{"id":"933b680f-def1-45e6-8ff1-1f752a5cc645","arxiv_id":"1908.04300","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In-situ planet-planet scattering at 0.1 to 1 AU naturally produces eccentric warm Jupiters, and a two-population mixture with about 60% scattered systems matches the observed solitary warm Jupiter eccentricity distribution.","lead":"This paper uses N-body simulations to show that warm Jupiter planets scattering off each other near their star can produce the stretched, non-circular orbits seen in many real systems. It finds that the observed eccentricities of solitary warm Jupiters are consistent with about 60% of them having had a violent scattering history.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The inferred ~60% scattering fraction is contingent on treating all planet-planet collisions as perfect mergers (Appendix A); the paper itself flags this simplification, and the grazing, vcoll/v0~1 collisions it finds are likely to be hit-and-run in reality, altering f_scat and the mixture-model…","rationale":"The reader's verdict was CONDITIONAL, and the reader's weakest assumption identified the same load-bearing concern: the simulation outcomes, and hence f_scat, depend on the treatment of planet-planet collisions as completely inelastic mergers. My stress-test pass agrees that this is the most load-bearing weakness of the central claim. The paper is otherwise careful: it systematically explores parameter dependence, uses a well-tested N-body integrator (rebound/IAS15 and Mercurius), checks timestep convergence, and presents a transparent likelihood analysis in Section 3.4. The two-population model is a reasonable way to reconcile the observed low-eccentricity peak with the simulated high-eccentricity tail. However, the numerical value of the scattered fraction is not a direct observable; it is an output of a model whose physical collision treatment is idealized. Appendix A explicitly acknowledges this simplification and even provides the diagnostic distributions (impact parameter and vcoll/v0) needed to gauge its importance. Since those diagnostics sit in the hit-and-run regime, the concern is concrete and testable. A dedicated rerun with a more realistic collision prescription would either confirm the 60% fraction or show that it shifts beyond the quoted uncertainty. Either outcome is scientifically useful, and the paper's existing CONDITIONAL verdict appropriately reflects this unresolved dependence.","tokens_in":21559,"tokens_out":7163,"duration_ms":75617,"concrete_test":"Re-run the fiducial 3313-system sample (or a statistically matched subset of ~1000 systems) with a collision model that distinguishes merging from hit-and-run, using a standard giant-impact criterion based on impact parameter and vcoll/v0 (e.g., Leinhardt & Stewart 2012 or Genda et al. 2012), applied to the distributions shown in Fig. A1. Then refit the mixture model of Section 3.4 to the new one-planet eccentricity distribution and compare the MLE and 95% confidence interval of F. If the inferred F shifts by more than the quoted 95% interval (i.e., the scattered fraction 1-F moves outside [0.46, 0.82]), the 'roughly 60%' claim is not robust to the collision prescription.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (abstract, Section 3.4) is a quantitative statement: the observed eccentricity distribution of solitary WJs is consistent with roughly 60% of systems having undergone in-situ scattering. This number is the result of a two-parameter mixture fit (Eq. 4-5) in which f_scat is the eccentricity distribution of one-planet survivors from the fiducial N-body sample. That template depends directly on the collision prescription: rebound's built-in routine merges any pair whose separation falls below the sum of the physical radii, conserving mass and momentum. Appendix A states this treatment is 'clearly a simplification' and shows that the simulated collisions are predominantly grazing, with bcoll/(R1+R2) ~ 1 and vcoll/v0 ~ 1 (Fig. A1). In giant-planet impact physics, this parameter regime is precisely where hit-and-run encounters, not mergers, are expected for a substantial fraction of impacts. If hit-and-run occurs, an encounter that currently removes a planet and creates a low-eccentricity remnant would instead leave both planets on altered orbits; the continued scattering would typically produce more ejections and higher survivor eccentricities, changing the shape of f_scat. Because the mixture fit uses only the shape of f_scat, the inferred F (and hence the claimed ~60% scattered fraction) is directly contingent on the merger assumption. The quoted 95% confidence interval for F (0.18-0.54) accounts only for sampling noise under the assumed model, not for this physical uncertainty in the collision outcome.