{"id":"73b6d16a-b81b-4bdc-9871-cabbde409a1c","arxiv_id":"2411.18570","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"For light multiply scattered in atomic vapor, finite vapor size truncates the frequency-redistribution wings and hence the Lévy step-size distribution, while alkali hyperfine structure creates oscillations in the derived Lévy exponent.","lead":"This paper calculates how the distance a photon travels between scatterings in an atomic vapor depends on the vapor's size and on the multi-level structure of alkali atoms. It shows that finite-size effects cut off the largest jumps and that hyperfine structure makes the Lévy exponent oscillate as a function of step size, which matters for interpreting light-transport experiments and astrophysical spectra.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The finite-size cutoff in α(r) rests on the [1−T(x)] single-big-jump factor in Eq. 9; the Fig. 1 Monte Carlo validates spectra but not the step-size/α predictions, so a direct step-size benchmark is needed.","rationale":"The reader's weakest assumption correctly identifies the [1−T(x)] factor in Eq. 9 as the load-bearing element for the finite-size predictions. However, the reader's statement that the Monte Carlo check is 'not an independent test' is too strong: the Monte Carlo does solve the full boundary-value random walk and therefore tests the recurrence approximation of Eq. 9, even though it does not test the underlying physical scattering model (RII/RIII, Doppler widths, etc.). The genuine remaining gap is that the Monte Carlo comparison is only for the scattered spectra, while the headline results are the step-size distribution and the associated α(r) curves. Since α(r) is a logarithmic derivative of P(r), small discrepancies in the spectrum or in the escape treatment can be amplified in the derived exponent. The experimental comparisons are acknowledged by the authors to be qualitative and use different parameters, so they do not close this gap. A direct Monte Carlo measurement of the step-size distribution is the minimal, concrete check that would settle whether Eq. 9 is quantitatively accurate for the finite-size cutoff. Until that check is performed, the conditional verdict is appropriate, but no stronger criticism is warranted by the text.","tokens_in":13699,"tokens_out":6223,"duration_ms":64032,"concrete_test":"Run the paper's own Monte Carlo for the RII case in the finite cylindrical cell (radius and thickness r_L) exactly as described in §3, but record the length of every step for photons that survive to n=10 scatterings; form P(r) and compute 1+α(r) with the same log-derivative definition used in Fig. 3(b). Repeat for r_L=10^2, 10^3, and 10^4. If the Monte Carlo P(r) cutoff and the resulting α(r) curves match the Eq. 9/Fig. 3 predictions within statistical error, the single-big-jump truncation is quantitatively sound; if the cutoff shifts or the α values differ beyond errors, the central finite-size claim is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central new quantitative claim—that finite vapor size truncates the step-size distribution and raises α(r) (Fig. 3)—is obtained from Eq. 9 by inserting the ballistic escape factor 1−T(x)=1−exp[−φ(x)r_L]. This replaces the actual boundary-value random walk with a position-independent Beer-Lambert survival probability. The Single-Big-Jump principle motivates the replacement, but it does not fix the quantitative cutoff shape for n=10, for a cylindrical cell, or for the multilevel RII case, where direction-frequency correlations are modified by hyperfine peaks. The Monte Carlo in Fig. 1 is an independent numerical solution of the random-walk model with the same physical scattering rules, and its agreement with Eq. 9 for the scattered spectra is genuine supporting evidence. However, the figure checks only Θ_n(x), not the derived P(r) or 1+α(r). The comparison with Cs/Rb experiments is qualitative and uses different Γ, Γ_D, and r_L than the calculations, so it cannot independently confirm the predicted cutoff. Thus the load-bearing assumption remains quantitatively unvalidated for the very quantities the paper headlines.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the step-size distribution of photons multiply scattered in resonant alkali vapors, described as a Lévy flight. Starting from Hummer's RII/RIII redistribution functions, the authors write a recurrence for the scattered spectrum Θ_n(x) (Eq. 9). They introduce a finite-size cutoff into this recurrence via the single-flight survival factor 1−T(x) = 1−exp(−ϕ(x) r_L), and then obtain P(r) and the local Lévy exponent 1+α(r) from Eq. 3. They report that finite size truncates P(r) and raises α(r) for RII but not for RIII, and that the hyperfine structure of Cs and Rb induces oscillations in α(r). The RII spectra are compared with Monte Carlo simulations in Fig. 1, while the alkali results are compared qualitatively with existing Cs and Rb experiments.","tokens_in":13912,"tokens_out":17334,"duration_ms":153103,"significance":"The calculation is a useful extension