{"id":"84548229-0955-42eb-8a53-99d64ba3075d","arxiv_id":"2510.21914","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Old and dark-accreted dark matter fits a Maxwell-Boltzmann shape, while recent massive merger debris can be traced by boosted stellar kinematics, yielding a Milky Way speed distribution whose high-speed tail is ~20% suppressed relative to the standard halo model.","lead":"This paper builds a recipe for estimating the local dark matter speed distribution from the motions of nearby stars, and applies it to the Milky Way using Gaia data. If the recipe transfers from simulations to our Galaxy, direct-detection experiments get a more realistic input for the dark matter speeds they expect.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"MW result depends on TNG50-calibrated boost and traceable fraction that FIRE-2 comparisons suggest are not universal; no external validation supports transfer.","rationale":"The strongest claim is that the empirical reconstruction works in TNG50 and, when applied to the Milky Way, yields a specific speed distribution with a 11 km/s slower mode and ~20% tail suppression. For that claim to hold for the MW, the TNG50-calibrated parameters (Δσ and w_tr) must transfer to the real Galaxy. The paper itself provides direct evidence that they may not: FIRE-2 halos show a tighter correlation and smaller offsets, and Appendix B explains why TNG50's assembly history inflates Δσ. The reader's weakest assumption identifies exactly this transferability problem, and I agree. The concern is not that the TNG50 validation is internally wrong; it is that the MW headline numbers inherit an unvalidated extrapolation. A secondary, reinforcing issue is that the TNG50 validation is in-sample—Δσ is the average over the same mergers used to evaluate the EMD—so a split-sample test would strengthen the methodology. The proposed FIRE-2 recalibration directly tests whether the MW conclusions survive a different, physically motivated simulation calibration; the leave-one-out test would at least quantify in-sample optimism independent of code choice. Given the high quality of the paper and its transparent discussion of the FIRE-2 discrepancy, the conditional verdict is appropriate and should be retained until such a test is performed.","tokens_in":28131,"tokens_out":4128,"duration_ms":44283,"concrete_test":"Recompute the MW reconstruction in Section 4 using Δσ and w_tr measured from GSE-like mergers in FIRE-2 (or the Zhang et al. in prep. sample) instead of TNG50, keeping the same Gaia GSE stars. If the resulting mode shift and 95th-percentile tail suppression move outside the quoted 16–84% bands (e.g., mode shift <6 km/s or >16 km/s, tail suppression <10% or >30%), the TNG50 calibration is not transferable. As an internal check, perform leave-one-out calibration on the 26 GSE-like TNG50 mergers: calibrate Δσ on 25 and test on the held-out merger; compare the EMDs to the in-sample 11^{+6}_{-4} km/s distribution.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The MW reconstruction in Section 4 uses two simulation-calibrated inputs: the dispersion boost Δσ (Eqs. 5–6) and the traceable fraction w_tr, both taken from TNG50. The paper's own Section 5 states that FIRE-2 (Necib et al. 2019b; Zhang et al. in prep.) finds a tighter stellar–DM correlation and smaller offsets, and Appendix B attributes the TNG50 offset to early host assembly and deeper potentials. Since the MW is relatively isolated and may have assembled later than typical TNG50 analogues, the GSE-specific Δσ = 43^{+11}_{-10} km/s could overestimate the true GSE DM boost. The GSE stellar mode is ~90 km/s and rises to ~140 km/s after boosting; a 15–20 km/s overestimate in Δσ would change both the quoted 11 km/s mode shift and the ~20% high-speed-tail suppression. Similarly, w_tr is taken from single-Traceable-merger TNG50 halos (18^{+15}_{-5}%) and may not transfer to the MW's specific accretion history. This is not an internal inconsistency, but it means the headline MW numbers are extrapolations from one simulation code without external validation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an empirical procedure for reconstructing the local dark-matter speed distribution in the solar neighborhood. Using 98 Milky Way analogues from TNG50, it separates the DM into three components: 'Old Untraceable' DM accreted before z=3, 'Young Untraceable' DM from later diffuse/low-mass accretion, and 'Traceable' DM from massive recent mergers. The first two are modeled jointly by a Maxwell–Boltzmann (SHM) distribution; the Traceable component is modeled from the stellar debris of the same merger after applying a velocity-dispersion boost. The full reconstruction is the weighted sum of these pieces (Eq. 7). The method is validated in TNG50 with Earth Mover's Distance metrics, showing median EMDs of 10–11 km/s against the exact simulated distributions. The paper then applies the procedure to the Milky Way using Gaia GSE stellar