{"id":"1145c9b6-47d2-43ab-a72f-d9dcdf6e6022","arxiv_id":"hep-th/0310221","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Causality in the dual AdS geometry imposes a speed limit on the N=4 SYM scalar moduli, and coupling that DBI-corrected field theory to gravity yields power-law inflation from an ordinary quadratic potential at subPlanckian field values.","lead":"A rolling scalar field in strongly coupled N=4 super Yang Mills cannot move arbitrarily fast through its moduli space: bulk causality in the dual AdS geometry caps its speed. The authors show that this cap, when the field theory is coupled to four-dimensional gravity, can drive a period of accelerated expansion from a steep quadratic potential, a mechanism they call D-cceleration.","discovery_kind":"unclear","skeptic_critique":{"model":"selfhosted/deepseek-v4.1-flash","headline":"The abstract's 'subPlanckian VEV' claim (abstract:s4) is contradicted by the manuscript's own footnote 3 in Section 5.2, which concedes that observational constraints require Planck-scale φ — the regime where Section 5.3 says curvature corrections need fine tuning.","rationale":"Reading the derivation line by line, the core chain is sound: (5.39) and (5.42) follow from the Friedmann equation plus the DBI energy density, and (5.46)–(5.48) are algebraically consistent with the small-φ expansion of (5.42). The displayed forms of (5.33), (5.34), (5.40) and (5.45) contain f↔1/f and √λ↔1/√λ slips relative to what (5.38), (5.42) and (5.46) require, but these are transcription-level and the downstream equations are self-consistent, so they are not load-bearing. The reader's weakest_assumption — the hand-inserted mass term — is real but fully acknowledged by the authors (Sections 4.1.1–4.1.2, 'we have not constructed an explicit example'), so it is not a new finding. The sharper, unresolved issue is the internal tension flagged above: the abstract's subPlanckian claim versus footnote 3 and Section 5.3's parenthetical. That tension is what a referee should press on, because it determines whether the paper's stated advantage over [2] survives. The Section 3.2.1 backreaction estimates differ by λ^{1/4} between the Aichelburg-Sexl and domain-wall methods, but both leave a strong-coupling window, so that is sub-leading. Verdict stays CONDITIONAL: the mechanism is a genuine contribution, but the headline novelty is not established as stated.","tokens_in":28373,"tokens_out":12265,"duration_ms":214259,"concrete_test":"Using the Case B background a(t) = a0 t^p with p = h1√λ from (5.47) and the mode equation (6.63), compute the comoving curvature perturbation amplitude and the non-Gaussianity parameter f_NL as functions of (p, λ, gs, m/Mp). Fix the amplitude to the observed COBE value at 60 e-foldings before the end of inflation and ask whether the required initial φ satisfies φ < Mp. If the amplitude forces φ_start ≳ Mp, footnote 3 is confirmed and the abstract's 'subPlanckian VEV' claim should be withdrawn or qualified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The selected main claim (abstract:s4) asserts slow-roll inflation 'at subPlanckian VEV.' Two passages in the manuscript undercut exactly that qualifier. Footnote 3 (Section 5.2) states: 'In our subsequent analysis of density perturbations [39], we discovered that to satisfy all the observational constraints including those on non-Gaussianity, we require starting inflation at Planck scale φ, which may require fine tuning.' Section 5.3 then argues that curvature corrections Rφ² are harmless 'in the regime of our solutions where φ ≪ Mp,' adding parenthetically that 'in other regimes more fine tuning would be required [39].' Together these say: the observationally viable regime is φ ~ Mp, and in that regime the paper's central selling point — that DBI kinetics evades the curvature-correction problems of [2] without fine tuning — does not hold. The mechanism (speed limit → power-law inflation from a quadratic potential, (5.47)–(5.49)) survives; the distinctive 'subPlanckian, no fine tuning' claim does not. This is not the acknowledged mass-term gap (Sections 4.1.1–4.1.2, where the authors openly say they have no explicit compactification); it is an internal contradiction between the abstract and the manuscript's own