{"id":"12fecf09-604b-4fb9-bb55-0bf51cc1229c","arxiv_id":"1908.02326","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In viscous steady state, gap-opening planets pile up disk gas exterior to their orbit, and the resulting torque yields a new Type-II migration rate that connects continuously to Type I.","lead":"This paper uses 2D simulations to show that a gap-opening planet in a low-mass gas disk piles up gas outside its orbit, acting as a leaky dam. The pileup sets a new Type-II migration rate that connects to the well-known Type-I rate, and should be visible in ALMA images.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Deep-gap Type-II migration rate rests on a power-law fit that excludes the deepest-gap point; including it may overturn the claimed mass-independence.","rationale":"The reader's weakest assumption was that ΔT is insensitive to the artificial inner boundary and wave-killing zone. The paper gives a plausible angular-momentum-conservation argument for this in §4.1, and the good agreement of the torque-balance checks (constant Mdot, Fν + Mdot l - Tdep ≈ 0) plus the similarity to Dürrmann & Kley non-migrating torques in Fig. 12 suggest that ΔT is not strongly boundary-dependent. The deeper vulnerability is the empirical fit. The abstract's new Type-II migration rate is derived solely from Eq. (32), and Eq. (32) is a fit that excludes an outlier. The excluded point is not an arbitrary outlier: it is the deepest gap simulated, produces the largest pileup, and is the most relevant to the observable regime. The paper uses this point to claim a pileup factor of ~10 while simultaneously withholding it from the fit that sets the migration rate; that is internally inconsistent. The quoted errors are statistical only and ignore this selection, as well as the omission of eccentric/non-converged runs. The inner-boundary issue, by contrast, would only affect ΔT if it changed the torque excitation region, and the paper's checks suggest it does not. I therefore see the fit as the most load-bearing concern, while agreeing with the reader's overall CONDITIONAL verdict.","tokens_in":33898,"tokens_out":13930,"duration_ms":143213,"concrete_test":"Refit Eq. (32) twice: once including the excluded K=1.07×10^4 simulation q1x3a3x4, and once using only K>100 simulations; in both cases recompute the implied exponents and the migration-rate scaling Eq. (38). If the q or α exponents move outside the quoted 1σ ranges, or if the migration rate's q/α dependence deviates from the claimed near-independence, the headline Type-II rate is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline deep-gap results, especially the new Type-II migration rate Eq. (38), rest on the empirical fit Eq. (32): ΔT/(Mdot l_p) = 4.36 q^1.05±0.06 α^-0.91±0.04. This fit is made to all simulations with q > 1e-4 except the deepest-gap point q1x3a3x4 (K = 1.07e4, pileup factor ~10), which is excluded as 'unusually large.' The claimed near-independence of the migration rate from planet mass and viscosity follows because the q exponent is close to 1 and the α exponent close to -1, yielding q^0.05 α^0.09 in Eq. (38). But the quoted statistical errors do not include the systematic effect of discarding the single most extreme and most observable point, nor the selection bias from omitting eccentric and non-converged simulations (§3.2). Including the excluded point would push the fit upward at high q and low α, likely steepening both exponents; if the q exponent exceeds 1.1, the migration rate becomes mass-dependent, contradicting the central claim. The fit also spans K<100 points, where theory requires ΔT ∝ q^2/α (Eq. 29), and deep-gap points, so a single power law has no theoretical basis. Since Eq. (38) is the primary new quantitative result, this is the most load-bearing weak point.