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper investigates whether in-situ planet-planet scattering among 3-4 giant planets at 0.1-1 AU can explain the eccentricity distribution of warm Jupiters. The authors run a large suite of rebound N-body simulations with a range of masses, spacings, and initial semi-major axes, including general relativistic precession, and catalogue the branching ratios into one-planet and two-planet systems and the resulting eccentricity distributions. They find that two-planet systems arise almost exclusively from planet-planet collisions and are dynamically cold, while one-planet systems come from a mix of collisions and ejections and have higher eccentricities. Comparing with the observed sample of 83 solitary WJs and 24 WJs with an external giant companion, they find that the combined eccentricity distribution is reproduced, but the separate one- and two-planet distributions are not. They then fit a two-population mixture model to the solitary WJ eccentricities, with a quiescent half-Gaussian component and a scattered component taken from the fiducial simulations, obtaining a maximum-likelihood quiescent fraction F≈0.35 and thus concluding that roughly 60% of solitary WJs have undergone in-situ scattering.","tokens_in":21897,"tokens_out":5887,"duration_ms":65937,"significance":"If the central claim holds, the paper establishes planet-planet scattering at sub-AU distances as a common and effective mechanism for producing eccentric warm Jupiters, with a sizeable quiescent component. The study is systematic and clearly presented: it uses standard, publicly available tools (rebound/reboundx), checks timestep convergence, integrates two-planet systems to 10^8 orbits and a subset to 10^9 orbits, and provides branching ratios as a function of mass, spacing, and semi-major axis. The mixture-model analysis is transparent, with a maximum-likelihood fit and confidence contours. The main quantitative conclusion, however, is contingent on the treatment of planet-planet collisions as perfect inelastic mergers, a simplification the paper explicitly flags in Appendix A but does not quantify. Because the simulated single-planet eccentricity template directly controls the inferred 60% fraction, this is a load-bearing uncertainty.","major_comments":[{"comment":"The headline result that roughly 60% of solitary WJs have experienced in-situ scattering is obtained by fitting a mixture model (Eq. 4-5) in which fscat is the single-planet eccentricity template from the fiducial simulations. Those simulations adopt rebound's built-in collision routine, which merges any pair whose separation falls below the sum of the physical radii, conserving mass and momentum. Appendix A itself calls this treatment 'clearly a simplification,' and Fig. A1 shows that the collisions are predominantly grazing, with bcoll/(R1+R2)~1 and vcoll/v0~1. In giant-planet impact physics, this is precisely the regime where a substantial fraction of encounters are expected to be hit-and-run rather than mergers. A hit-and-run outcome would change the branching ratios (fpp, fej, f1p/f2p) and alter the survivor eccentricity distribution, directly shifting the fitted F. The quoted 95% interval F=0.18-0.54 does not include this systematic effect. Please provide a sensitivity test, for example a simple hit-and-run prescription or a bracketing model based on published giant-impact results, or at minimum rephrase the abstract and Section 3.4 to state explicitly that the 60% estimate is conditional on the perfect-merger assumption.","section":"Section 3.4 and Appendix A"},{"comment":"The likelihood treats the simulated template fscat as fixed, so the confidence contours in Fig. 11 reflect only Poisson/sampling noise in the observed eccentricities, not uncertainty in the shape of fscat. The template is sensitive to the assumed initial planet masses and planet number: for example, the near-eq-mass simulations produce one-planet systems with mean eccentricity 0.47 versus 0.30 in fiducial (Table 3), and the 4-planet run has f1p=0.75 versus 0.52 for fiducial (Tables 1 and 4). Since F is inferred by matching the shape of fscat, the true uncertainty range for