{"id":"7da84302-228b-4b5f-b0bf-ecac43d1a2d8","arxiv_id":"2411.12465","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"An alternating drift-and-kick 'zoom-in' scheme for stochastic inflation is equivalent to the Itô interpretation, and the Itô-Stratonovich difference vanishes in the full non-Markovian setup.","lead":"This paper explains that the random noise in stochastic inflation should be added using the Itô scheme, not the Stratonovich scheme, because Itô matches a physical 'zoom-in' process where the observer's coarse-graining scale changes. The two schemes give the same result once the full, non-Markovian physics is included.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Non-Markovian Itô–Stratonovich equality is tied to the H0-anchored coarse-graining scale; with the common local-H prescription the claimed vanishing correction is nonzero.","rationale":"The reader and I converge on the same load-bearing premise: the H0-anchored coarse-graining scale. My reading of the paper confirms that this choice is what makes the diffusion matrix in (4.2) independent of the directly noisy background variables, which is exactly the condition that kills the Stratonovich correction in (4.8). The Markovian equivalence of the alternating scheme to Itô (Section 3.2) is a clean, parameter-free argument and is not threatened by this concern; the numerical demonstration in Section 3.4 corroborates it. The non-Markovian equality, however, is presented in the abstract and in the closing paragraph of Section 4.1 as a general property of the physical system ('these results are general, and apply...'), when in fact it relies on a specific coarse-graining convention. The local-H prescription is not exotic: it is the standard way to define the coarse-graining scale in much of the stochastic-inflation literature, and it would introduce a nonzero Itô–Stratonovich drift through the direct dependence of kσ on φ_R and π_R. Since the paper's physical advocacy for Itô leans partly on the non-Markovian equality, the overgeneralization is material. The fix is inexpensive: qualify the result as holding for the H0-anchored comoving scale, or argue why that scale is the unique physical one. My recommendation is therefore CONDITIONAL rather than a flat rejection: the core contributions (zoom-in scheme, Markovian Itô interpretation) stand, but the abstract and Section 4.1 need a caveat or a justification of the coarse-graining choice. The proposed test—an explicit computation of the extra drift for the local-H prescription, or a two-scheme numerical comparison—settles whether the concern truly lands.","tokens_in":27028,"tokens_out":15158,"duration_ms":156532,"concrete_test":"Re-derive the Stratonovich expansion of Section 4.1 for the alternative coarse-graining scale kσ = σ a H(φ_R, π_R), keeping the same mode promotion. Differentiate the diffusion coefficients in (4.4) with respect to φ_R and π_R and compute the extra drift terms (∂σ_δφ/∂φ_R) σ_δφ + (∂σ_δπ/∂φ_R) σ_δφ + ... If these are nonzero, the Itô–Stratonovich difference does not vanish for this prescription. A complementary numerical check: evolve a non-slow-roll model (e.g., an ultraslow-roll plateau) in the full promoted system with the local-H coarse-graining scale, using both the Itô update (4.5) and the Stratonovich update (4.6), and compare the resulting distributions of φ_R and π_R; any statistically significant difference would falsify the paper's claim that the two interpretations coincide in the full non-Markovian setup.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's Section 4.1 shows that in the promoted non-Markovian system the Stratonovich correction (4.8) vanishes because the diffusion coefficients σα_i depend only on the mode variables δφ_kσ and δπ_kσ, which carry no direct noise. This structural property is not intrinsic to stochastic inflation; it follows from the choice in footnote 3 that the coarse-graining scale is Rσ = 1/(σ a H0), anchored to the non-stochastic initial Hubble rate. If instead one uses the equally common prescription R = 1/(σ a H) with the local stochastic H(φ_R, π_R), then kσ depends on φ_R and π_R, so σα_i acquires explicit dependence on the directly noisy background variables. In the Stratonovich expansion, the term (∂σ_i/∂φ_R) σ_φ_R ξ² dN (and the π_R analogue) is generically nonzero, producing an O(dN) extra drift. The claimed equality 'in the full non-Markovian setup' therefore does not hold for that prescription. The paper notes the H0 convention in a footnote but presents the vanishing difference in the abstract and Section 4.1 as a general result, and Section 4.1 explicitly claims generality