{"id":"3e7295e8-5638-4782-b123-35da4edb703b","arxiv_id":"2611.16325","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"An intermediate number of discrete steps in an antibiotic gradient minimizes the time for bacteria to evolve full resistance, with the optimum scaling as (r^2 / μ v K0)^{1/3}.","lead":"This paper studies how the number of discrete steps in a spatial antibiotic gradient affects the time bacteria take to evolve full resistance. It finds that an intermediate number of steps minimizes the adaptation time, revealing a trade-off between faster search and repeated population saturation.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"Analytical FAT derivation uses a constant-population approximation that is not tested against simulations for the parameter regime where the optimum is predicted.","rationale":"The reader's verdict is CONDITIONAL with high confidence, citing the limited simulation test (six parameter sets, k ≤ 10) and the reliance on approximations. The reader's weakest_assumption is the sequential-adaptation assumption (no clonal interference). While that is a valid concern, the paper explicitly acknowledges this regime and states it is not considered. My stress-test focuses on a different, more concrete load-bearing concern: the analytical derivation itself is tested only in a narrow parameter range, and the approximations (constant population, Laplace method, saturation time) are not validated in the regime where the scaling law for k* is predicted. The paper's own Fig. 4 shows that the approximate analytical expression (orange dashed) systematically overestimates FAT compared to the numerically evaluated theory (blue solid) and the Gillespie simulations, with discrepancies up to 50%. The paper attributes this to the Laplace approximation being marginal (a ~ 4.5) for some parameter sets. However, the scaling law (Fig. 5) and the speedup map (Fig. 3) are based entirely on the approximate analytical expression, not on the numerically evaluated theory or simulations. The paper does not test whether the approximations affect the location of the minimum. The concrete test I propose would directly check whether the predicted optimum exists and scales as predicted in the regime where the theory is claimed to apply. This is a concrete, feasible computational test that would settle whether the central claim is robust. I partially agree with the reader because the reader's weakest_assumption (sequential adaptation) is a different concern, but my concern is more specific and testable. The verdict should remain CONDITIONAL because the paper's central claim is not yet fully validated, but the condition is more specific: the analytical derivation needs to be tested in the regime where the scaling law is predicted, not just in the limited simulation range shown.","tokens_in":20344,"tokens_out":1104,"duration_ms":11550,"concrete_test":"Perform Gillespie simulations for at least three parameter sets where the analytical expression predicts a clear interior optimum with k* > 10 (e.g., from the regime of Fig. 5, with r=1, d=0.1, K0=10^9, μ=10^{-9}, v=10^{-6}, giving r^2/(μvK0) ~ 10^6, predicted k*+1 ~ 100). For each set, simulate FAT(k) for k = 1, 5, 10, 20, 50, 100, 200 (or up to kmax) with at least 300 independent trials per k. Compare the simulated FAT values to the analytical Eq. (18) and to the numerically evaluated theory (blue curve in Fig. 4). If the simulated minimum occurs at a k that differs from the analytical prediction by more than a factor of 2, or if no minimum is observed, then the analytical derivation is not reliable in the regime where the scaling law is claimed to hold.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central result—the existence and scaling of an optimal number of steps k*—rests on the analytical expression for FAT(k), Eq. (18). This expression is derived under several approximations: (i) the population at each step is treated as constant at its saturated size N (Eq. (5)), neglecting demographic fluctuations; (ii) the Laplace approximation (a ≫ 1) is used to evaluate the adaptation time T_adapt; (iii) the saturation time T_sat is derived from a birth–death process that neglects mutation and approximates the inverse generator by keeping only the leading term (Appendix C, Eq. (C6)). The paper tests Eq. (18) against Gillespie simulations for only six parameter sets, with k ≤ 10 and a limited range of parameters (Figs. 4a–f). In these tests, the approximate analytical expression systematically overestimates FAT by up to 50% (ratio 1.0–1.5). More importantly, the scaling law for k* (Eq. (24)) and the speedup map (Fig. 3) are derived entirely from Eq. (18) and are not validated by simulations in the regime where the optimum is