footnote, left unresolved in the text.","agreement_with_reader":"agree"},"referee_report":{"model":"selfhosted/deepseek-v4.1-flash","summary":"The paper proposes that the Dirac-Born-Infeld kinetic term for a probe D3-brane moving in the AdS Poincaré patch provides a microphysical model of the N=4 SYM Coulomb-branch scalar in which bulk causality imposes a speed limit, φ̇ ≤ φ²/√λ, on the modulus. In the global CFT the late-time trajectory saturates this limit, φ → √λ/t (Section 3.2). Coupling the sector to four-dimensional gravity and adding a potential V = V0 + V2φ² + V4φ⁴ + …, the authors derive Hamilton-Jacobi equations (5.39)–(5.40) and solve them case by case. The central result is Case B of Section 5.1: with V2 ≠ 0, the ansatz H = h1φ gives φ → √λ/t and a(t) ~ t^{h1√λ}, with h1√λ ≈ √(V2λ/(3gsMp²)), so power-law inflation obtains when V2λ/(gsMp²) > 1 (Eq. 5.49). This is acceleration from a steep quadratic potential rather than a flat one. Other potentials yield dust-like or transiently accelerating cosmologies. Sections 3.2.1 and 6 check backreaction, proper acceleration, and particle production.","tokens_in":28683,"tokens_out":18345,"duration_ms":344253,"significance":"If the mechanism holds, it is a microphysical realization of k-inflation with a genuinely distinctive feature: the DBI kinetic term, not a flat potential, supplies slow roll, so acceleration strengthens with a steeper mass. The derivation is explicit and checkable; the Hamilton-Jacobi reduction and the case-by-case late-time solutions follow cleanly from the DBI action, and the field redefinition in Section 5.4 connects the construction to tachyon-matter cosmology in a useful way. The paper is candid about its main gap: no explicit compactification generates the required mass term. The value of the paper is the mechanism and the worked cosmology, not a complete model, and it should be read on those terms.","major_comments":[{"comment":"The abstract claims slow-roll inflation 'at subPlanckian VEV without need for a flat potential,' but footnote 3 states that satisfying observational constraints, including non-Gaussianity, requires starting inflation at Planck-scale φ, 'which may require fine tuning,' and Section 5.3 says the curvature corrections Rφ² are harmless only for φ ≪ Mp, with 'more fine tuning required' otherwise. The mechanism of Eqs. (5.47)–(5.49) survives, but the distinctive subPlanckian and no-fine-tuning claim does not. Please revise the abstract to match the body and state explicitly in Sections 5.2–5.3 what the observationally viable window is and what tuning it requires.","section":"Abstract [abstract:s4]; footnote 3 (Section 5.2); Section 5.3","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The only concrete mass estimate in the paper, m ~ 1/(l_s λ^{1/4}) from throat-to-throat couplings in Section 4.1.2, combined with the e-folding criterion n ~ m√λ/(√gs Mp) ≥ 100 in Eq. (5.52), requires λ^{1/4} ≳ 100 √gs Mp l_s. For typical values of Mp l_s this is an extremely large 't Hooft coupling, and the paper does not check that its own estimate satisfies (5.52). Please quantify the coupling and cutoff needed, or provide an estimate that does satisfy the bound; otherwise the inflation window (5.51) is not demonstrated with the scales the paper itself supplies.","section":"Section 4.1.2; Eq. (5.52)","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The expansion of the energy gives ½φ̇² + (3/8)λφ̇⁴/φ⁴ + …, not the printed (1/8) coefficient; please check the second-order term.","section":"Eq. (3.11)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The printed solution h1 = (1/(3√λ))(1 + √(1 + 3V2λ/(gsMp²))) does not appear to satisfy (5.45) exactly; the exact root is h1 = (√λ/3)(1 + √(1 + 3V2/(λgsMp²))). The leading approximation in (5.48) is unaffected, but the exact form should be corrected.","section":"Eqs. (5.45)–(5.48)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Reference [22] (Maldacena 1998) is cited for the claim that the DBI form holds at weak coupling 'without recourse to a dual gravitational description'; [22] is the dual gravitational description, so the citation appears misplaced and should be checked.","section":"Section 3.1","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The exit from