of earlier infinite-medium work [10,18,25] to finite samples and to alkali hyperfine structure. If the finite-size model is quantitatively valid, the predicted r_L-dependence of α(r) is a falsifiable signature that can be tested in transmission or imaging experiments. The use of established Hummer functions, the dimensionless formulation, and the Monte Carlo check of the scattered spectra are strengths. The multilevel treatment and the explicit α(r) curves for Cs D1/D2 and Rb D2 give concrete targets for future experiments. However, the central finite-size prediction is not yet independently validated for the headline quantity α(r), and some numerical and parameter statements in the Rb section are inconsistent.","major_comments":[{"comment":"The finite-size cutoff is obtained by replacing the boundary-value random walk with a position-independent Beer-Lambert survival factor T(x) = exp(−ϕ(x) r_L). The Single-Big-Jump principle motivates this replacement, but it does not fix the quantitative cutoff shape for n = 10, for the cylindrical cell used in the Monte Carlo, or for the multilevel cases. The Monte Carlo in Fig. 1 is a genuine check of the scattered spectra, but it is not an independent check of the step-size distribution P(r) or of 1+α(r), which are the quantities used to state the main finite-size result. The authors should provide a direct Monte Carlo comparison of the simulated step-size distribution and of 1+α(r) with Eqs. (3) and (9), and discuss the sensitivity to the escape geometry; the mean chord length of the simulated cylinder is not r_L, so the agreement for Θ_n(x) does not automatically validate the one-flight factor in Eq. (10).","section":"§3, Eqs. (9)–(10) and Fig. 3"},{"comment":"The claim that the [1−T(x)] factor does not impact the emission in the RIII case is not justified by the statement that emission in the atomic rest frame is independent of the incident frequency. In the laboratory frame, RIII(x,x′)/ϕ(x) still depends on x through the velocity selection encoded in Eq. (6); suppressing large-|x| incident photons changes the distribution of atomic velocities that scatter and therefore the emitted spectrum. The authors should either demonstrate numerically that this effect is negligible for the parameters of Fig. 2 or revise the physical explanation.","section":"§3, paragraph after Eq. (10) and Fig. 2"},{"comment":"There is a numerical inconsistency in the Rb D2 calculations: the text states that the low-density calculation uses r_L = 20 and the high-density calculation uses r_L = 10^5, while the Fig. 12 caption reports r_L = 100 and r_L = 10^4, respectively. This discrepancy directly affects the claimed consistency with the Rb measurements of Refs. [11–13] and must be corrected and unified.","section":"§4.3 and Fig. 12 caption"},{"comment":"The text first reports that Ref. [13] obtained α = 1.03 ± 0.15 for system sizes up to r_L ∼ 10^5 and later states that at r_L ∼ 10^5 “α < 1 was measured.” These two statements need to be reconciled or rephrased to describe the actual r-dependence of the reported α, since as written they appear contradictory.","section":"§4.3, paragraph on Rb measurements"}],"minor_comments":[{"comment":"The definitions of r and r_L should be stated together explicitly: r is the step length in units of the line-center mean free path and r_L is the sample size in the same units; the current text introduces them in neighboring sentences but would benefit from a single clarifying sentence.","section":"§2, Eq. (3)"},{"comment":"The caption writes “rl ∼ 100” for the finite-size cases; this should read “r_L = 100” (and similarly for the other values) to match the notation used in the text.","section":"Fig. 3 caption"},{"comment":"The recurrence for Θ_n no longer preserves normalization when [1−T(x)] is included. The authors should state explicitly that the normalization constant drops out of the logarithmic derivative defining α(r), so that the plotted P(r) and α(r) curves are unaffected by this overall factor.","section":"§3, Eq. (9)"},{"comment":"Several typographical errors remain, including “scaterring” (p. 8), “diferent” (Fig. 4 caption), and the non-standard spacing in “L´evy flight” in the title; these should be corrected in the final version.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of JQSRT and the topic is timely. The main concern is that the finite-size approximation in Eq. (9) is load-bearing but is validated only for the scattered spectrum, not for the step-size distribution and α(r) that the paper headlines. The authors already have a Monte Carlo code that can produce the missing benchmark, so this should be feasible in revision. The RIII finite-size discussion also needs a numerical check, and the Rb r_L inconsistency should be fixed. I do not see grounds for rejection if these points are adequately addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this paper does something genuinely useful: it adds a finite-size cutoff to the recurrence that generates scattered spectra in partial frequency redistribution, and it works out the multilevel hyperfine structure for Cs and Rb. The resulting α(r) curves show the expected qualitative behavior—finite size raises α by truncating large steps, and hyperfine peaks superimpose oscillations. Second, the load-bearing piece—the [1−T(x)] factor in Eq. 9—is a physically motivated guess, and the Monte Carlo check only validates the spectra, not the step-size distributions or α(r) that the paper headlines.