{"id":"65b2b4b0-9fd6-493d-acf1-c0c271ebe599","arxiv_id":"2412.15864","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Supercooling terminated by curvature flipping still proceeds by bubble nucleation and expansion, not by smooth phase mixing, according to 3D lattice simulations.","lead":"This paper asks whether a supercooled universe, whose phase transition is forced to end when the potential curvature flips sign at temperature T2, finishes by nucleating bubbles or by smoothly rolling to the true vacuum. Lattice simulations with barriers smaller than the temperature show bubbles form and expand, so the standard first-order transition picture and its gravitational wave predictions survive.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The generic bubble-nucleation conclusion is under-validated: it relies on visual identification of bubbles in three toy-potential runs, with no extracted lattice nucleation rate and no simulation of the actual flaton potential.","rationale":"The reader's verdict is CONDITIONAL, and my stress-test does not move it. The simulation results are internally consistent: with a barrier, localized true-vacuum regions appear; without a barrier, the field rolls down homogeneously. This is strong qualitative evidence for bubble-mediated transitions. However, the central claim is framed as generic, and the quantitative link to critical-bubble nucleation is not established. The authors themselves acknowledge that no lattice nucleation rate is extracted and that the representative potential is not the actual flaton potential. My concrete check targets exactly that gap. I do not think this requires rejection: the paper is honest about its limitations, and the qualitative conclusion is plausible. But the conditional verdict remains appropriate because the missing rate comparison and the untested mapping to the SUSY potential are real, addressable weaknesses rather than fatal flaws. The percolation threshold lambda > 0.05 in Section 4.3 is an additional approximate statement; my own check of the integral I(t2) suggests the threshold is not obviously wrong, but it is derived from a fit at only two coupling values and would also benefit from direct evaluation on a grid of lambda. This reinforces, rather than replaces, the need for quantitative validation of the simulation claim.","tokens_in":25445,"tokens_out":19953,"duration_ms":191555,"concrete_test":"Using the same Langevin lattice code, repeat benchmark A on a volume large enough to contain several bubbles, run at least 20 independent seeds, and extract Gamma_lattice by measuring the time of first bubble appearance per unit volume. Compare Gamma_lattice with the analytic Gamma from Eq. (2.1), evaluated with BubbleDet for the same potential Eq. (3.1). If Gamma_lattice agrees with Gamma_analytic to within an order of magnitude and the subcritical/supercritical field profiles match the bounce solution phi_3(r), the bubble-nucleation interpretation is validated; if not, the central claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's Section 5 central claim goes beyond what the simulations can currently verify. Section 3.2 explicitly states that extracting the lattice nucleation rate Gamma_lattice is 'beyond the scope of the current work', and no code or raw data are released. The evidence for bubble nucleation in benchmarks A-C is therefore the visual appearance of localized expanding regions in Figs. 6-8, together with the contrast with the barrier-less benchmark M. That contrast is suggestive, but it does not by itself prove that the localized regions are the critical-bubble configurations of Eq. (2.3), as opposed to rare large thermal fluctuations that locally cross a shallow barrier and then grow because the potential is almost flat. The mapping from Eq. (3.1) to realistic flaton potentials is also an assumption (footnote 1 and Sec. 3.1), justified only by the hierarchy checks |V''(phi_c)| < (2 pi T)^2 and phi_cutoff/phi_c > 1. These checks ensure heavy Matsubara modes are not active at the critical bubble, but they do not test whether the actual SUSY potential of Sec. 4.1, with its field-dependent damping and non-white noise structure, produces the same channel. Since the conclusion is phrased as 'generically', the missing quantitative validation is load-bearing.