inflation and reheating are acknowledged as uncontrolled ('we have not yet controlled the details of the exit'). Since the whole window (5.51) depends on μ ≪ φ ≪ Mp, a sentence on what happens as φ reaches μ and whether the transition is consistent with the assumed background would help.","section":"Section 5.2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The claimed solutions α ~ t² and t^{3/ε0} of (6.63) should be checked against the printed equation; with H = ε0/t the modes appear to be t^{-2} and t^{-3-3ε0}, which is consistent with the text's statement that they decay faster than in ordinary slow roll but not with the printed exponents.","section":"Section 6.2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Typographical errors: 'confont' (Section 2.1), 'accelaration' (Section 3.2), 'stategy' (Section 5), and 'there there are three possib le cases' (Section 5.1).","section":"Sections 2.1, 3.2, 5, 5.1","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision","confidential_remarks":"The central mechanism is sound and the derivations check out, but the abstract's 'subPlanckian VEV' claim is not reconciled with footnote 3, which was added in v3 after the companion paper [39]. This is the main editorial issue and should be resolved before publication. The mass-term gap is acknowledged openly and is not a novelty concern; the paper's contribution relative to [2] is the DBI kinetic term, which is clearly stated."},"author_rebuttal":null,"desk_editor":{"model":"selfhosted/deepseek-v4.1-flash","letter":"Worth your time, with one caveat about the abstract's framing. The real content is a clean, explicit derivation. A probe D3 in the AdS Poincaré patch has a DBI kinetic term that imposes a speed limit on the modulus, φ̇ ≤ φ²/√λ, and when you couple that to 4d gravity the late-time attractor is φ → √λ/t with a(t) ~ t^{h₁√λ}. For a quadratic potential that gives power-law inflation with h₁√λ ≈ √(V₂λ/(3 g_s M_p²)), so acceleration needs a large mass, condition (5.49) — the opposite of the usual flat-potential requirement. The Hamilton-Jacobi reduction to (5.39)–(5.40) is correct, the case taxonomy in 5.1 (V₀, V₂, V₄, negative V) is a genuine addition, and the perturbation analysis in 6.2 is more careful than most k-inflation papers bother to be. The relation to tachyon matter via the field redefinition in 5.4 is honest about what is shared and what is not. Soft spots, in order of how much they matter. First, the mass term: everything in Case B rests on a V₂φ² the paper inserts by hand. Section 4.1.1 argues couplings φ²η² can generate it without spoiling the DBI terms, and 4.1.2 estimates m ~ 1/(l_s λ^{1/4}), but the authors say outright they have no explicit compactification and that KK-mode tuning may be needed. They also note the hard mass from the cutoff geometry, m_DBI ~ μ/√λ, is too small to inflate. So the mechanism is real but the model is not built. That is acknowledged, not concealed. Second, and this is the one I would raise with them: the abstract's 'subPlanckian VEV' is undercut by their own footnote 3 in 5.2, which says their later density-perturbation analysis [39] requires starting inflation at Planck-scale φ. Section 5.3 then says curvature corrections are harmless only for φ ≪ M_p. The mechanism survives; the distinctive 'no fine tuning, sub-Planckian' qualifier does not. Third, minor: the two backreaction estimates in 3.2.1 differ by λ^{1/4} (3.15 versus 3.17). Both give a window, so conclusions are unchanged, but it is sloppy. Who should read it: anyone working on string inflation, k-inflation, or DBI cosmology. It is a foundational reference in that literature and deserves to be. Send it out.","headline":"A clean, explicit derivation of DBI-driven power-law inflation from a quadratic potential — the mechanism holds up, but the abstract's 'subPlanckian, no fine tuning' claim is undercut by the paper's own footnote.","tokens_in":29308,"tokens_out":1664,"would_cite":true,"duration_ms":33802,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":[],"pacs":["98.80.Cq","11.25.Tq"],"model":"selfhosted/deepseek-v4.1-flash","headline":"The paper argues that the Dirac-Born-Infeld kinetic term of a probe D3-brane falling down an AdS throat, coupled to four-dimensional gravity, produces