\n\nThe new content is real. Previous calculations (Pereira, Alves-Pereira, Mercadier, Lopez) were all infinite-medium, two-level. The finite-size factor, justified by the single-big-jump principle, is a natural extension. The multilevel treatment is systematic: the sum over hyperfine transitions in Eq. 11, the absorption profile in Eq. 12, and the decomposition of the Cs D1 oscillations by selectively keeping transitions is instructive. The paper is clearly written, parameters are specified, and the comparison with experiment is appropriately qualitative.\n\nThe soft spot is not fatal, but it is real. The factor 1−exp(−ϕ(x) r_L) treats escape as a Beer-Lambert survival probability for a single flight through a slab of length L. The Monte Carlo, by contrast, uses a cylindrical cell and checks at each step whether the photon escapes. The agreement in Fig. 1 for the scattered spectra is encouraging—it shows the escape probability is not wildly wrong—but it is not a direct test of P(r) or α(r). The step-size distribution is a functional of Θ(x), so the spectra agreement is nontrivial, but the geometry mismatch and the fact that the MC uses the same scattering rules mean the validation is indirect. I'd like to see a plot of the MC step-size distribution overlaid on Fig. 3(a). That would settle it.\n\nThe D1 section is purely theoretical; no experiment exists for that line. The Rb comparison uses different Γ, Γ_D, and r_L than the calculations, so it's consistency, not confirmation.\n\nWho is this for? The atomic-vapor Lévy-flight community and people doing radiative transfer in optically thick media. It's a solid extension, not a paradigm shift. I'd send it to a referee—with the request to add the direct step-size benchmark. If that comes out clean, the paper is definitely publishable.","headline":"A clean extension of the infinite-medium Lévy-flight picture to finite vapors and alkali hyperfine structure, but the finite-size cutoff rests on a factor that the Monte Carlo only partially validates.","tokens_in":14491,"tokens_out":3993,"would_cite":true,"duration_ms":35391,"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":"Finite vapor size clips the far-wing photons that make resonant light transport a Lévy flight, raising the effective Lévy exponent and adding hyperfine-structure oscillations to its step-size dependence.","keywords":["photon Lévy flight","frequency redistribution","partial frequency redistribution","step-size distribution","finite-size effects","alkali hyperfine structure","resonant atomic vapor"],"falsifier":"The finite-size claim could be settled by a photon-by-photon random-walk simulation that tracks each photon's position, draws free paths from the exponential distribution $N\\sigma(\\delta)e^{-N\\sigma(\\delta)l}$, and stops a photon only when it crosses an absorbing boundary, without ever inserting the $[1-T(x)]$ factor; if the resulting spectrum at scattering event $n=10$ or the step-size distribution differs systematically from the paper's finite-vapor curves, the escape-factor approximation is not capturing the boundary. Experimentally, the same test is to measure the step-size distribution or transmitted spectrum in the same vapor at two different cell thicknesses and check that the $\\alpha(r)$ increase appears at the $r$ value predicted by $r_L$.","tokens_in":13443,"feed_emoji":"⚛️","tokens_out":12218,"duration_ms":102369,"temperature":0.7,"pith_summary":"The paper claims that two features of real resonant vapors change the Lévy-flight description of multiply scattered light: the finite size of the vapor and the multi-level hyperfine structure of alkali atoms. It computes the frequency redistribution function for partial frequency redistribution and uses a recurrence to evolve the scattered spectrum from one scattering event to the next, then derives the step-size distribution and a local Lévy exponent $\\alpha(r)$. The central result is that a finite vapor cuts off the far-wing frequencies at each scattering event, truncating the largest steps and making $\\alpha(r)$ larger than in an infinite vapor, while the hyperfine and isotope structure of Cs and Rb imprints oscillations on $\\alpha(r)$. These effects matter because measured $\\alpha$ values depend on cell size and level structure; the paper connects its calculations to existing Cs and Rb experiments.","feed_headline":"Atomic-vapor size clips the heavy tails of photon Lévy flights","feed_subtitle":"Step-size distributions for Cs and Rb show the finite-cell cutoff and hyperfine oscillations that experiments should see.","key_machinery":"The object that carries the argument is the recurrence for the