the quoted scattering fraction is larger than the plotted 95% contour. A propagation of the mass-function and planet-number variations, and ideally the collision prescription, is needed before the claim 'roughly 60% of systems' can be taken at face value.","section":"Section 3.4, Eq. (5), and Tables 3-4"},{"comment":"The mixture model is fit only to the eccentricities of 83 solitary WJs, but the same scattering model has a strong implication for the relative numbers of one- and two-planet systems that is not incorporated into the inference. The fiducial simulation yields N1/N2 ≈ 1 due to f1p=0.52, f2p=0.48, and all other three-planet runs give N1/N2 between roughly 0.5 and 1.5, whereas the observed ratio is N1/N2 ≈ 2.3-3.5 (Section 3.3). If 65% of solitary WJs are scattered remnants, the same scattering events would produce a large population of compact two-planet systems, which is not observed. The paper discusses this discrepancy qualitatively, but a joint likelihood over both the eccentricity distributions and the multiplicity ratio would be a more robust basis for the two-population conclusion. As written, the 'scattered + quiescent' model is not self-consistently tested against all the data presented.","section":"Sections 3.1, 3.3, and 4.2"}],"minor_comments":[{"comment":"The Gaussian kernel-density estimate used to construct a smooth fscat(e) is not accompanied by a stated bandwidth; the inferred F may depend on this choice, so please report the bandwidth and check the sensitivity.","section":"Section 3.4"},{"comment":"The likelihood neglects uncertainties in the observed eccentricities; a short Monte Carlo test perturbing the eccentricities within their error bars would justify this assumption.","section":"Section 3.4"},{"comment":"For fiducial-K-5 the initial eccentricities are increased to make instabilities practical, so the K-dependence is convolved with a change in initial eccentricity; this is explained in the text but should be flagged more prominently in Table 1 or its caption.","section":"Section 2.4 and Table 1"},{"comment":"The abstract states 'roughly 60%' scattered, which corresponds to 1-F at the MLE, but the 95% confidence range (F=0.18-0.54, i.e., 46-82% scattered) is quite wide; quoting this range in the abstract or conclusions would be more informative.","section":"Abstract and Section 3.4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a solid, well-organized contribution that fits the scope of MNRAS. The main risk is the perfect-merger collision assumption in the simulations that provide the fscat template; if the authors can bracket this effect with a sensitivity test, the central claim would be much more robust. I do not have concerns about novelty or overlap, but the mixture-model confidence intervals should be presented as conditional on the simulation assumptions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper fills the gap between the cold-Jupiter scattering literature (several AU) and the hot-Jupiter scattering studies with tight spacings: it is the first systematic N-body exploration of the 0.1–1 AU regime where the Safronov number is order unity and collisions compete with ejections. The thing to know before reading: the qualitative picture is solid, but the headline 'roughly 60% scattered' is model-dependent and should not be quoted as a measured fraction.\n\nThe new content is genuinely useful. For the fiducial three-planet, K=4 setup, ~65% of first planet losses are planet-planet collisions and ~29% ejections; two-planet survivors arise exclusively from collisions, are compact (90% with aout/ain < 3), and have low eccentricities and mutual inclinations; one-planet survivors peak near e ≈ 0.25 with a tail to ~0.9. These branching ratios will be reference numbers. I also credit the observational honesty: the combined eccentricity distribution of one- and two-planet systems matches the observed WJs, but the authors split the samples, show scattering cannot separately reproduce either, test the undetected-companion rescue via RV cuts, and reject it. The numerics are standard and careful — REBOUND/IAS15 plus Mercurius, GR apsidal precession, timestep convergence checked, long-term integrations of a subsample — and Appendix A quantifies the collision parameters rather than hiding them.