tracers, finding that the GSE contribution shifts the mode of the local speed distribution by 11 km/s and suppresses the high-speed tail by about 20%.","tokens_in":28451,"tokens_out":3529,"duration_ms":38606,"significance":"If the method holds up, this is a valuable step toward an observationally grounded local DM speed distribution for direct-detection analyses. The paper's strengths are its large, homogeneous sample of 98 MW analogues; the explicit treatment of dark accretion as distinct from luminous mergers; the systematic parameter-robustness tests in Appendices A–C; and the public release of the inferred MW speed distributions. The three-way decomposition is physically well motivated, and the finding that the untraceable background is Maxwellian even when young dark-accretion is included is a useful result. The application to the GSE is timely and connects to active literature. However, the validation is in-sample, the improvement over the SHM is modest in EMD terms (11 vs 14 km/s), and the Milky Way application relies on TNG50-calibrated parameters that may not transfer to the MW's specific assembly history.","major_comments":[{"comment":"The central validation is in-sample: the distributions for Δσ and w_tr used in Eq. (7) are computed from the same 98 TNG50 halos against which the reconstruction is tested in Fig. 7. The EMDs of 10–11 km/s therefore measure self-consistency of the calibration, not out-of-sample predictive skill. Since the paper advertises a reconstruction procedure, I ask for an explicit out-of-sample test: e.g., calibrate on half the analogues and apply to the other half, or calibrate Δσ on the GSE-like mergers and apply it to non-GSE mergers (and vice versa). Appendix A tests tagging parameters but does not address this circularity.","section":"§3.3, Eq. (7), Fig. 7"},{"comment":"The Milky Way result — the 11 km/s mode shift and the ~20% high-speed-tail suppression — is controlled by Δσ = 43^{+11}_{-10} km/s and w_tr = 18^{+15}_{-5}%, both taken from TNG50. Section 5 explicitly states that FIRE-2 finds a tighter stellar–DM correlation and smaller offsets, and Appendix B attributes the TNG50 offset to earlier assembly and deeper potentials. Since 81% of the analogues are within 16 Mpc of a Virgo-mass cluster (Sec. 2.1) while the MW is not, the TNG50-calibrated boost is an extrapolation. The shaded bands in Fig. 8 reflect only internal TNG50 scatter, not this code/systematic uncertainty. Please quantify the MW speed distribution for a range of Δσ and w_tr spanning the FIRE-2 expectations (including Δσ = 0), and report the resulting spread in the mode shift, 95th percentile, and tail suppression.","section":"§4, Fig. 8"},{"comment":"The EMD is the only validation metric used in the TNG50 tests, and the paper itself notes that EMD is 'not very sensitive to changes at the high-speed tail' (Sec. 3). However, the abstract and conclusions highlight the high-speed tail suppression as a key consequence. The reported median EMD improvement over the SHM (11 vs 14 km/s) is modest and does not directly test whether the reconstruction captures the tail. I recommend adding a tail-sensitive statistic (e.g., the 95th percentile speed, the fraction of DM above 400 or 500 km/s, or a rate-weighted integral) to the TNG50 validation in Fig. 7, and reporting it alongside the EMD.","section":"§3, EMD metric; Fig. 7"}],"minor_comments":[{"comment":"The classification into Old/Young/Traceable depends on the z_acc = 3 threshold and the 2 Gyr/70% tagging choices; Appendix A notes that w_tr varies with these choices. Please add a sentence in the main text reminding the reader that w_tr is defined relative to this operational classification.","section":"§2.2 / Appendix A"},{"comment":"The sentence 'The SHM alone (not shown here)' is confusing because the caption and the figure appear to include black SHM curves in other panels; specify that the left panel omits the SHM-only curve for clarity.","section":"Fig. 7, left panel"},{"comment":"The conclusion quotes a 'median speed of 180 km/s' for the GSE stars, while Section 4 quotes a mode of 90 km/s. These are different statistics; please use consistent terminology to avoid apparent contradiction.","section":"§5"},{"comment":"The notation v^i_b and Δσ could be defined more explicitly: state that Δσ is a single scalar applied equally to all three spherical components, and that the directionally averaged definition in Eq. (6) is used.","section":"Eq. (5)"}],"recommendation":"major_revision","confidential_remarks":"This is a solid and honest paper. The main issue is not internal inconsistency but a load-bearing extrapolation from TNG50 to the MW, combined with in-sample validation. The requested out-of-sample and systematic-sensitivity checks are feasible within the manuscript's scope and would substantially strengthen the claims. I would be happy to see a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on arXiv:2510.21914. It's a careful, well-written method paper: the authors lay out a recipe for building the local DM speed distribution from stellar tracers, and the genuinely new piece is treating dark accretion explicitly and testing on 98 TNG50 analogues rather than two FIRE halos. The three-way split (Old Untraceable, Young Untraceable, Traceable) and the one-parameter dispersion boost are sensible, and the validation in TNG50 is transparent — Appendices A–C test the merger-tagging systematics and compare the GSE star selection to the literature. I trust the simulation-side result: boosted stars do trace the Traceable DM well, and the Untraceable component is close enough to a Maxwellian in most halos.