scattered spectrum after scattering event $n$: $\\Theta_n(x') = \\int dx\\, [1-T(x)]\\,\\Theta_{n-1}(x)\\, R(x,x')/\\phi(x)$, where $R(x,x')$ is the frequency redistribution function (RII for coherent scattering in the atomic rest frame, RIII for complete redistribution in the atomic rest frame), $\\phi(x)$ is the normalized Voigt absorption profile, and $T(x)=\\exp[-\\phi(x)r_L]$ is the single-pass ballistic transmission through a sample of dimensionless size $r_L$. The factor $[1-T(x)]$ is the finite-size innovation, justified by the Single-Big-Jump principle rather than by a boundary-condition solution of the random walk. The step-size distribution is then $P(r)=\\int dx\\, \\Theta(x)\\,\\phi(x)\\,e^{-\\phi(x)r}$, and the reported Lévy exponent is the local slope $1+\\alpha(r) = -d\\log P/d\\log r$. For alkali vapors the same machinery is summed over hyperfine transitions with relative strengths and detunings, producing the side peaks and oscillations.","core_discovery":"The paper's central claim is that the step-size distribution of light scattered in atomic vapor is not a fixed power law: the local Lévy exponent, defined by $1+\\alpha(r) = -d\\log P(r)/d\\log r$, varies with step size and is controlled by the vapor's size and level scheme. For the RII redistribution case (coherent scattering in the atomic rest frame), the recurrence for the scattered spectrum contains a factor $[1-T(x)]$ with $T(x)=\\exp[-\\phi(x)r_L]$, so photons emitted far in the wings, which would produce the longest steps, are preferentially removed from subsequent scattering in a finite sample. This shifts the cutoff in $P(r)$ to smaller $r$ and raises $\\alpha(r)$ relative to an infinite vapor; for RIII (complete redistribution in the atomic rest frame) no such shift occurs because the emitted frequency does not depend on the incident frequency. In alkali vapors, summing the same recurrence over hyperfine transitions produces side peaks spaced by ground-state hyperfine splittings, which create oscillations in $\\alpha(r)$ in the range $r\\sim 10^3$--$10^6$ depending on the line; the paper reports that the calculated values are consistent with previous Cs and Rb measurements.","pith_inferences":["The paper leaves implicit that the same escape-factor recurrence could be used to reinterpret existing transmission measurements made in cells of different thickness; if the finite-size effect is real, a thin cell at fixed density should show a higher effective $\\alpha$ than a thick cell.","Because the Monte Carlo validation in the paper uses the same scattering rules and escape criterion as the recurrence, an independent photon-tracking simulation with an explicit absorbing boundary would be a stronger test of the finite-size approximation.","The oscillations in $\\alpha(r)$ imply that a single-number Lévy exponent is not enough to characterize photon transport in alkali vapors; experiments and radiative-transfer models should report the full step-size distribution or the scale-dependent exponent.","The same finite-size truncation may appear in astrophysical resonance-line transfer, where the RII wing approaches the RIII Lorentzian wing after many scatterings; this is a direct extension of the paper's two-level result to very large systems."],"forward_implications":["In a finite two-level vapor described by RII, the Lévy exponent extracted at a given step size is systematically larger than in an infinite vapor, and the lowest values of $\\alpha$ near 0.3 are not reached for $r_L \\sim 100$.","For RIII, the finite size of the vapor leaves the scattered spectrum and the step-size distribution unchanged, so the finite-size cutoff does not affect the Lévy exponent.","For alkali vapors, hyperfine ground-state structure, resolved excited levels (Cs D1), and the two Rb isotopes introduce oscillations in $\\alpha(r)$ whose position depends on the line and on vapor density.","The calculations are consistent with measured Lévy exponents in Cs and Rb, including the trend from $\\alpha \\approx 0.8$ at $r\\sim 400$ to $\\alpha \\approx 0.6$ at $r\\sim 4\\times 10^4$ for Cs D2.","With RII and a finite vapor, the scattered spectrum after many events becomes independent of whether the incident spectrum is a delta function or a Lorentzian, because wing photons that would preserve the incident shape are removed by the escape factor."],"supporting_citations":[{"why":"Supplies the Single-Big-Jump principle used to justify replacing the sample boundary with the ballistic escape factor $1-\\exp[-\\phi(x)r_L]$.","marker":"[17]"},{"why":"Provides the RII and RIII frequency redistribution functions $R(x,x')$ used in the recurrence for the scattered spectrum.","marker":"[21]"},{"why":"Gives the baseline two-level Lévy exponents under complete frequency redistribution: $\\alpha=0.5$ for Lorentz and Voigt profiles and $\\alpha\\approx 1$ for a Doppler profile.","marker":"[10]"},{"why":"Introduces the partial-frequency-redistribution recurrence for the emitted spectrum in an infinite vapor, which this paper extends to finite size.","marker":"[18]"},{"why":"Applies