\n\nThe soft spot is the fraction. Section 3.4 fits a mixture of a half-Gaussian quiescent component and the fiducial simulation's scattered component; the MLE is F ≈ 0.35 with F in 0.18–0.54 at 95% confidence, and that interval covers only sampling noise. The scattered template assumes every planet-planet collision is a perfect inelastic merger, while Appendix A shows the collisions are mostly grazing with vcoll/v0 near unity — the regime where giant-planet impact physics predicts substantial hit-and-run. Hit-and-run would change the branching ratios, the survivor eccentricities, and hence the fitted F. The authors flag the merger assumption, which I credit; the abstract's 'roughly 60%' just overstates the certainty. Minor: observed eccentricity uncertainties are neglected and the quiescent component's shape is assumed; both are standard here and neither changes my read.\n\nThis is for exoplanet dynamicists and anyone building formation or population models for close-in giants. It deserves a serious referee. The referee should ask for a sensitivity test on the collision prescription and an error budget on F that includes model uncertainty; I would expect minor-to-moderate revision, not rejection. My bottom line: send it to peer review, and ask the authors to temper the abstract.","headline":"First systematic N-body study of scattering in the warm-Jupiter regime; the qualitative two-population picture holds, but the ~60% scattered fraction is softer than the abstract implies.","tokens_in":22396,"tokens_out":10709,"would_cite":true,"duration_ms":82206,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Planet-planet scattering at sub-AU distances can excite the eccentricities of warm Jupiters, and the observed eccentricity distribution of solitary warm Jupiters implies that roughly 60 percent of them underwent such scattering.","keywords":["warm Jupiters","planet-planet scattering","eccentricity excitation","N-body simulations","planetary collisions","planet ejections","two-population model"],"falsifier":"A decisive observational check would be long-baseline radial-velocity monitoring of the 83 solitary warm Jupiters: the scattering model predicts that many of these systems actually carry close companions with velocity semi-amplitudes of 10 to 100 m/s, and the mixture model predicts a specific full shape for the eccentricity distribution. If the companions are not found and the low-eccentricity peak is wider or narrower than the half-Gaussian-plus-scattering-tail mixture predicts, the 60-percent fraction would need substantial revision. A complementary calculation would simulate the grazing giant-planet collisions that dominate these experiments to test whether mergers, rather than hit-and-run events, set the true branching ratios.","tokens_in":21373,"feed_emoji":"🪐","tokens_out":14379,"duration_ms":128989,"temperature":0.7,"pith_summary":"This paper asks whether the elongated orbits of warm Jupiters — giant planets with orbital periods of roughly 10 to 300 days — can be explained by the planets having violently scattered one another soon after formation. The authors simulate systems of three or four giant planets packed into initially unstable orbits between 0.1 and 1 AU, a distance range where the balance between planet collisions and ejections is roughly even, and catalogue what the surviving systems look like. They find that scattering yields two distinct survivor types: compact two-planet systems with low eccentricities, which form only when planets collide and merge, and single planets with higher eccentricities, which form once at least one planet is ejected. Comparing the simulated eccentricities with radial-velocity observations, they conclude that the eccentricity distribution of solitary warm Jupiters is best matched by a mix of two populations: roughly 60 percent that underwent scattering and 40 percent that had a quiet dynamical history. If correct, this establishes in-situ planet-planet scattering as a common mechanism for producing eccentric warm Jupiters, alongside a quiescent low-eccentricity population.","feed_headline":"Planet scattering explains 6 in 10 eccentric warm Jupiters","feed_subtitle":"Close-in planet collisions and ejections, plus a quiet population, match observed orbits.","key_machinery":"The central object is the N-body scattering experiment: three or four giant planets started on nearly circular, nearly coplanar orbits spaced by a fixed number of mutual Hill radii (3 to 5), with the innermost planet drawn uniformly from 0.1 to 1 AU, integrated with a high-accuracy integrator that includes general-relativistic apsidal precession and treats any planet-planet contact as a perfectly