predicted (e.g., k* > 10, large K0, small μ, v). The paper acknowledges that the Laplace approximation parameter a is only ~4.5 at k=1 for the parameter sets in Figs. 4(a)–(c), which is marginal for a ≫ 1. The paper also notes that the saturation time approximation neglects paths where the population size decreases (Appendix C), and that this approximation is not uniformly valid for i ∼ N. These approximations are load-bearing because the existence and location of the optimum depend on the balance between the search time (∼ k^{-1/2}) and the saturation time (∼ k ln(K0/k)). If the approximations systematically bias either term, the predicted optimum could shift or disappear. The paper does not provide a sensitivity analysis or error bounds on Eq. (18).","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"The paper studies a stochastic model of bacterial adaptation in a discretized antibiotic gradient, treating the number of steps k as a control parameter while keeping overall spatial and concentration ranges fixed. The authors derive an approximate analytical expression for the mean first-arrival time (FAT) to the fully resistant state, decomposing it into adaptation and saturation times. They find that FAT(k) can be minimized at an intermediate k* due to a trade-off between faster search per step (via increased mutation and migration rates) and the cumulative cost of repeated population saturation. The scaling k*+1 ~ (r^2/(μ v K0))^{1/3} is derived in an asymptotic regime. The analytical predictions are compared with numerical evaluation of the exact expressions and with Gillespie simulations for a limited set of parameters (k ≤ 10), showing qualitative agreement. The paper also presents a speedup map and a scaling plot based on numerical minimization of the analytical expression.","tokens_in":20766,"tokens_out":2426,"duration_ms":19590,"significance":"The paper addresses a timely question in evolutionary dynamics: how the discretization of a spatial antibiotic gradient affects the speed of resistance evolution. The central idea—that an intermediate number of steps can optimize adaptation time—is novel and potentially testable. The analytical derivation is detailed and self-contained, with careful treatment of the scaling of rates to keep end-to-end first-passage times fixed. The paper provides a falsifiable scaling prediction (Eq. 24) and identifies the key dimensionless parameter controlling the optimum. If the result holds under more realistic conditions (fitness costs, asymmetric rates, spatial structure), it could inform the design of antibiotic landscapes to slow resistance evolution. The authors are transparent about limitations and assumptions.","major_comments":[{"comment":"The validation of the analytical FAT expression against Gillespie simulations is limited to six parameter sets with k ≤ 10. The paper's central claim—the existence and scaling of an optimal k*—relies on the analytical expression (Eq. 18) for larger k (k* can exceed 10, as seen in Fig. 5). The simulations do not probe the regime where the optimum is predicted (e.g., k* > 10, large K0, small μ, v). The authors acknowledge this implicitly by using the analytical expression for the scaling plot (Fig. 5) and the speedup map (Fig. 3), but the lack of direct simulation support for the optimum at larger k is a significant gap. The paper should either provide simulation evidence for at least one parameter set where k* > 10, or explicitly discuss why the analytical expression is expected to remain reliable in that regime despite the approximations (constant N, Laplace approximation, neglect of clonal interference).","section":"§IV D 1, Fig. 4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The Laplace approximation used to derive T_adapt (Eq. 13) assumes a ≫ 1, where a = μ_k N / (r' - μ_k). The paper notes that at k=1 for the parameter sets in Figs. 4(a)–(c), a ≃ 4.5, which is marginal. While the discrepancy between the numerical and approximate T_adapt is only 5.4% in those cases, the approximation is used to derive the scaling law (Eq. 24) and the speedup map (Fig. 3). The paper should quantify how the error in T_adapt propagates to the predicted k* and the scaling exponent, especially for parameter sets where a is not large (e.g., a ~ 10). A sensitivity analysis or a bound on the error would strengthen the claim.","section":"§III A, Eq. (13); §IV D 1, Fig. 4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The derivation of T_sat (Eq. 16) neglects paths where the population size decreases (Appendix C, Eq. C6). The authors state that this approximation is not uniformly valid for i ∼ N but claim the effect on T'_sat is small based on numerical checks. However, no numerical evidence is provided in