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies planet-disk interaction in the low-disk-mass limit, where the planet's migration is slower than the disk's viscous evolution and the disk can be treated as passing through a sequence of viscous steady states (VSS). The authors introduce boundary conditions that allow 2D hydrodynamical simulations to reach VSS, and they show that the two-sided planet torque ΔT controls both the exterior pileup and the planet's migration rate. For moderately deep gaps they derive a first-principles prediction for ΔT (Eqs. 21-29) that agrees with their simulations at K ≲ 100. For deeper gaps they present an empirical power-law fit (Eq. 32) and use it to obtain a new Type-II migration rate (Eq. 38) that is nearly independent of planet mass and viscosity. They also show that previous 1D local-deposition models greatly overpredict the pileup, and they identify the difficulty of constructing a complete theory of very deep gaps.","tokens_in":1748,"tokens_out":2707,"duration_ms":89227,"significance":"If the central claims hold, the paper provides a clean organizing framework for planet-disk interaction in low-mass disks: a single quantity, ΔT, links the observable pileup, the gap profile, and the migration rate. The torque bookkeeping is carefully checked (Fig. 4), the moderate-gap prediction matches the simulations without fitted coefficients, and the resolution study gives 10% average (30% worst) agreement in ΔT. The paper also gives a clear explanation of why previous 1D local-deposition models fail and why earlier 2D simulations missed the pileup. These are substantial contributions. However, the headline deep-gap result, the new Type-II migration rate, rests on an empirical fit whose selection and regime boundaries are not robustly justified, so the quantitative deep-gap conclusions are not yet established at the same level as the moderate-gap results.","major_comments":[{"comment":"The empirical fit underlying the deep-gap migration rate is not restricted to the deep-gap regime. Eq. (32) is fitted to all runs with q > 10^-4, which includes many K < 100 points where the moderate-gap scaling ΔT ∝ q^2/α (Eq. 29) is known to hold and has a different dependence. The deepest-gap point q1x3a3x4 (K = 1.07e4, ΔT/(Ṁ l_p) = 11.0) is excluded as 'unusually large,' yet this is the most extreme and most observationally relevant point. At the quoted fit this point lies a factor of roughly 2 above the prediction, so its exclusion is consequential. Since Eq. (38) and the claim of near-independence from q and α follow directly from the fitted exponents q^1.05 and α^-0.91, the authors should refit using only K ≳ 100 points, report the fit with and without q1x3a3x4, and show how Eq. (38) changes. As written, the central deep-gap result is not robust to reasonable changes in the fitting procedure.","section":"Section 4.5 and Eq. (32); used in Section 5.1, Eq. (38)"},{"comment":"The assertion that the artificial inner wave-killing zone has 'negligible effect on the value of ΔT' is load-bearing for the deep-gap results, but it is not demonstrated by a domain-size or wave-killing-zone test. For the highest-K runs the gap extends past the inner boundary (Section 5.2 and Fig. 11), and the deepest-gap point q1x3a3x4 is also the point with the largest low-resolution deviation in ΔT (14.2 vs. 11.0 in Table 1). Because ΔT at high K is the small difference of larger one-sided torques, a test that moves ri,wkz inward (or enlarges the domain) is needed to show that the deepest-gap pileup and the fitted Eq. (32) are not biased by the inner boundary treatment.","section":"Section 4.1, Section 5.4, and Table 1"},{"comment":"The VSS convergence criterion is a 10% global consistency of Ṁ, and runs that become eccentric or fail to converge are omitted from the analysis. This selection could bias the fitted q-α scaling if convergence correlates with q or α, which is plausible given that the deepest-gap and lowest-α runs are the most expensive and the least converged. The authors should show that the excluded or longer-time runs do not change the fitted exponents in Eq. (32), or at least quantify how the fit depends on the adopted convergence threshold.","section":"Section 3.2 and Section 5.1"}],"minor_comments":[{"comment":"The displayed uncertainty on the coefficient in Eq. (32) is garbled ('4.36.6 2.8'); the statistical errors on the exponents are quoted, but the notation should be cleaned up and the covariance of the fitted parameters should be reported.","section":"Section 4.5, Eq. (32)"},{"comment":"The description of the inner and outer wave-killing zones would be clearer if the radii ri,wkz and ro,wkz were marked consistently on all panels