beyond the exact forms of µ and σ. If the local-H prescription is the one used in most numerical implementations, the conclusion that Itô and Stratonovich coincide in the complete system is at best convention-dependent, weakening the physical motivation for Itô drawn from the non-Markovian argument. The Markovian alternating-scheme result (Section 3.2) is unaffected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the Itô versus Stratonovich ambiguity in stochastic inflation and proposes a physical interpretation based on a \"zoom-in\" picture. The coarse-graining scale is changed in discrete steps: deterministic evolution with fixed comoving scale alternates with instantaneous kicks that add newly exited short-wavelength modes to the coarse-grained field. The paper claims that in the Markovian slow-roll limit this alternating scheme reduces to the Itô prescription, whereas the Stratonovich prescription has no analogous physical picture; it also claims that in the full non-Markovian setup, after promoting the mode functions to dynamical variables, the Itô and Stratonovich prescriptions coincide and both match the alternating scheme. A numerical example with quadratic inflation and a pedagogical appendix on stochastic calculus support the presentation.","tokens_in":27378,"tokens_out":4294,"duration_ms":49077,"significance":"If the central claims survive scrutiny, the paper would give a concrete, physically motivated resolution of a long-standing interpretational question in stochastic inflation, and it would clarify the connection between stochastic inflation, linear perturbation theory, and the classical ΔN formalism. The paper is clearly written, the Markovian expansion in Section 3.2 is correct, and the numerical demonstration in Section 3.4 usefully shows where the two interpretations differ. The main weakness is that the non-Markovian equality of Itô and Stratonovich is tied to a specific choice of coarse-graining scale, a point that the paper states only in a footnote and then presents in the abstract and conclusions as a general result.","major_comments":[{"comment":"The claimed equality of Itô and Stratonovich in the full non-Markovian setup is convention-dependent. The argument that the correction in Eq. (4.8) vanishes uses the structural property in Eq. (4.2) that σα_i depends only on δφ_k and δπ_k, not on φ_R and π_R. This property follows from the choice Rσ = 1/(σ a H0) with non-stochastic H0. If one instead uses the equally common local prescription R = 1/(σ a H(φ_R, π_R)), then kσ depends on φ_R and π_R, so σα_i in Eq. (4.4) acquires explicit dependence on the directly noisy background variables. The term (∂σα_i/∂φ_R) σφ_R ξ² dN (and its π_R analogue) is then generically nonzero, so the O(dN) Stratonovich correction does not vanish. The abstract, Section 4.1, and Section 5 present the vanishing difference as a general result; this should be qualified to the H0-anchored coarse-graining convention, and the local-H case should be discussed explicitly.","section":"§4.1 and footnote 3"},{"comment":"The statement that the result is general \"beyond the exact forms of μ and σ\" because of the shapes in Eq. (4.2) overreaches. The vanishing of (4.8) relies on two simultaneous properties: the diffusion coefficients do not depend on the directly noisy variables, and the mode functions receive no direct noise. These properties are not dictated by stochastic inflation itself but by the chosen split between background and modes and by the anchoring of the coarse-graining scale to H0. For a different window function or a different split where the mode equations carry noise, the same expansion would produce a nonzero correction. The generality claim in Section 4.1 should be restricted to the explicit setup constructed in Sections 2.2–2.3.","section":"§4.1, Eqs. (4.2) and (4.8)"}],"minor_comments":[{"comment":"There are several typos: \"phi2 inflation\" in the Introduction should be \"φ² inflation\"; \"secion 3.3\" in Section 4.1 should be \"section 3.3\"; \"corresponsing\" in Section 3.5 should be \"corresponding\".","section":"Throughout"},{"comment":"The axis label \"ar=H-1\" is easy to misread; using \"ar = H^{-1}\" and \"R_σ\" consistently would improve clarity.","section":"Figure 1"},{"comment":"In the alternating scheme for the non-Markovian case, σα_i is evaluated at N rather than N+ in the kick step, whereas the analogous Markovian scheme (3.9) evaluates the noise amplitude after the drift step. The difference is subleading in dN, but the notation would be clearer if the two lines of (4.9) used the same time argument for the post-drift