slow-roll inflation from an ordinary quadratic mass term at sub-Planckian field values, a","keywords":["AdS/CFT correspondence","Dirac-Born-Infeld action","k-inflation","moduli space speed limit","slow roll inflation","D3-brane cosmology","tachyon matter","FRW cosmology"],"falsifier":"Compute, in an explicit throat compactification, the coefficient of $varphi^2$ generated by couplings to other sectors and the quartic coefficient generated alongside it: if the mass cannot reach $V_2lambda > g_sM_p^2$ without the induced $V_4$ violating $varphi^2 < m^2/V_4$ somewhere in the window $mu ll varphi ll M_p$, the accelerating phase disappears. The authors' own footnote already reports that the follow-up analysis of density perturbations and non-Gaussianity pushes the start of inflation back to Planck-scale $varphi$, which would remove the sub-Planckian feature that motivates the construction.","tokens_in":29722,"feed_emoji":"🌌","tokens_out":4892,"duration_ms":89227,"temperature":0.7,"pith_summary":"In the strong-coupling limit of $N=4$ super Yang-Mills, the scalar fields that parametrize the moduli space cannot roll arbitrarily fast. Causality on the gravity side of the correspondence caps the velocity at $|dot varphi| le varphi^2/sqrt{lambda}$, so the field takes an infinite time, as measured by a boundary observer, to reach the origin. The paper's aim is to show that this cap, which comes from the Dirac-Born-Infeld kinetic term of a probe D3-brane descending the AdS throat, survives the coupling of the gauge theory to four-dimensional gravity and changes which cosmologies are available. The headline consequence is a mechanism for slow-roll inflation from an ordinary quadratic mass term at sub-Planckian field values: because the kinetic term shuts down as the field approaches the origin, the potential barely has to be flat, and acceleration in fact gets stronger as the mass grows. The authors also catalogue the non-inflating cosmologies of the same system, including a dust-like phase resembling tachyon matter, and check that particle production, string production, and back reaction stay under control at large 't Hooft coupling.","feed_headline":"A speed limit on scalar fields can drive inflation","feed_subtitle":"Strong-coupling kinetic terms let a steep quadratic potential accelerate the universe without a flat direction.","key_machinery":"The engine is the Dirac-Born-Infeld Lagrangian for a probe D3-brane moving radially in the AdS Poincare patch, written in field-theory variables with $f(varphi) = lambda/varphi^4$, together with the Lorentz-like factor $gamma = (1-f(varphi)|dotvarphi^2)^{-1/2}$ built from it. Expanding the square root gives the canonical kinetic term plus a series in $lambda|dotvarphi^2/varphi^4$ that becomes order one near the origin; this is what enforces the speed limit and what makes $gamma$ grow without bound at late times. The cosmological solutions are then obtained through the Hamilton-Jacobi trick of treating $varphi$ as the time variable, which turns the Friedmann equations into a first-order pair: an algebraic equation fixing $V(varphi)$ in terms of $H(varphi)$ and $H'(varphi)$, and a separate expression for $|dotvarphi$ in terms of $H'$ and $f$. That separation is what allows the authors to solve sequentially and read off the scale-factor exponent directly from the coefficient $h_1$.","core_discovery":"In Case B of Section 5.1 the authors take a potential $V = V_2varphi^2$ and an ansatz $H = h_1varphi$, solve the Hamilton-Jacobi form of the Friedmann equations, and find $varphi to sqrt{lambda}/t$ together with $a(t) propto t^{h_1sqrt{lambda}}$, where $h_1sqrt{lambda} = (1/3)(1+sqrt{1+3V_2lambda/(g_sM_p^2)})$. Inflation therefore requires $V_2lambda/(g_sM_p^2) > 1$, which is the reverse of the usual condition: a steeper potential accelerates more, because the speed limit rather than the flatness of $V$ supplies the slow roll. The same mechanism defuses the curvature coupling $Rvarphi^2$, since the factor $gamma = (1-lambda|dotvarphi^2/varphi^4)^{-1/2}$ that blows up at late times suppresses both the potential and curvature terms in the scalar equation of motion, leaving $M_p^2 + varphi^2$ as a small shift to the effective Planck mass when $varphi ll M_p$. Different potentials