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the end of strongly supercooled phase transitions in models where the curvature of the finite-temperature effective potential at the origin flips sign at a temperature T2, as in m^2-type flaton potentials. The authors argue that the two competing exit channels are thermal bubble nucleation and a phase-mixing-like tachyonic roll-down, and they simulate a 3D Langevin dynamics on the lattice for a representative polynomial potential (3.1) with a sub-thermal barrier width (phi_b<T). From visual inspection of the field snapshots for benchmarks A, B, C (Table 1), they conclude that localized true-vacuum regions nucleate and expand, in contrast to the barrier-less benchmark M where a homogeneous roll-down occurs. They then analyze a SUSY toy model (Secs 4.1-4.2), compute the bounce action S3/T near T2 (Sec 4.3), find percolation before T2 for lambda>0.05, and estimate the gravitational wave spectrum (Sec 4.4). The paper concludes that such transitions generically proceed via critical-bubble nucleation and that the standard first-order transition description remains applicable.","tokens_in":25716,"tokens_out":14208,"duration_ms":120395,"significance":"If correct, the paper resolves an important open question: it supports the use of the standard O(3) bounce formalism and the associated gravitational-wave predictions for m^2-type flaton potentials, even when the barrier width in field space is below the thermal scale. The lattice simulations are a useful first step and include a suggestive control benchmark without a barrier. The paper also provides a concrete SUSY model, a quantitative estimate of the percolation threshold, and an explicit GW spectrum, making the scenario falsifiable. However, the central 'generically' claim rests on qualitative visual evidence from three toy-potential runs with no extracted nucleation rate, and the mapping to realistic flaton potentials is assumed rather than demonstrated. The strengths are the clear physical question, the balanced treatment of the competing channels, and the explicit admission of the limitations in Sec 3.2.","major_comments":[{"comment":"The central conclusion that the phase transition 'generically proceeds via the nucleation and expansion of critical bubbles' is based on visual identification of localized regions in benchmarks A-C (Figs 6-8) and the contrast with the barrier-less benchmark M (Fig 9). The paper explicitly states in Sec 3.2 that extracting the lattice nucleation rate Gamma_lattice is 'beyond the scope of the current work', and no comparison of the observed field profiles to the O(3) bounce solution of Eq. (2.3) is made. Consequently, the localized regions could in principle be rare large thermal fluctuations that cross the shallow barrier and then expand in a nearly flat potential; they are not demonstrated to be the critical-bubble saddle-point configurations. Because the paper's main claim is about the nature of the transition, this missing quantitative diagnostic is load-bearing. Please either provide such a diagnostic (e.g., comparing a nucleating profile to Eq. (2.3), or extracting Gamma_lattice and comparing with Eq. (2.1)) or restrict the conclusion to what the snapshots show and soften the words 'generically' and 'critical bubbles' accordingly.","section":"Secs. 3.2, 3.3, and 5"},{"comment":"The simulated potential (Eq. (3.1)) is a representative toy potential, not the SUSY flaton potential of Sec 4.1. The checks in Table 1 (|V''(phi_c)|/(2pi T)<1 and phi_cutoff/phi_c>1) validate the 3D EFT for the toy potential at the critical bubble, but they do not test the effects of field-dependent damping eta(phi), multiplicative noise, or the non-white noise structure that would be present in the realistic model. Since the abstract and conclusion use the word 'generically', the transfer of the toy-potential result to the class of m^2-type flat potentials is an assumption. Please either simulate a closer-to-realistic potential (e.g., using the daisy-resummed potential of Sec 4.2 for a benchmark with moderate lambda) or explicitly restrict the claim to the toy class and discuss the possible sensitivity to the neglected effects.","section":"Sec. 3.1 and footnote 1"},{"comment":"The toy potential (3.1) is Z2-symmetric with two degenerate minima, leading to simultaneous red and blue bubbles separated by domain walls in Figs 6-8. The realistic flaton potential of Sec 4.1 has a single true minimum. The presence of domain walls may affect the expansion and collision dynamics and complicates the interpretation of the snapshots as isolated true-vacuum bubbles. Please either simulate a potential with a single global minimum (e.g., by adding a small cubic term) or provide