state.","section":"§4.1, Eq. (4.9)"},{"comment":"The equivalence between the alternating scheme and Itô is close to being definitional, since the drift step contains no O(√dN) component and therefore σ evaluated after the drift agrees with σ evaluated before it to the relevant order. The paper could state this more explicitly to avoid the impression that the expansion in Eq. (3.10) is doing additional work.","section":"§3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is worth publishing after revision, but the abstract and conclusions currently overstate the non-Markovian result. The Markovian part is solid; the non-Markovian equality needs to be presented as a property of the H0-anchored coarse-graining convention rather than as a general consequence of the enlarged phase-space formulation. A referee-confidential observation: the paper would benefit from a short discussion of which coarse-graining convention is used in recent numerical implementations, since that determines how practically relevant the local-H caveat is."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new piece is the explicit alternating zoom-in scheme—deterministic evolution interleaved with instantaneous kicks—and the demonstration that it reduces to Itô in the Markovian slow-roll limit. That is a clean and useful result, and the numerical example supports it. The non-Markovian argument in Section 4.1 is also structurally nice: once the mode functions are promoted to dynamical variables, the diffusion matrix has zeros in the block that would produce a Stratonovich correction, so the two interpretations coincide. That is a real insight, and the paper is honest about many of its assumptions. But one assumption deserves more weight than it gets. The vanishing of the non-Markovian correction hinges on the diffusion coefficients depending only on the mode variables, not on the directly noisy background variables. This is true because the coarse-graining scale is anchored to a non-stochastic H0 (footnote 3). If one instead uses the common local-H prescription, k_sigma picks up a dependence on phi_R and pi_R, and the sigma coefficients acquire explicit stochastic background dependence. The Stratonovich expansion then has an extra drift term and the claimed equality does not follow. The paper presents the Section 4.1 result as general, but it is really specific to the H0-anchored scheme. This does not damage the Markovian alternating-scheme result, but it weakens the broader conclusion that Itô and Stratonovich coincide in the full system. Minor: the numerical example is a simple phi^2 model and no code is shipped, but the method is straightforward and the analytic argument carries the weight. The appendix pedagogy is fine, though familiar to most practitioners. Who should read it: anyone implementing stochastic inflation numerically, and people arguing about the discretisation ambiguity in the literature. The paper deserves serious peer review; the main claim is clarified and the H0-anchoring caveat should be pushed in revision. I'd send it to a competent referee rather than desk-reject it.","headline":"A clean formalization of the alternating zoom-in scheme as Itô, but the claimed non-Markovian Itô–Stratonovich equality only holds for the H0-anchored coarse-graining prescription.","tokens_in":639,"tokens_out":743,"would_cite":true,"duration_ms":22515,"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":"Alternating between classical drift and quantum kicks, stochastic inflation's zoom-in scheme matches Itô integration in the slow-roll limit, and the two rules coincide when modes are tracked exactly.","keywords":["stochastic inflation","Itô versus Stratonovich","zeno-in scheme","coarse-graining scale","separate universe approximation","non-Markovian noise","slow-roll approximation","stochastic differential equations"],"falsifier":"Recompute the expansion with a coarse-graining scale defined as $R = 1/(\\sigma a H(\\phi_R,\\pi_R))$ so the diffusion components depend on the stochastic background; if non-vanishing extra drift terms appear at order $dN$, the claimed equality of Itô and Stratonovich in the non-Markovian setup fails. A numerical cross-check is to evolve the full equations for a steep potential where the slow-roll parameters change abruptly and compare the alternating scheme against the Stratonovich midpoint rule; discrepancies would show the scheme choice still matters.","tokens_in":26794,"feed_emoji":"🪐","tokens_out":9997,"duration_ms":84233,"temperature":0.7,"pith_summary":"The noise in stochastic inflation is not an external bath: it is generated by the coarse-graining scale itself, so the correct way