give different endpoints: a pure quartic, which is what an anti-D3-brane generates, yields $a sim t^{2/3}$ and a kinetic dust equation of state, while a negative $V_2$ permits a burst of acceleration that settles into a steady state with $a sim exp(-c/t^2)$. A toy capped throat with $f(varphi) = lambda/(varphi^2+mu^2)^2$ lets the field reach the origin in finite time, so it can oscillate and reheat.","pith_inferences":["The $gamma$-suppression argument is not specific to this brane: any modulus whose kinetic term descends from a warped DBI action should get the same protection from curvature couplings, which would make speed-limit inflation a generic feature of throat compactifications rather than a single model.","Because the velocity bound is causal and hard, the total field-space distance the scalar can cover is tied to the elapsed time; this suggests an observable relation between the tensor-to-scalar ratio and the field excursion that would distinguish the scenario from ordinary slow roll.","The construction points to a search criterion for explicit flux compactifications: look for sectors that generate a soft mass for a throat modulus without generating a comparable quartic coupling, since that combination is precisely what opens the accelerating window.","The non-Gaussianity constraint reported in the authors' own footnote suggests the sub-Planckian feature and the observed density-perturbation amplitude may be in tension; a dedicated $f_{NL}$ calculation for Case B would decide whether the mechanism survives contact with the CMB."],"forward_implications":["Inflationary model building in string compactifications need not hunt for flat directions: a quadratic mass term suffices, and the required condition is a lower bound on the mass, $V_2lambda > g_sM_p^2$, rather than an upper bound.","The number of e-foldings is $n sim h_1sqrt{lambda} sim msqrt{lambda}/(sqrt{g_s}M_p)$, so a large enough soft mass gives more than 100 e-foldings over the window $mu ll varphi ll M_p$.","Curvature corrections of the form $Rvarphi^2$ and $H^2varphi^2$, which obstruct other brane-inflation models, are suppressed by powers of $gamma^{-1}$ in this regime, so sub-Planckian inflation is protected.","An anti-D3-brane in the throat produces $a sim t^{2/3}$ with particle production suppressed by $1/sqrt{lambda}$ and $1/lambda^{1/4}$, a cleaner relative of tachyon matter with the same dust equation of state.","A capped throat lets the scalar pass through the origin at finite speed, so the field oscillates about the minimum afterwards and ordinary reheating becomes available."],"supporting_citations":[{"why":"Prior identification of the bulk causal speed limit on moduli space, the effect the paper builds its cosmology on.","marker":"[1]"},{"why":"Supplies the conventions and the DBI action for a probe D3-brane in $AdS_5 times S^5$ used as the starting Lagrangian.","marker":"[20]"},{"why":"Grounds the claim that the higher-derivative DBI form is fixed by nonlinearly realized conformal invariance plus non-renormalization theorems, even at weak coupling.","marker":"[22]"},{"why":"The k-inflation class the paper says it realizes in a microphysical framework.","marker":"[3]"},{"why":"The brane-inflation obstacles (curvature couplings, super-Planckian VEVs) that the speed-limit mechanism is meant to circumvent.","marker":"[2]"},{"why":"The cosmological precedent for Case B: power-law inflation from an inverse-power potential in the rolling-tachyon system.","marker":"[9]"},{"why":"The rolling-tachyon cosmology the paper's dust-like Case C is compared against.","marker":"[8]"},{"why":"The prototype capped throat motivating the mass-gap toy model $f(varphi) = lambda/(varphi^2+mu^2)^2$.","marker":"[31]"},{"why":"The Hamilton-Jacobi formalism used to reduce the Friedmann equations to the first-order pair that the solutions are read off from.","marker":"[34]"}],"fun_headline_variants":["Speed limit, not flat potential, drives this inflation","Steep quadratic potential inflates via strong-coupling speed limit","DBI speed limit turns a steep quadratic into slow-roll inflation","Acceleration from D-cceleration: speed limit powers inflation","Scalar speed limit reverses inflation's flatness requirement"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The