the manuscript. Given that T_sat contributes significantly to FAT for larger k (the saturation term grows as k ln(K0/k)), the accuracy of this approximation should be tested, at least for a representative parameter set. The paper should include a comparison of the approximate T_sat with the exact numerical solution of the birth–death process (or with simulation) for a few cases.","section":"§III B, Appendix C","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The asymptotic criterion for an interior optimum (Eq. 20) and the scaling law (Eq. 24) are derived under the assumption v k^2 ≪ d and d ≪ r. The paper then uses these to produce Fig. 5, which shows agreement between the numerical minimization of Eq. (18) and the 1/3 power law. However, the parameter sets in Fig. 5 are filtered to satisfy k* < 0.9 k_max, which ensures v (k*)^2 ≪ d is not necessarily satisfied for all points (since k_max ~ sqrt(2d/v), so k* < 0.9 sqrt(2d/v) implies v (k*)^2 < 1.62 d, not necessarily ≪ d). The paper should clarify the range of validity of the scaling and discuss whether deviations from the 1/3 slope are expected when v (k*)^2 is not negligible compared to d.","section":"§IV A, Eq. (20); §IV B, Eq. (24)","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The scaling relations v_k = k(k+1) v / 2 and μ_k = k(k+1) μ / 2 are derived from the first-passage time of a single bacterium. It would be helpful to note explicitly that this scaling assumes symmetric nearest-neighbor transitions and that the end-to-end first-passage time is τ_k = k(k+1)/(2 v_k). The derivation in Appendix B is clear, but a brief comment in the main text would improve readability.","section":"§II, Eq. (1) and (2)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The master equation (8) uses r' = d + 2 v_k and δ = d + 2 v_k + μ_k. The notation is a bit confusing: r' is not a division rate but a loss rate. Consider renaming to avoid confusion with the division rate r.","section":"§III A, Eq. (8)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The expression for T_sat includes a term C T_adapt, where C is defined in Eq. (D11). The constant C is derived in the limit K0 → ∞. The paper should note that for finite K0, C may differ, and quantify the error if possible.","section":"§III B, Eq. (17)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The figure caption says the blue solid curve is the result of computing Eq. (18) with Eqs. (12) and (14) evaluated by numerical integration. But Eq. (18) already includes the Laplace approximation; the numerical integration replaces the Laplace approximation for T_adapt and T_readapt. The description is slightly ambiguous. Clarify that the blue curve uses numerical integration of the exact survival probability (Eq. 9) to compute T_adapt and T_readapt, while the orange curve uses the approximate closed forms (Eqs. 13 and 15).","section":"§IV D 1, Fig. 4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The figure shows the scaling of k*+1 vs r^2/(μ v K0) with a fitted line of slope 1/3. The intercept is not given; providing the fitted constant A in Eq. (24) would be useful for reproducibility. Also, the parameter range for Fig. 5 is K0 ∈ (10^9, 10^10), which is larger than the K0 used in the simulations (10^6–10^8). This is fine, but the paper should note that the scaling is tested only in the regime of very large K0.","section":"§IV D 2, Fig. 5","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The discussion of the geometric mean in Appendix F is helpful. However, the AND circuit model assumes that the intermediate state is reached before the goal, which corresponds to the sequential mutation–migration process. The paper should note that this is an idealization; in the actual model, mutation and migration can occur in any order, but the dominant path is mutation first then migration (as argued in Ref. [5]).","section":"§V A, Appendix F","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The paper uses the term 'FAT' for first-arrival time. This is fine, but it might be confused with 'fitness-associated trait' in some contexts. Consider spelling out 'first-arrival time' at first use and using 'FAT' thereafter.","section":"General","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Figure 1 is a schematic of the staircase model. It would be helpful to label the axes more clearly (e.g., 'Patch i' and 'Phenotype p') and to indicate the direction of increasing antibiotic concentration.","section":"§II, Fig. 1","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The error bars for the Gillespie simulations are said to be mostly smaller than the plotting symbols. It would be useful to report the standard deviation or coefficient of variation of the FAT across trials, as