to integrate the stochastic equations should follow from the physics of that 'zoom-in' process. This paper makes that link explicit by introducing the alternating zoom-in scheme, in which classical deterministic evolution and instantaneous kicks from newly stretched modes alternate. In the Markovian slow-roll limit the scheme reduces to the Itô interpretation, while Stratonovich integration would make the kick depend on the post-kick field value and has no similar physical picture. In the full non-Markovian system, where the short-wavelength mode functions are promoted to dynamical variables, Itô and Stratonovich coincide, and again match the alternating scheme. If these claims hold, the Itô convention is the physically motivated default for stochastic inflation, and any apparent Stratonovich shift is an artifact of the Markovian approximation.","feed_headline":"Zoom-in rule picks Itô over Stratonovich in inflation","feed_subtitle":"Alternating kicks match Itô, and the schemes merge once modes are tracked exactly.","key_machinery":"The central object is the alternating zoom-in scheme, a discretization in which the coarse-graining scale $R$ stays fixed during classical evolution and then jumps to its new value instantaneously, injecting the modes between the two scales as delta-function kicks; in the $dN \\to 0$ limit the kicks become the white noises $\\xi_\\phi$ and $\\xi_\\pi$ in the stochastic equations. The load-bearing identity is that in the Markovian case the kick's diffusion coefficient is evaluated at the pre-kick value after drift but without noise, so no $\\xi^2$ term survives, matching the Itô rule; for Stratonovich the midpoint evaluation would generate the extra drift $(\\sigma/2)\\,\\partial_\\varphi\\sigma$. In the non-Markovian case the same expansion gives a contribution $\\partial\\sigma/\\partial\\Phi_l \\, \\sigma_l$ that vanishes because $\\sigma$ depends only on the perturbation components $\\delta\\phi_k$, $\\delta\\pi_k$, whose drift receives no noise. That vanishing of the correction term is what makes Itô, Stratonovich, and the alternating scheme agree in the full system.","core_discovery":"Stochastic inflation's noise comes from a fixed comoving coarse-graining radius $R_\\sigma = 1/(\\sigma a H_0)$ shrinking relative to physical scales, so modes continually cross the scale and join the coarse-grained local universe. The paper's central claim is that this zoom-in process, implemented as alternating steps of classical drift and instantaneous kicks, defines the stochastic integral: in the slow-roll Markovian limit the step-wise rule reduces exactly to Itô's scheme, with the kick amplitude evaluated before the kick, whereas Stratonovich's midpoint rule would require the noise to know the post-kick field value. In the general non-Markovian case the mode functions $\\delta\\phi_k$ and $\\delta\\pi_k$ are promoted to first-class stochastic variables; the diffusion matrix then depends on the current perturbation values, whose own noise components vanish, so the extra Stratonovich drift term disappears and Itô and Stratonovich give the same evolution. The alternating scheme remains valid in both regimes. The paper also shows the Itô–Stratonovich drift difference is suppressed by $H/M_{\\mathrm{Pl}}$ and only becomes significant at super-Planckian energy densities, where the Markovian slow-roll equation is no longer trustworthy.","pith_inferences":["A testable extension: recompute the non-Markovian expansion with the coarse-graining scale anchored to the local stochastic Hubble parameter $H$ instead of the initial value $H_0$; a non-vanishing extra drift would delimit exactly where the Itô–Stratonovich coincidence stops.","The paper states that the multifield generalization is trivial; spelling it out would give a concrete check of whether the vanishing correction survives extra fields and spectator sectors.","The lighthouse analogy suggests a criterion for when a stochastic variable is 'fictitious': if its rapid variation arises purely from changing a bookkeeping scale, it carries no energy, a criterion that could be imported to other effective theories with moving cutoffs.","A two-step finite zoom-in benchmark could make the optimal-scheme question quantitative: comparing a classical $\\Delta N$ calculation with a single macroscopic jump against full stochastic inflation on the same model would show how much accuracy the $dN \\to 0$ limit actually buys."],"forward_implications":["Numerical codes that implement stochastic inflation in the slow-roll limit can safely use the simple Euler/Itô update or the alternating scheme; both give the same distribution, as