an argument that the Z2 symmetry and domain walls do not alter the qualitative conclusion about bubble nucleation vs. phase mixing.","section":"Secs. 3.1 and 3.3"}],"minor_comments":[{"comment":"The text contains a typo: 'modeled through the the Langevin equation' should read 'modeled through the Langevin equation'.","section":"Sec. 3.1"},{"comment":"The color scheme for benchmark M is changed (white at the origin, black around |phi|~T, red/blue near the minima) without noting that the color map differs from Figs 6-8; this makes direct visual comparison across figures more difficult.","section":"Sec. 3.3, Fig. 9"},{"comment":"The entry for benchmark M lists lambda = -0.011, while lambda is otherwise the quartic coupling in Eq. (3.1); the sign convention for this entry should be clarified in the text.","section":"Table 1"},{"comment":"The approximation Tn/T2 - 1 ~ lambda^6 drops the numerical coefficient from the logarithm in Eq. (4.26); a brief estimate of this coefficient for the benchmark parameters would improve reproducibility of the percolation claim.","section":"Sec. 4.3, Eq. (4.27)"},{"comment":"The paper would benefit from a data or code availability statement, as the quantitative claims depend on simulation snapshots that are only displayed qualitatively.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely correct in its qualitative conclusion, but the abstract and conclusion overstate the evidence. The authors should either substantially soften the 'generically' and 'critical bubbles' wording or add a quantitative validation (e.g., a profile comparison or a lattice nucleation-rate extraction for at least one benchmark). The Z2-symmetric toy potential and the assumed transferability to the SUSY potential are the two main modeling gaps. A revision along these lines would, in my view, make the paper suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the real news is the first direct lattice test of whether curvature-flip termination nucleates bubbles or phase-mixes, and it lands on bubble nucleation for any barrier. The barrier-less benchmark M producing homogeneous roll-down is the right control. That is a genuine step beyond the argument in Ref. [60].\n\nWhat is done well: the simulations use a dimensionally reduced Langevin equation with lattice counterterms, they check the hierarchy |V''(phi_c)| < (2 pi T)^2 and phi_cutoff/phi_c > 1 for each benchmark, and they are explicit that extracting the lattice nucleation rate is beyond scope. The analytic bounce fit near T2 is coherent with the simulation timescales. The colour snapshots are honestly qualitative. I think the qualitative conclusion is credible.\n\nThe soft spots: the word 'generically' is doing more work than the evidence. Three benchmark potentials, two with non-perturbative couplings (lambda = 10, epsilon = 10 T^2), and no quantitative extraction of the bubble nucleation rate. The visual identification of 'bubbles' cannot by itself rule out that the localized expanding regions are rare thermal fluctuations that crossed a shallow barrier—though in a thermal field theory that is functionally what bubble nucleation is, so this concern is milder than the stress-test note suggests. The bigger gap is the representative-potential mapping to realistic flaton potentials: field-dependent damping and non-white noise could in principle change the channel, and the hierarchy checks only validate the equilibrium effective potential at the critical bubble, not the full Langevin dynamics.\n\nThe authors acknowledge these limitations openly. The claim would be hardened by a wider parameter scan, a fitted Gamma_lattice compared to the bounce prediction, and ideally a simulation of a potential closer to the SUSY model. But as a first test, it serves its purpose.\n\nWho it is for: phase transition phenomenologists, especially those working on supercooled models, gravitational waves from phase transitions, and anyone relying on the standard bubble description in m^2-type flat directions. It deserves a serious referee. I would recommend conditional acceptance: the qualitative result is significant and likely correct, but the 'generically' claim needs to be either substantiated with more benchmarks and rate extraction or toned down to 'for the simulated potentials'.","headline":"First lattice test of curvature-flip termination: bubbles win over phase mixing for any barrier, but 'generically' outruns the three benchmarks.","tokens_in":26244,"tokens_out":2550,"would_cite":true,"duration_ms":25061,"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":"When