inflationary window needs a quadratic mass term for $varphi$ large enough to satisfy $V_2lambda > g_sM_p^2$, and the paper inserts that term by hand: it argues that couplings to other sectors of a compactification can generate it while leaving the DBI kinetic terms intact, but states plainly that no explicit compactification example has been constructed, that tuning of Kaluza-Klein couplings may be required, and that the hard mass available from the cutoff geometry itself, $m_{DBI} sim mu/sqrt{lambda}$, is far too small to inflate.","fun_headline_variants_meta":{"raw":{"variants":["Speed limit, not flat potential, drives this inflation","Steep quadratic potential inflates via strong-coupling speed limit","DBI speed limit turns a steep quadratic into slow-roll inflation","Acceleration from D-cceleration: speed limit powers inflation","Scalar speed limit reverses inflation's flatness requirement"]},"model":"selfhosted/deepseek-v4.1-flash","effort":"low","attempts":1,"cost_usd":0.0,"raw_usage":{"total_tokens":2613,"prompt_tokens":1099,"completion_tokens":1514,"prompt_tokens_details":null,"completion_tokens_details":{"reasoning_tokens":1444}},"tokens_in":1099,"tokens_out":1514,"duration_ms":30341,"image_pages":[],"temperature":1.0,"reasoning_budget":50,"reasoning_tokens":1444,"transport_provider":"selfhosted","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-10-29T17:01:36.219118+00:00","model_set":{"reader":"selfhosted/deepseek-v4.1-flash"},"falsifier":"Compute, in an explicit throat compactification, the coefficient of $varphi^2$ generated by couplings to other sectors and the quartic coefficient generated alongside it: if the mass cannot reach $V_2lambda > g_sM_p^2$ without the induced $V_4$ violating $varphi^2 < m^2/V_4$ somewhere in the window $mu ll varphi ll M_p$, the accelerating phase disappears. The authors' own footnote already reports that the follow-up analysis of density perturbations and non-Gaussianity pushes the start of inflation back to Planck-scale $varphi$, which would remove the sub-Planckian feature that motivates the construction.","supporting_citations":[{"cited_title":"Gauge theory origins of supergravity causal structure","cited_arxiv_id":"hep-th/9902073","evidence_quote":"Prior identification of the bulk causal speed limit on moduli space, the effect the paper builds its cosmology on."},{"cited_title":"Large N Field Theories, String Theory and Gravity","cited_arxiv_id":"hep-th/9905111","evidence_quote":"Supplies the conventions and the DBI action for a probe D3-brane in $AdS_5 times S^5$ used as the starting Lagrangian."},{"cited_title":"The large N limit of superconformal field theories and superg ravity","cited_arxiv_id":null,"evidence_quote":"Grounds the claim that the higher-derivative DBI form is fixed by nonlinearly realized conformal invariance plus non-renormalization theorems, even at weak coupling."},{"cited_title":"k-Inflation","cited_arxiv_id":"hep-th/9904075","evidence_quote":"The k-inflation class the paper says it realizes in a microphysical framework."},{"cited_title":"Towards Inflation in String Theory","cited_arxiv_id":"hep-th/0308055","evidence_quote":"The brane-inflation obstacles (curvature couplings, super-Planckian VEVs) that the speed-limit mechanism is meant to circumvent."},{"cited_title":"Power-Law Inflation from the Rolling Tachyon","cited_arxiv_id":"hep-th/0204140","evidence_quote":"The cosmological precedent for Case B: power-law inflation from an inverse-power potential in the rolling-tachyon system."},{"cited_title":"Cosmological Evolution of the Rolling Tachyon","cited_arxiv_id":"hep-th/0204008","evidence_quote":"The rolling-tachyon cosmology the paper's dust-like Case C is compared against."},{"cited_title":"Supergravity and a confining gauge theory: Duality cascades and χSB-resolution of naked singularities","cited_arxiv_id":null,"evidence_quote":"The prototype capped throat motivating the mass-gap toy model $f(varphi) = lambda/(varphi^2+mu^2)^2$."},{"cited_title":"A Hamilton-Jacobi approach to non-slow-roll inflation","cited_arxiv_id":"hep-ph/9702427","evidence_quote":"The Hamilton-Jacobi formalism used to reduce the Friedmann equations to the first-order pair that the solutions are read off from."}],"review_version":1}