demonstrated for a $\\phi^2$ potential.","In non-Markovian simulations that evolve mode functions alongside the background, the choice of Itô versus Stratonovich is irrelevant; the physical result is scheme-independent.","Any significant difference between Itô and Stratonovich in a Markovian calculation signals that the Markovian approximation is breaking down and the full mode-resolved equations should be used.","The two-step finite zoom-in scheme, with a single macroscopic jump between classical evolution periods, reproduces the setup of the classical $\\Delta N$ formalism and fixes which super-Hubble modes contribute to the initial distribution.","Stochastic velocities and energies are not physical: because kicks are changes of coarse-graining perspective, a large stochastic field velocity carries no energy scale and cannot by itself invalidate the effective theory."],"supporting_citations":[{"why":"early formulation of stochastic inflation as deterministic drift interrupted by impulses; direct predecessor of the alternating zoom-in scheme and an early advocacy for Itô.","marker":"[52]"},{"why":"defines the Itô stochastic integral whose Euler step is the rule the alternating scheme is shown to match.","marker":"[47]"},{"why":"defines the Stratonovich integral and its midpoint rule, from which the extra drift term is derived and then contrasted.","marker":"[48]"},{"why":"numerical implementation of stochastic inflation that uses the alternating description and provides the non-Markovian noise setup this paper generalizes.","marker":"[24]"},{"why":"detailed numerical framework in which mode functions evolve in the stochastic background and backreact; source of the full system of equations used here.","marker":"[33]"},{"why":"stochastic $\\Delta N$ correlation-function framework that notes the Itô–Stratonovich ambiguity persists beyond slow roll; a consistency target for the new interpretation.","marker":"[11]"},{"why":"shows noise and momentum align with the slow-roll attractor, justifying the reduction to the single-field Markovian equation.","marker":"[39]"},{"why":"demonstrates that colored-noise limits do not uniquely select Stratonovich, supporting the paper's case that the physical zoom-in picture must decide.","marker":"[50]"}],"fun_headline_variants":["Zoom-in scheme identifies Itô rule in inflation","Itô wins in stochastic inflation via zoom-in steps","Zoom-in procedure pins down Itô integral for inflation","Non-Markovian merges Itô and Stratonovich in inflation","Zoom-in kicks: Itô in Markovian, merged beyond"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rests on anchoring the comoving coarse-graining scale to a non-stochastic initial Hubble parameter $H_0$ rather than to the local stochastic Hubble parameter $H$; if the scale followed $H$, the noise would depend directly on the stochastic field variables and the vanishing of the Itô–Stratonovich correction in the full setup would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Zoom-in scheme identifies Itô rule in inflation","Itô wins in stochastic inflation via zoom-in steps","Zoom-in procedure pins down Itô integral for inflation","Non-Markovian merges Itô and Stratonovich in inflation","Zoom-in kicks: Itô in Markovian, merged beyond"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000279,"raw_usage":{"total_tokens":1665,"prompt_tokens":964,"completion_tokens":701,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":580,"completion_tokens_details":{"reasoning_tokens":619}},"tokens_in":580,"tokens_out":701,"duration_ms":6277,"temperature":1.0,"reasoning_tokens":619,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:30:06.264721+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the expansion with a coarse-graining scale defined as $R = 1/(\\sigma a H(\\phi_R,\\pi_R))$ so the diffusion components depend on the stochastic background; if non-vanishing extra drift terms appear at order $dN$, the claimed equality of Itô and Stratonovich in the non-Markovian setup fails. A numerical cross-check is to evolve the full equations for a steep potential where the slow-roll parameters change abruptly and compare the alternating scheme against the Stratonovich midpoint rule; discrepancies would show the scheme choice still matters.","supporting_citations":[{"cited_title":"Stratonovich,A new representation for stochastic integrals and equations, SIAM Journal on Control 4 (1966) 362 [https://doi.org/10.1137/0304028]","cited_arxiv_id":null,"evidence_quote":"defines the Stratonovich integral and its midpoint rule, from which the extra drift term is derived and then contrasted."}],"review_version":1}