inelastic merger conserving mass and momentum. The physical quantity that determines the outcome is the Safronov number — the squared ratio of surface escape velocity to orbital velocity — which is of order unity for warm Jupiters, placing them in the regime where encounters split between collisions and ejections. The branching between those two channels controls both the final multiplicities and the eccentricity distribution: collisions produce compact, low-eccentricity two-planet systems, while ejections raise the eccentricities of surviving single planets. The comparison tool is a two-population mixture model, $f(e) = F f_{\\mathrm{circ}}(e) + (1-F) f_{\\mathrm{scat}}(e)$, in which the quiescent component is a half-Gaussian peaked at $e = 0$ with a fitted width and the scattered component is the simulated eccentricity distribution; maximizing the likelihood over $F$ and the width gives the roughly 60-percent scattered fraction.","core_discovery":"At distances of 0.1 to 1 AU, the escape velocity from a giant planet's surface is comparable to its orbital velocity, so close encounters between planets end in a mixture of collisions and ejections rather than the ejection-dominated outcome familiar from cold Jupiters at several AU. In the authors' fiducial simulations of three planets with masses 0.5, 1, and 2 Jupiter masses spaced by four mutual Hill radii, the first planet loss is a planet-planet collision 65 percent of the time, an ejection 29 percent of the time, and a collision with the star 6 percent of the time; the final state is 52 percent one-planet and 48 percent two-planet systems. Every surviving two-planet system was produced by a collision, which gives these systems low eccentricities (average inner-planet eccentricity 0.15, with 90 percent below 0.34), low mutual inclinations (90 percent below about 13 degrees), and compact spacing (90 percent with a semi-major axis ratio below 3). One-planet systems carry higher eccentricities (average 0.30, with 90 percent below 0.55) because each suffered at least one ejection. When the simulated single-planet eccentricity distribution is combined with a quiescent, low-eccentricity population in a mixture model, the observed eccentricities of 83 solitary warm Jupiters are reproduced with roughly 60 percent of systems having scattered and the rest quiescent; the same model also exposes tensions, since scattering produces an excess of circular two-planet systems and more compact pairs than are observed.","pith_inferences":["A sharper test that the authors do not perform: the scattered fraction should rise with host-star metallicity, since more metals favor forming multiple giant planets; measuring eccentricity distributions in metal-poor versus metal-rich subsamples would test the 60-percent split directly.","If hot and warm Jupiters have distinct histories as the paper argues, a testable consequence is that hot Jupiters should show a very different close-companion frequency than the compact two-planet systems this model produces.","Because the simulations show that grazing collisions dominate the merger events, the assumption of perfect mergers could be the first thing to revisit: if many grazing encounters are hit-and-run, more planets survive and the collision-to-ejection ratio shifts, which would change the fitted scattering fraction.","An observational follow-up suggested by the model: eccentric solitary warm Jupiters should more often show signs of a depleted outer region (no distant companions) than circular ones, a difference that could be tested with extended radial-velocity monitoring."],"forward_implications":["Planet-planet scattering at 0.1 to 1 AU is an effective eccentricity-excitation mechanism: the simulated single-planet eccentricity distribution reproduces the moderate-to-high eccentricity tail observed among solitary warm Jupiters.","Scattered systems carry recognizable hallmarks: compact two-planet systems with low mutual inclinations arise exclusively from collisions, while a high-eccentricity single planet implies at least one ejection occurred.","The observed solitary-warm-Jupiter eccentricities imply a two-population history: roughly 60 percent scattered in situ and roughly 40 percent remained quiescent, the latter consistent with the observed warm Jupiters that host low-mass neighbors.","Extrapolating the same scattering inward to hot-Jupiter distances would overproduce compact two-planet systems that are not observed, suggesting hot Jupiters and