":["5.66M Moon-bound paths reveal 12 transfer families","12 families of Earth-Moon transfers now fully mapped","5.6M trajectories chart low-energy lunar transfers","Cislunar transfer atlas: 5.66M solutions, 12 families","From LEO to lunar orbit: 5.66 million path families"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the chosen grid spacing and continuation steps are fine enough to separate the solution set into genuine pieces, so the twelve reported families are not just fragments of one larger connected set of transfers; the paper itself notes that families with the same number of Earth revolutions could belong to a single family.","fun_headline_variants_meta":{"raw":{"variants":["5.66M Moon-bound paths reveal 12 transfer families","12 families of Earth-Moon transfers now fully mapped","5.6M trajectories chart low-energy lunar transfers","Cislunar transfer atlas: 5.66M solutions, 12 families","From LEO to lunar orbit: 5.66 million path families"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000196,"raw_usage":{"total_tokens":1383,"prompt_tokens":990,"completion_tokens":393,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":606,"completion_tokens_details":{"reasoning_tokens":305}},"tokens_in":606,"tokens_out":393,"duration_ms":4497,"temperature":1.0,"reasoning_tokens":305,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T05:23:44.094621+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A finer grid search (e.g., $\\tau_f$ step below $\\pi/5000$ and $\\beta_f$ step below $0.0001$) that finds a continuous solution curve bridging any pair among F6-F9 or F11-F12 would collapse the corresponding family labels and change the reported ranges.","supporting_citations":[{"cited_title":"Hénon, Numerical exploration of the restricted problem, v, Astron","cited_arxiv_id":null,"evidence_quote":"Original discovery of the g and g' families in the Hill problem, the source of the DPO target orbit."},{"cited_title":"Control Dyn","cited_arxiv_id":null,"evidence_quote":"Shows the two-branch structure of g/g' families in the PCR3BP, used to describe the selected 1:1 DPO."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the nonlinear predictor for trajectory continuation, the context against which the paper's simpler linear predictor is chosen."}],"review_version":1}