the standard error of the mean can be small due to 300 trials even if the distribution is broad.","section":"§IV D 1, Fig. 4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The limitations section is thorough. One additional limitation worth mentioning is the assumption of symmetric mutation and migration rates; the paper does mention this. Also, the assumption that the population at each step reaches local saturation before the next adaptation occurs (sequential adaptation) is central and is stated, but the paper could briefly discuss the conditions under which this assumption might break down (e.g., high mutation supply, clonal interference) and how the results would change.","section":"§V B","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision"},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"Colleague,\n\nThis paper extends the Hermsen et al. staircase model by treating the number of steps k as a control parameter, holding the overall spatial and concentration ranges fixed, and scaling migration and mutation rates to keep end-to-end first-passage times constant. The central result is that the mean first-arrival time (FAT) can be minimized at an intermediate k, with a scaling law k*+1 ~ (r^2/(μ v K0))^{1/3}. This is a clean theoretical extension and the trade-off between search time (which decreases with k) and saturation cost (which increases with k) is physically intuitive.\n\nThe paper does several things well. The analytical derivation is laid out in detail, with appendices covering the first-passage time calculations, the saturation time, and the Laplace approximation. The authors are upfront about their assumptions: sequential adaptation, strong-selection weak-mutation regime, constant population size approximation, and the regime conditions r > d+2v_k and d > v_k >> μ_k. They also discuss limitations—fitness costs, asymmetric rates, well-mixed patch assumption—in a dedicated section. The scaling analysis is internally consistent: the numerical minimization of the analytical expression (Eq. 18) over a broad parameter range yields points that fall on the predicted 1/3 slope (Fig. 5).\n\nThe soft spots are in the validation. The Gillespie simulations cover only six parameter sets with k up to 10, and the analytical expression overestimates FAT by up to 50% in those tests. The scaling law for k* is derived from and compared to the same approximate analytical expression, not to independent simulations or data. The paper acknowledges that the Laplace approximation parameter a is only ~4.5 at k=1 for some parameter sets, which is marginal for a >> 1. The stress-test note about the constant-population approximation not being tested in the regime where the optimum is predicted (k* > 10, large K0, small μ, v) is a fair concern. The paper does not provide a sensitivity analysis or error bounds on Eq. 18.\n\nWho is this for? Researchers in evolutionary dynamics who work on spatial structure and antibiotic resistance. The paper is a solid theoretical contribution within an established program. It deserves a serious referee—the math is careful, the assumptions are stated, and the result is nontrivial. But the referee should push for more extensive simulation validation, especially in the regime where the optimum is predicted, and for a discussion of how the approximations might affect the predicted scaling.\n\nRecommendation: send to peer review. The paper is not ready as is—the simulation support is too thin—but the theoretical core is worth engaging with.","headline":"Solid theoretical extension of the staircase model with a clear trade-off and scaling law; the simulation validation is narrower than the claims warrant.","tokens_in":21365,"tokens_out":646,"duration_ms":10359},"pith_extraction":{"msc":["92D15","92D25","60J70","82C31"],"pacs":["87.23.Cc","87.23.Kg","87.10.Mn","05.40.-a"],"model":"deepseek/deepseek-v4-flash","headline":"Staircase of antibiotic steps can accelerate bacterial evolution—up to a point","keywords":["antibiotic resistance","staircase model","evolutionary adaptation","first-passage time","optimal discretization","spatial heterogeneity","mutation-migration trade-off","Gillespie simulation"],"falsifier":"A direct Gillespie simulation or microfluidic experiment with a parameter set where v $k^{2}$ is comparable to d (violating the v $k^{2}$ << d condition) would show no minimum in FAT(k) or a minimum at k=1, contradicting the predicted optimum. Alternatively, if clonal interference is present (e.g., high mutation supply), the sequential decomposition fails and the optimum may disappear.","tokens_in":21721,"feed_emoji":"🦠","tokens_out":1648,"duration_ms":13223,"temperature":0.7,"pith_summary":"This paper studies how the number of discrete steps in