supercooling is terminated by curvature flipping of the effective potential, the phase transition proceeds by bubble nucleation rather than by smooth phase mixing.","keywords":["supercooled phase transition","bubble nucleation","phase mixing","curvature flipping","Langevin dynamics","lattice simulation","gravitational waves","flaton"],"falsifier":"Simulate the actual finite-temperature effective potential of the toy SUSY model, including the full one-loop and daisy-resummed terms, on the same lattice and watch a spatial slice: if the field variance grows to the true vacuum without forming distinct round true-vacuum regions, or if a benchmark with $\\lambda<0.05$ completes by homogeneous roll-down before bubbles percolate, the bubble-nucleation conclusion would fail. A cheaper check is to compute the tachyonic growth time from Eq. (2.32) using the real $V''$ just below $T_2$ and compare it with the inverse bubble nucleation rate per Hubble volume.","tokens_in":25206,"feed_emoji":"🫧","tokens_out":8025,"duration_ms":65114,"temperature":0.7,"pith_summary":"Supercooled first-order phase transitions can be forced to end because the curvature of the finite-temperature effective potential at the origin flips sign at a temperature $T_2$, quickly erasing the potential barrier. This paper asks whether such a transition proceeds by nucleating and expanding true-vacuum bubbles or by a smooth, phase-mixing-like roll of the field. The answer, from lattice simulations of a representative flat potential with barrier width $\\phi_b<T$, is bubble nucleation: even a very small barrier localizes the field long enough for critical bubbles to form and collide. If true, the standard first-order transition description, including the bounce action $S_3/T$ and gravitational-wave emission from colliding bubbles, remains valid for $m^2$-type flat potentials, and the models remain testable in future detectors.","feed_headline":"Expect bubbles, not phase mixing, when supercooling ends","feed_subtitle":"Lattice simulations show a sub-temperature barrier traps the field long enough for true-vacuum bubbles to nucleate and collide.","key_machinery":"The load-bearing objects are the classical Langevin equation for the zero Matsubara mode, $\\partial_t^2\\phi+\\eta\\partial_t\\phi-\\nabla^2\\phi+\\partial_\\phi V_T(\\phi)=\\xi$, with white noise $\\langle\\xi\\xi\\rangle=2T\\eta\\,\\delta\\delta$, and the representative finite-temperature potential $V_T(\\phi)=\\frac{1}{2}m_T^2\\phi^2-\\frac{\\lambda_T}{4!}\\phi^4+\\frac{\\epsilon}{6!}\\phi^6$, with $m_T^2=\\lambda_T\\phi_b^2/12$ so that the barrier width $\\phi_b$ is an input. The escape rate is controlled by the O(3) bounce action $S_3/T$, the exponential cost of forming a critical thermal bubble, whose small-barrier form $S_3/T\\simeq 19\\sqrt{3}\\,\\phi_b/\\sqrt{\\lambda_T}$ sets the benchmark parameters. The machinery works by showing that the metastable origin suppresses the tachyonic instability even when thermal fluctuations ($\\langle\\phi^2\\rangle\\sim T^2$) exceed the barrier size, so the field waits for rare local fluctuations rather than rolling globally.","core_discovery":"On the paper's own terms, the central claim is that when supercooling is terminated by curvature flipping of the effective potential, the phase transition generically proceeds via the nucleation and expansion of critical bubbles rather than via a phase-mixing-like transition. The lattice simulations compare three barrier benchmarks and one barrier-less Mexican hat; the barrier-less case rolls down homogeneously, while all barrier cases show localized true-vacuum regions that expand and collide, even though $\\phi_b$ is below the thermal scale. For the toy SUSY-inspired model, percolation completes before the temperature reaches $T_2$ whenever the quartic coupling $\\lambda$ is above about $0.05$. Consequently the usual first-order phase transition description, including the O(3) bounce action and the resulting gravitational-wave spectrum, continues to apply in this regime, and the bubble wall does not run away because the vacuum-energy difference is too small.","pith_inferences":["The paper simulates a representative $Z_2$ polynomial rather than the actual SUSY flaton potential, so a natural extension is to rerun the same benchmarks with the full daisy-resummed potential and field-dependent damping; the bubble conclusion is likely robust but should be verified quantitatively.","Curvature flipping is one instance