\n\nThe soft spots are exactly where the stress-test lands. The reconstruction uses Δσ and w_tr calibrated on TNG50 and then validates on the same TNG50 halos; the improvement over SHM is modest (median EMD 11 vs 14 km/s). The Milky Way application then leans on the TNG50-derived boost, Δσ = 43 ± 11 km/s for GSE-like mergers, and a w_tr of 18%. The paper's own Section 5 and Appendix B acknowledge that FIRE-2 gives tighter stellar-DM correlations and smaller offsets, and they attribute this to earlier assembly and deeper potentials in TNG50. So the MW numbers are extrapolations from one simulation code, and the quoted 11 km/s mode shift and ~20% tail suppression inherit that uncertainty. The stress-test concern — that the GSE boost may be overestimated — holds up, though the paper is open about it. The environment mismatch (TNG50 has Virgo-like clusters near many analogues) reinforces the worry that the MW's quiescent, later assembly is not well represented.\n\nNone of this kills the paper. The method is a useful framework, and the code and inferred distributions are public. But I'd want to see a split-sample test or a cross-simulation calibration before adopting the MW speed distribution as a standard input. The empirical distribution is a step forward; the uncertainty on the correction factor should be reported more prominently.\n\nWho is this for? Direct-detection phenomenologists who care about the tail shape, and anyone building empirical DM models from Gaia-like data. It deserves a serious referee — the structure is sound and the limitations are honestly stated. I'd send it to review with a request for a more explicit statement about the transferability of Δσ and w_tr, and ideally a test leaving out some of the TNG50 sample.","headline":"Solid, honest method paper for reconstructing the local DM speed distribution from stellar kinematics; the Milky Way application is a TNG50-calibrated extrapolation that deserves review but needs external validation.","tokens_in":28945,"tokens_out":2556,"would_cite":true,"duration_ms":24424,"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":"Local dark-matter speeds can be reconstructed empirically from observed stars: a Maxwellian background plus dispersion-boosted merger debris matches simulated galaxies and shifts the Milky Way's peak 11 km/s slower.","keywords":["dark matter speed distribution","direct detection","stellar kinematics","galactic mergers","solar neighborhood","Maxwell-Boltzmann distribution","local standard of rest"],"falsifier":"Compute the same stellar–dark-matter velocity-dispersion offset for recent massive mergers in high-resolution Milky Way-mass simulations that use different baryonic feedback and a different halo-tracking algorithm; if the offset is not near 30 km/s, or if the Earth Mover's Distance between boosted stars and dark matter exceeds ~20 km/s, the one-parameter boost is not portable. A second check is to redo the Milky Way reconstruction with alternative selections of the last major merger's stars; the 11 km/s peak shift should not move by more than a few km/s.","tokens_in":28049,"feed_emoji":"🌌","tokens_out":9283,"duration_ms":86736,"temperature":0.7,"pith_summary":"The paper claims that the dark matter speed distribution near the Sun can be built empirically from stars: early-accreted and dark-accreted dark matter follows a Maxwell–Boltzmann distribution, while dark matter from the last few massive mergers is traced by the stellar debris of those mergers after a one-parameter velocity-dispersion boost. Using 98 simulated Milky Way-like galaxies, the authors show the two-component reconstruction matches the true simulated speed distribution to roughly 10 km/s in Earth Mover's Distance, even when the boost and the traceable fraction are uncertain. Applied to the Milky Way's last major merger, the recipe puts the peak of the local speed distribution 11 km/s below the standard Maxwellian and suppresses the high-speed tail by about 20%. This matters because direct dark-matter detection rates and inferred cross sections depend directly on this speed distribution.","feed_headline":"Last big merger slows local dark matter by 11 km/s","feed_subtitle":"Observed merger debris plus a Maxwellian background rebuilds local dark-matter speeds, trimming ~20% off the tail.","key_machinery":"The