a spatial antibiotic gradient affects the time it takes for a bacterial population to evolve full resistance. The authors extend a known staircase model to treat the number of steps as a control parameter, while holding the total spatial extent and concentration range fixed. They derive an approximate analytical expression for the mean first-arrival time at the fully resistant state, decomposing it into a search time for each new adapted mutant and a saturation time for the population to fill each patch. The competition between these two contributions—faster per-step search when steps are many, but more saturation events—produces an optimal intermediate number of steps that minimizes the total adaptation time. The optimum scales as k*+1 ~ ($r^{2}$/(μ v K0))^{1/3}, where r is the division rate, μ and v are mutation and migration rates, and K0 is the total carrying capacity. Direct Gillespie simulations support the predicted dependence, including the location of the optimum.","feed_headline":"Optimal staircase structure for bacterial adaptation in antibiotic landscapes","feed_subtitle":"How the number of discrete steps in a drug gradient controls the speed of resistance evolution","key_machinery":"The central object is the staircase model on a two-dimensional lattice (physical patch i, resistance phenotype p), where bacteria migrate along the physical axis and mutate along the phenotype axis, with division possible only when i ≤ p. The key machinery is the decomposition of the adaptation time into per-step search time (first-passage time for a mutant to appear and migrate) and per-step saturation time (time for the population to grow to carrying capacity), and the scaling relations v_k = k(k+1)v/2 and μ_k = k(k+1)μ/2 that keep the mean single-bacterium first-passage times independent of k when the number of steps k is varied. The competition between the search time (which decreases as $k^{{-1/2}}$) and the saturation time (which increases as k ln(K0/k)) yields the optimum.","core_discovery":"The paper claims that the mean first-arrival time (FAT) for a bacterial population to evolve full resistance across a discretized antibiotic gradient can be minimized at an intermediate number of steps k*, and this optimum scales as k*+1 ~ ($r^{2}$/(μ v K0))^{1/3} in the asymptotic regime where v $k^{2}$ << d and d << r. The discovery is that the competition between faster per-step search (due to increased mutation and migration rates when the gradient is more finely resolved) and the cumulative cost of repeated population saturation can produce a non-monotonic FAT, with a minimum at an intermediate resolution. This is an extension of the staircase model, which previously treated the number of steps as given, to treat k as a control parameter, revealing an optimal discretization.","pith_inferences":["The result suggests that in natural or clinical antibiotic gradients, the spatial discretization (e.g., patchiness of drug concentration) could be an underappreciated factor controlling the speed of resistance evolution. If the gradient is too fine-grained, the repeated population saturation cost dominates; if too coarse, the search for the next mutant is slow.","The scaling law might be testable in microfluidic devices where the number of discrete concentration steps can be engineered, providing a direct experimental test of the trade-off between search and saturation.","The assumption that the population saturates locally before the next adaptation step (strong-selection weak-mutation regime) is the load-bearing premise; if clonal interference or multiple mutations arise, the decomposition fails and the optimum may shift or disappear."],"forward_implications":["If the optimal staircase exists, then antibiotic gradients with an intermediate number of discrete concentration steps should produce faster resistance evolution than either a single step (sharp boundary) or a very finely resolved gradient.","The scaling law k*+1 ~ (r^2/(μ v K0))^{1/3} provides a testable prediction: the optimal number of steps decreases when the product μ v K0 is large (fast adaptation) and increases when the intrinsic growth rate r is large.","The analytical expression for FAT at k=1 can be used to estimate the effective mutation rate μ from experimentally measured adaptation times, if the other parameters are independently known.","The model predicts that the speedup ratio FAT(1)/FAT(k*) can be substantial (up to ~5 in the explored parameter range), implying that staircase design could be relevant for understanding and possibly controlling resistance evolution in spatially structured environments."],"supporting_citations":[{"why":"Hermsen and Hwa 2010: Introduces the stochastic source–sink model that the staircase