of Figure 4; currently the reader must look back and forth between the text and the figure.","section":"Section 3.1 and Fig. 4"},{"comment":"The statement that the deep-gap migration rate is 'roughly independent of their mass and the disk's viscosity' is only as good as the fitted exponents in Eq. (32). A short sensitivity statement around Eq. (38), showing how the q^0.05 α^0.09 factors change under the alternative fits suggested above, would help the reader assess the strength of this claim.","section":"Section 5.1, Eq. (38)"},{"comment":"The paper is commendably explicit about the limitations of the deep-gap theory and about the inner-boundary issue. These statements should be retained in the final version; they correctly frame the empirical fit as a starting point rather than a complete theory.","section":"Section 7, Open Questions"}],"recommendation":"major_revision","confidential_remarks":"The moderate-gap theory and the VSS framework are solid and worth publishing. The main risk is the deep-gap empirical fit: the exclusion of the deepest point, the inclusion of moderate-gap points in the same power-law fit, and the absence of a domain-size test together make the headline Type-II rate (Eq. 38) insufficiently supported as it stands. These are fixable with additional fits and convergence tests, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the paper shows that the missing pileup in 2D hydro was largely a boundary-condition problem, and that once you let the disk settle into viscous steady state, the outer pileup appears naturally. Second, the moderate-gap theory — the two-sided torque from linear C± — is clean and checks out against simulation at K ≲ 100. That half of the paper is in good shape.\n\nThe deep-gap half is where I get cautious. The headline result, the Type-II migration rate in Eq. (38), is built on the empirical fit Eq. (32) to ΔT. The fit runs from K~3 to K~10^4 in a single power law, even though the moderate-gap theory says ΔT scales as q^2/α and the deep-gap points are where the exponent has to change. And the one point that would most constrain the deep end (q1x3a3x4, pileup ~10) is discarded as 'unusually large' with statistical errors only and no sensitivity to that exclusion. The stress-test concern is genuine: include that point and the q exponent could easily move past 1.1, which erodes the claimed mass-independence. Since Eq. (38) is the main quantitative novelty, this is the load-bearing flaw.\n\nThe paper is also a bit quick to discard the eccentric and non-converged simulations; that is standard, but it is a selection step with no error assigned. And the inner wave-killing truncates the inner gap at high K; the authors argue ΔT is insensitive, which is plausible for moderate gaps but less obviously true when the gap edge runs into the damping zone.\n\nCredit where it is due: the torque bookkeeping closes (Fig. 4e), Ṁ is constant to ~10%, the resolution check is minimal but real, and the comparison with Kanagawa et al. (2018) and Dürmann & Kley (2015) is a useful step toward understanding why earlier simulations missed the pileup and overestimated ΔT.\n\nAudience: anyone working on gap-opening planets, Type-I/II migration, or reading ALMA rings as pileups. Worth a serious referee. I would send it out, and in revision push hard on the deep-gap fit: refit with the excluded point, bootstrap the selection, and try to derive the deep-gap scaling rather than fit it.","headline":"Solid moderate-gap theory and a real pileup result, but the deep-gap Type-II migration rate rests on a fit with an excluded extreme point.","tokens_in":34772,"tokens_out":5747,"would_cite":true,"duration_ms":66171,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"In a low-mass disk, a gap-opening planet acts as a leaky dam: the same two-sided torque that creates the exterior gas pileup sets the planet's inward migration rate.","keywords":["planet-disk interactions","protoplanetary disks","accretion disks","planet migration","Type II migration","gap opening","surface density pileup","viscous steady state"],"falsifier":"Run a viscous-steady-state simulation with the inner boundary moved to a much smaller radius, or with the inner wave-killing zone removed, and check whether $\\Delta T/(\\dot