load-bearing construction is the two-component formula f_tot(v) = (1 - w_tr) SHM(v|v0) + w_tr Σ (m_*/M_*) f_b(v). The first term is a Maxwell–Boltzmann 'Standard Halo Model' with v0 set by the circular speed at 8 kpc, covering old and dark-accreted dark matter. The second term is a kernel-density estimate built from the observed velocities of stars from each massive merger, after each velocity component is shifted by the boost v_i^b = (Δσ + σ_i^★)/σ_i^★ (v_i^★ − ⟨v_i^★⟩) + ⟨v_i^★⟩, with Δσ ≈ 30 km/s the mean difference between dark-matter and stellar velocity dispersions. The construction is validated with the Earth Mover's Distance between the true and reconstructed speed distributions,","core_discovery":"The central discovery is that the total local dark-matter speed distribution splits cleanly into two pieces. The 'Untraceable' piece—old accreted dark matter plus recent dark accretion—is Maxwell–Boltzmann with a scale set by the mass enclosed at the solar radius. The 'Traceable' piece from recent massive mergers can be modeled from the observed velocities of the merger's stars, provided the stellar velocity dispersion is boosted by an empirical factor of about 30 km/s. The boost works because stars are stripped later and land deeper in the potential well, while dark matter is stripped earlier and retains higher orbital speeds. The authors verify the reconstruction on 98 simulated Milky Way-","pith_inferences":["If the offset between dark-matter and stellar velocity dispersions depends on galaxy assembly history or baryonic feedback, the ~30 km/s boost is best treated as a prior from this simulation set; a multi-simulation calibration would tell whether the Milky Way value is stable.","The same two-component strategy could be extended beyond the solar neighborhood once future wide-field spectroscopic surveys provide clean samples of accreted stars for more merger events.","Because the stellar debris from massive mergers carries nonzero azimuthal velocity, the full dark-matter velocity distribution is likely not isotropic even where the speed distribution looks Maxwellian—an opportunity for directional detectors.","The predicted ~20% suppression of the high-speed tail is the part of the distribution that high-recoil-energy searches probe; experiments with different thresholds should see opposite-signed rate shifts if this reconstruction is right."],"forward_implications":["Direct detection experiments can replace the pure Maxwellian assumption with this empirical mixture, changing predicted recoil spectra near threshold and at high recoil energy.","The Milky Way's last major merger shifts the local speed-distribution mode 11 km/s lower and suppresses the fastest portion of the tail by about 20%, altering sensitivity projections for low- and high-mass dark matter.","Dark matter that cannot be traced by stars—early accretion and dark accretion—can safely be left as a Maxwellian, removing a major uncertainty in empirical models of the local dark matter.","The reconstructed Milky Way speed distributions are released for public use, so detector analyses can adopt them without re-running galaxy formation simulations."],"fun_headline_variants":["Stellar debris reveals dark matter speed boost","Gaia stars decode local dark matter speeds","Merger stars map dark matter velocities near Sun","Ancient crash slows dark matter near Sun"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the velocity-dispersion boost and the traceable dark-matter fraction measured in simulated Milky Way analogues transfer to our own Galaxy; if the stellar-to-dark-matter offset in the Milky Way differs, the reconstructed peak shift and tail suppression change.","fun_headline_variants_meta":{"raw":{"variants":["Stellar debris reveals dark matter speed boost","Gaia stars decode local dark matter speeds","Merger stars map dark matter velocities near Sun","Ancient crash slows dark matter near Sun"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000294,"raw_usage":{"total_tokens":1554,"prompt_tokens":760,"completion_tokens":794,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":504,"completion_tokens_details":{"reasoning_tokens":739}},"tokens_in":504,"tokens_out":794,"duration_ms":8534,"temperature":1.0,"reasoning_tokens":739,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T08:11:08.218457+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the same stellar–dark-matter velocity-dispersion offset for recent massive mergers in high-resolution Milky Way-mass simulations that use different baryonic feedback and a different halo-tracking algorithm; if the offset is not near 30 km/s, or if the Earth Mover's Distance between boosted stars and dark matter exceeds ~20 km/s, the one-parameter boost is not portable. A second check is to redo the Milky Way reconstruction with alternative selections of the last major merger's stars; the 11 km/s peak shift should not move by more than a few km/s.","supporting_citations":[],"review_version":1}