model builds on.","marker":"[5]"},{"why":"Hermsen, Deris, and Hwa 2012: Introduces the staircase model and shows that a concentration gradient accelerates adaptation; the present paper extends that model.","marker":"[6]"},{"why":"Gillespie 1983, 1984: Provides the strong-selection weak-mutation regime assumption that underlies the sequential-adaptation decomposition.","marker":"[10, 11]"},{"why":"Sniegowski and Gerrish 2010, Desai and Fisher 2007: Discuss clonal interference, which the paper explicitly excludes from its regime.","marker":"[13, 14]"},{"why":"Sekimoto 2021: Provides the method for first-passage time of population growth used in the saturation time derivation.","marker":"[15]"},{"why":"Sezonov et al. 2007: Provides the intrinsic division rate r for E. coli K-12 used as a reference parameter.","marker":"[16]"},{"why":"Kurzthaler et al. 2024: Provides the measured effective diffusion coefficient for E. coli used to relate migration rate v to experimental quantities.","marker":"[17]"},{"why":"Lee et al. 2012: Provides the measured spontaneous mutation rate for E. coli, though the paper notes that μ is an effective transition rate between resistance phenotypes, not the total mutation rate.","marker":"[18]"}],"fun_headline_variants":[],"cache_read_input_tokens":0,"weakest_assumption_plain":"The population must reach local saturation at each step before the next adaptation occurs (sequential adaptation), which requires that mutations are rare compared to migration and that clonal interference is absent; the paper states that this regime, known as strong-selection weak-mutation, is assumed and that clonal interference is not considered.","fun_headline_variants_meta":{"raw":{"variants":[]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":8.3e-05,"raw_usage":{"cost":8.3083e-05,"is_byok":false,"cost_details":{"upstream_inference_cost":8.3083e-05,"upstream_inference_prompt_cost":8.2719e-05,"upstream_inference_completions_cost":3.64e-07},"total_tokens":911,"prompt_tokens":909,"completion_tokens":2,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":0,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":0}},"tokens_in":909,"tokens_out":2,"cost_source":"openrouter_usage","duration_ms":4890,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":0,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-13T06:36:06.342962+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"A direct Gillespie simulation or microfluidic experiment with a parameter set where v $k^{2}$ is comparable to d (violating the v $k^{2}$ << d condition) would show no minimum in FAT(k) or a minimum at k=1, contradicting the predicted optimum. Alternatively, if clonal interference is present (e.g., high mutation supply), the sequential decomposition fails and the optimum may disappear.","supporting_citations":[{"cited_title":"Sources and sinks: a stochastic model of evolution in heterogeneous environments,Phys. Rev. Lett.105, 248104","cited_arxiv_id":null,"evidence_quote":"Hermsen and Hwa 2010: Introduces the stochastic source–sink model that the staircase model builds on."},{"cited_title":"On the rapidity of antibiotic resistance evolution facilitated by a concentra- tion gradient,Proc. Natl. Acad. Sci. U.S.A.109, 10775","cited_arxiv_id":null,"evidence_quote":"Hermsen, Deris, and Hwa 2012: Introduces the staircase model and shows that a concentration gradient accelerates adaptation; the present paper extends that model."},{"cited_title":"Derivation of the First Passage Time Distribution for Markovian Process on Discrete Network","cited_arxiv_id":"2110.02216","evidence_quote":"Sekimoto 2021: Provides the method for first-passage time of population growth used in the saturation time derivation."},{"cited_title":"Escherichia coliphysiology in Luria–Bertani broth,J. Bacteriol.189, 8746","cited_arxiv_id":null,"evidence_quote":"Sezonov et al. 2007: Provides the intrinsic division rate r for E. coli K-12 used as a reference parameter."},{"cited_title":"Kurzthaler et al., Characterization and control of the run-and-tumble dynamics ofEscherichia coli,Phys","cited_arxiv_id":null,"evidence_quote":"Kurzthaler et al. 2024: Provides the measured effective diffusion coefficient for E. coli used to relate migration rate v to experimental quantities."},{"cited_title":"Rate and molecular spectrum of spontaneous mutations in the bac- teriumEscherichia colias determined by whole-genome sequencing,Proc. Natl. Acad. Sci. U.S.A.109, E2774","cited_arxiv_id":null,"evidence_quote":"Lee et al. 2012: Provides the measured spontaneous mutation rate for E. coli, though the paper notes that μ is an effective transition rate between resistance phenotypes, not the total mutation rate."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}