M\\ell_p)$ and the exterior pileup level stay within the paper's quoted accuracy; if they shift significantly, the empirical fit and the derived migration rate are biased by the boundary treatment.","tokens_in":33646,"feed_emoji":"🪐","tokens_out":10496,"duration_ms":92947,"temperature":0.7,"pith_summary":"The paper studies planets in disks so low in mass that the planet migrates more slowly than gas flows through the disk, so the disk settles into a viscous steady state while overflowing the planet's orbit. It argues that one quantity, the total two-sided torque $\\Delta T$ that the planet exerts on the disk, controls both the pileup of gas outside the planet's orbit and the planet's inward migration rate. Using 2D hydrodynamic simulations with boundary conditions that let the disk reach that steady state, the paper measures $\\Delta T$, quantifies the pileup, and derives a new Type-II migration rate that connects continuously to the well-tested Type-I rate. This matters because directly imaged protoplanetary disks may show the pileup, offering a diagnostic of planet mass or disk viscosity.","feed_headline":"One torque controls pileup and migration of gap-opening planets","feed_subtitle":"In low-mass disks, the same torque that piles up gas also sets a new migration rate that merges with Type I.","key_machinery":"The load-bearing object is $\\Delta T$, the total two-sided torque the planet applies to the disk, defined as the integral of the excited torque density over all radii. The argument separates angular-momentum transport into wave excitation, wave propagation, and deposition: waves carry angular momentum from Lindblad resonances near the planet to where they damp, and the deposited torque density $t_{\\rm dep}$ shapes the surface-density profile through the viscous-steady-state equation $F_\\nu = \\dot M\\ell + \\int^r t_{\\rm dep}\\,dr'$. Far from the planet this yields $\\Sigma = \\Sigma_Z$ inside and $\\Sigma = \\Sigma_Z(1+\\Delta T/(\\dot M\\ell))$ outside, so $\\Delta T$ sets the pileup height, and the same $\\Delta T$ enters the migration formula. The numerical boundary conditions are chosen to match these steady-state solutions so the pileup can survive.","core_discovery":"In a low-mass disk, a gap-opening planet acts as a leaky dam: gas keeps accreting inward across the planet's orbit, but the angular momentum the planet deposits makes the exterior disk denser than a planet-free disk by the factor $1+\\Delta T/(\\dot M\\ell_p)$. The paper shows that the same two-sided torque $\\Delta T$ fixes the migration rate through $\\dot r_p/r_p = -2\\Delta T/(M_p\\ell_p)$, so in viscous steady state the planet is not locked to the disk's viscous accretion but migrates at a rate set by the torque balance. For moderately deep gaps the paper derives $\\Delta T$ from a gap-depth argument based on the torque cutoff, matching simulations; for deep gaps the simulations give an empirical scaling, roughly $\\Delta T/(\\dot M\\ell_p)\\sim 4(q/\\alpha)$, so the deep-gap migration rate becomes nearly independent of planet mass and viscosity. These results explain why earlier 2D simulations missed the pileup (incorrect boundary conditions) and why 1D local-deposition models overpredicted it dramatically.","pith_inferences":["Going beyond the paper: if an observed ring or cavity in a directly imaged disk is a pileup, comparing the brightness contrast at large radius with independent estimates of $\\dot M$ and $\\alpha$ could break the degeneracy between planet mass and viscosity that a single gap-depth measurement carries.","Going beyond the paper: the empirical deep-gap scaling $\\Delta T/(\\dot M\\ell)\\propto q/\\alpha$ suggests the pileup may grow only sub-linearly as $K$ increases, so the leaky-dam picture implies an upper bound on pileup that could be tested by pushing simulations beyond $K\\sim10^4$.","Going beyond the paper: a migration rate that is nearly independent of disk viscosity would mean giant-planet migration timescales in low-mass disks are no longer tied to the disk's viscous time, a shift that would change interpretations of observed exoplanet orbits."],"forward_implications":["Gas exterior to a gap-opening planet in a low-mass disk should sit above the planet-free profile by $1+\\Delta