of a more general criterion: any phase transition whose barrier shrinks to sub-$T$ width while staying flat enough may still be first-order, because the metastable minimum itself, not the barrier height, is what gates nucleation.","If phase mixing had instead occurred, the gravitational-wave signal from colliding bubbles would be absent or strongly suppressed; therefore future stochastic gravitational-wave observations around Hz to kHz can discriminate between the two transition channels."],"forward_implications":["For $m^2$-type flat potentials, the standard first-order transition description, with rapid bubble nucleation, expansion, and collision, remains valid even when the barrier width $\\phi_b$ is below the thermal scale.","In the toy SUSY model, bubble percolation happens before curvature flipping at $T=T_2$ for $\\lambda\\gtrsim0.05$; for smaller $\\lambda$ the fate is unsettled but gravitational-wave signals are suppressed.","Bubble walls do not run away; the small vacuum-energy difference makes the transition sound-wave dominated, producing a gravitational-wave spectrum peaked above the Hz range.","The transition rapidity $\\beta/H$ ranges from about $10^2$ to $10^6$ depending on $\\lambda$ and the mass hierarchy, placing observable signals within reach of future detectors."],"supporting_citations":[{"why":"The earlier claim that thermal fluctuations dominate before percolation and force phase mixing; this paper's lattice simulations directly test and oppose that conclusion.","marker":"[60]"},{"why":"Establishes the thermal-escape picture and the Fokker-Planck/Langevin equivalence that underlies the bubble nucleation rate used throughout.","marker":"[61–69]"},{"why":"Provides the semi-analytic bounce action $S_3/T\\simeq 19\\sqrt{3}\\,\\phi_b/\\sqrt{\\lambda_T}$ used to set benchmark parameters and to approximate $S_3/T$ near $T_2$.","marker":"[88]"},{"why":"BubbleDet numerically computes the nucleation prefactor for small barriers, used for the benchmark nucleation rates.","marker":"[89]"},{"why":"Supplies the dimensionally reduced 3D effective theory and Langevin treatment of the zero Matsubara mode on which the simulation setup is based.","marker":"[72]"},{"why":"Provides lattice renormalisation counterterms and matching used to keep the simulated 3D theory consistent with the continuum.","marker":"[77]"},{"why":"CosmoTransitions computes the bounce action $S_3/T$ for the toy SUSY model, giving the sharp drop around $T_2$ and the percolation analysis.","marker":"[98]"},{"why":"Gives the sound-wave gravitational-wave spectrum model used for the predicted signal.","marker":"[100]"}],"fun_headline_variants":["Supercooling ends with bubbles, not smooth roll","Curvature flip still yields bubble nucleation","Lattice shows bubbles when supercooling ends","Even with barrier vanishing, bubbles nucleate"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole argument rests on one assumption: the simplified barrier potential and white-noise thermal fluctuations used in the simulations behave like the real scalar-field theory near the moment the barrier vanishes, including how the field evolves at larger field values.","fun_headline_variants_meta":{"raw":{"variants":["Supercooling ends with bubbles, not smooth roll","Curvature flip still yields bubble nucleation","Lattice shows bubbles when supercooling ends","Even with barrier vanishing, bubbles nucleate"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000492,"raw_usage":{"total_tokens":2431,"prompt_tokens":972,"completion_tokens":1459,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":588,"completion_tokens_details":{"reasoning_tokens":1401}},"tokens_in":588,"tokens_out":1459,"duration_ms":9721,"temperature":1.0,"reasoning_tokens":1401,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:01:05.340362+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate the actual finite-temperature effective potential of the toy SUSY model, including the full one-loop and daisy-resummed terms, on the same lattice and watch a spatial slice: if the field variance grows to the true vacuum without forming distinct round true-vacuum regions, or if a benchmark with $\\lambda<0.05$ completes by homogeneous roll-down before bubbles percolate, the bubble-nucleation conclusion would fail. A cheaper check is to compute the tachyonic growth time from Eq. (2.32) using the real $V''$ just below $T_2$ and compare it with the inverse bubble nucleation rate per Hubble volume.","supporting_citations":[],"review_version":1}