warm Jupiters often have distinct formation and dynamical histories.","The observed eccentric, hierarchical two-planet systems with semi-major axis ratios above 3 are not reproduced by sub-AU scattering, so those systems require a different origin."],"supporting_citations":[{"why":"Supplies the rebound N-body code in which all scattering experiments are integrated.","marker":"Rein & Liu 2012"},{"why":"Provides the IAS15 integrator used during the close-encounter phase of the simulations.","marker":"Rein & Spiegel 2015"},{"why":"Provides the Mercurius hybrid integrator used for the long-term follow-up of two-planet systems.","marker":"Rein & Tamayo 2015"},{"why":"Establishes that collision remnants emerge with low eccentricities, the expectation the two-planet results confirm.","marker":"Ford et al. 2001"},{"why":"Shows unequal-mass scattering produces fewer close encounters, the mass-dependence this paper's near-equal-mass runs build on.","marker":"Ford & Rasio 2008"},{"why":"Provides the standard cold-Jupiter scattering results at several AU that this paper contrasts with sub-AU outcomes.","marker":"Chatterjee et al. 2008"},{"why":"Previous scattering study of close-in planets that found inefficient eccentricity excitation; this paper explores a wider spacing parameter space.","marker":"Petrovich et al. 2014"},{"why":"Characterized observed warm Jupiters with giant companions, providing the two-planet sample and the finding that high-eccentricity migration is inconsistent with it.","marker":"Antonini et al. 2016"},{"why":"Finds that a substantial fraction of warm Jupiters have low-mass neighbors, cited as evidence for the quiescent component.","marker":"Huang et al. 2016"},{"why":"Proposed the two-population warm Jupiter picture based on metallicity, which the scattered-plus-quiescent model echoes.","marker":"Dawson & Murray-Clay 2013"}],"fun_headline_variants":["Warm Jupiter orbits hint 60% had violent scattering pasts","Collisions and ejections shape warm Jupiters' eccentricities","60% of solitary warm Jupiters may have been scattered","Warm Jupiter scatterings: collisions quench, ejections excite","Scattered warm Jupiters: mixtures of collisions and ejections"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole picture rests on treating every planet-planet collision as a perfect merger that conserves mass and momentum; if real giant-planet collisions are often hit-and-run or shed substantial mass, the balance between collisions and ejections changes and the fitted 60-percent scattering fraction would shift.","fun_headline_variants_meta":{"raw":{"variants":["Warm Jupiter orbits hint 60% had violent scattering pasts","Collisions and ejections shape warm Jupiters' eccentricities","60% of solitary warm Jupiters may have been scattered","Warm Jupiter scatterings: collisions quench, ejections excite","Scattered warm Jupiters: mixtures of collisions and ejections"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000349,"raw_usage":{"total_tokens":2010,"prompt_tokens":1153,"completion_tokens":857,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":769,"completion_tokens_details":{"reasoning_tokens":767}},"tokens_in":769,"tokens_out":857,"duration_ms":7533,"temperature":1.0,"reasoning_tokens":767,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:45:39.048693+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive observational check would be long-baseline radial-velocity monitoring of the 83 solitary warm Jupiters: the scattering model predicts that many of these systems actually carry close companions with velocity semi-amplitudes of 10 to 100 m/s, and the mixture model predicts a specific full shape for the eccentricity distribution. If the companions are not found and the low-eccentricity peak is wider or narrower than the half-Gaussian-plus-scattering-tail mixture predicts, the 60-percent fraction would need substantial revision. A complementary calculation would simulate the grazing giant-planet collisions that dominate these experiments to test whether mergers, rather than hit-and-run events, set the true branching ratios.","supporting_citations":[{"cited_title":"S., Lithwick Y., 2016, @doi [ ] 10.3847/0004-6256/152/6/174 , http://adsabs.harvard.edu/abs/2016AJ....152..174A 152, 174","cited_arxiv_id":null,"evidence_quote":"Characterized observed warm Jupiters with giant companions, providing the two-planet sample and the finding that high-eccentricity migration is inconsistent with it."}],"review_version":1}