T/(\\dot M\\ell)$, giving a directly observable pileup that traces the planet-disk torque.","The deep-gap migration rate is nearly independent of planet mass and disk viscosity, scaling mainly with the disk-to-star mass ratio, in contrast to classical Type-II 'locked to the disk' migration.","The Type-I and Type-II migration rates join smoothly near $K\\sim100$, so one torque formula covers the transition from low-mass to massive planets in low-mass disks.","Published 2D simulations that held all fluid quantities at their initial values at the boundaries would not produce the pileup, while 1D local-deposition models overpredict it by factors like $e^{200}$ at $K\\sim10^4$."],"supporting_citations":[{"why":"supplies the standard torque formula $t_{\\rm ex}\\propto\\Sigma q^2/|r-r_p|^4$ that anchors the excitation side of the argument.","marker":"Goldreich & Tremaine (1980)"},{"why":"provides the asymmetric torque-density correction used to estimate one-sided torques in deep gaps and the classical Type-II framework.","marker":"Ward (1997)"},{"why":"gives the zero-angular-momentum-flux disk solution $\\Sigma_Z=\\dot M/(3\\pi\\nu)$ that defines the pileup baseline.","marker":"Lynden-Bell & Pringle (1974)"},{"why":"supplies the moderately deep gap-depth scaling $\\Sigma_p/\\Sigma_{Z,p}\\approx 1/(1+0.04K)$ that the paper extends to a two-sided torque prediction.","marker":"Kanagawa et al. (2015b)"},{"why":"provides the linear Type-I torque coefficients and migration rate to which the new rate is shown to connect continuously.","marker":"Tanaka et al. (2002)"},{"why":"prior simulations whose torques and migrating-planet runs are compared to show that non-steady-state results differ from the VSS torque.","marker":"Dürmann & Kley (2015)"},{"why":"prior empirical $\\Delta T$ and migration-rate relation that the paper argues is not in viscous steady state.","marker":"Kanagawa et al. (2018)"}],"fun_headline_variants":["Leaky dam planets pile up gas and set migration speed","Same torque drives disk pileup and planet migration","Low-mass disk planets: pileup and migration tied to one torque","Planets as leaky dams: pileup reveals migration rate","Pileup torque sets new migration law for low-mass disks"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the measured two-sided torque $\\Delta T$ is set by wave excitation close to the planet, so the artificial inner boundary and angular-momentum-conserving wave-killing zone that distort the inner disk profile do not change the pileup or migration rate.","fun_headline_variants_meta":{"raw":{"variants":["Leaky dam planets pile up gas and set migration speed","Same torque drives disk pileup and planet migration","Low-mass disk planets: pileup and migration tied to one torque","Planets as leaky dams: pileup reveals migration rate","Pileup torque sets new migration law for low-mass disks"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000428,"raw_usage":{"total_tokens":2236,"prompt_tokens":1036,"completion_tokens":1200,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":652,"completion_tokens_details":{"reasoning_tokens":1117}},"tokens_in":652,"tokens_out":1200,"duration_ms":9100,"temperature":1.0,"reasoning_tokens":1117,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:48:13.115179+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a viscous-steady-state simulation with the inner boundary moved to a much smaller radius, or with the inner wave-killing zone removed, and check whether $\\Delta T/(\\dot M\\ell_p)$ and the exterior pileup level stay within the paper's quoted accuracy; if they shift significantly, the empirical fit and the derived migration rate are biased by the boundary treatment.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the asymmetric torque-density correction used to estimate one-sided torques in deep gaps and the classical Type-II framework."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the linear Type-I torque coefficients and migration rate to which the new rate is shown to connect continuously."},{"cited_title":"D., Tanaka, H., & Szuszkiewicz, E","cited_arxiv_id":null,"evidence_quote":"prior empirical $\\Delta T$ and migration-rate relation that the paper argues is not in viscous steady state."}],"review_version":1}