the recurrence to microscopic step-size measurements in Rb and supplies the Lorentzian incident-spectrum convention used in the alkali calculations.","marker":"[12]"},{"why":"Describes the Monte Carlo scattering algorithm used to validate the computed scattered spectra for infinite and finite vapor.","marker":"[19]"},{"why":"Reports Cs transmission measurements showing that the Lévy exponent depends on system size, the experimental trend the finite-vapor calculation addresses.","marker":"[15]"},{"why":"Reports step-size-dependent Lévy exponents and transmission measurements in Cs, used as a consistency check for the calculated $\\alpha(r)$ curves.","marker":"[16]"},{"why":"Provides the measured Rb Lévy exponent $\\alpha=1.09\\pm 0.15$ in the single-scattering regime, compared with the low-density Rb calculation.","marker":"[11]"},{"why":"Supplies the exponential free-path step-size distribution $N\\sigma e^{-N\\sigma l}$ that underlies Eq. 1 and all later $P(r)$ calculations.","marker":"[26]"}],"fun_headline_variants":["Vapor size clips heavy tails of photon Lévy flights","Finite vapor cuts longest steps in atomic light diffusion","Vapor size alters photon Lévy flight step-size law","Hyperfine structure creates oscillations in Lévy flight steps","Photon steps in atomic vapor: size and levels matter"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a photon's chance of being scattered one more time is given by the single-pass transmission factor $1-\\exp[-\\phi(x)r_L]$—the probability it does not cross the whole sample without absorption—and that this same factor remains accurate after many scattering events and for multilevel atoms; if that factor is not right, the finite-size cutoff and the predicted $\\alpha(r)$ curves do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Vapor size clips heavy tails of photon Lévy flights","Finite vapor cuts longest steps in atomic light diffusion","Vapor size alters photon Lévy flight step-size law","Hyperfine structure creates oscillations in Lévy flight steps","Photon steps in atomic vapor: size and levels matter"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000944,"raw_usage":{"total_tokens":4060,"prompt_tokens":1000,"completion_tokens":3060,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":616,"completion_tokens_details":{"reasoning_tokens":2981}},"tokens_in":616,"tokens_out":3060,"duration_ms":22298,"temperature":1.0,"reasoning_tokens":2981,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:05:03.012429+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The finite-size claim could be settled by a photon-by-photon random-walk simulation that tracks each photon's position, draws free paths from the exponential distribution $N\\sigma(\\delta)e^{-N\\sigma(\\delta)l}$, and stops a photon only when it crosses an absorbing boundary, without ever inserting the $[1-T(x)]$ factor; if the resulting spectrum at scattering event $n=10$ or the step-size distribution differs systematically from the paper's finite-vapor curves, the escape-factor approximation is not capturing the boundary. Experimentally, the same test is to measure the step-size distribution or transmitted spectrum in the same vapor at two different cell thicknesses and check that the $\\alpha(r)$ increase appears at the $r$ value predicted by $r_L$.","supporting_citations":[{"cited_title":"Pereira, J","cited_arxiv_id":null,"evidence_quote":"Gives the baseline two-level Lévy exponents under complete frequency redistribution: $\\alpha=0.5$ for Lorentz and Voigt profiles and $\\alpha\\approx 1$ for a Doppler profile."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the partial-frequency-redistribution recurrence for the emitted spectrum in an infinite vapor, which this paper extends to finite size."},{"cited_title":"Mercadier, M","cited_arxiv_id":null,"evidence_quote":"Applies the recurrence to microscopic step-size measurements in Rb and supplies the Lorentzian incident-spectrum convention used in the alkali calculations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the Monte Carlo scattering algorithm used to validate the computed scattered spectra for infinite and finite vapor."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports Cs transmission measurements showing that the Lévy exponent depends on system size, the experimental trend the finite-vapor calculation addresses."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports step-size-dependent Lévy exponents and transmission measurements in Cs, used as a consistency check for the calculated $\\alpha(r)$ curves."},{"cited_title":"Mercadier, W","cited_arxiv_id":null,"evidence_quote":"Provides the measured Rb Lévy exponent $\\alpha=1.09\\pm 0.15$ in the single-scattering regime, compared with the low-density Rb calculation."},{"cited_title":"Holstein, Imprisonment of resonance radiation in gases, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the exponential free-path step-size distribution $N\\sigma e^{-N\\sigma l}$ that underlies Eq. 1 and all later $P(r)$ calculations."}],"review_version":1}