{"id":"d987317b-1d41-4501-9f98-b39731e0744a","arxiv_id":"2606.20069","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":4.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":1,"one_line_summary":"Develops a double sequential sampling framework for two-sample tests on shifted exponential location parameters with unknown unequal scales, achieving first- and second-order asymptotic efficiencies.","lead":"The paper proposes a double sequential sampling procedure to test differences in location parameters of two shifted exponential distributions while controlling type I error and minimizing a combined loss from type II error and sampling cost. A smart generalist might read it to see how sequential designs can reduce data collection expenses in settings like environmental monitoring where scale parameters are unknown.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"No significant objection identified","rationale":"The reader's weakest assumption correctly isolates the motivation for replacing the unattainable fixed-sample optimum with a double sequential rule. Because the abstract already flags the unknown-scale obstacle and asserts the efficiencies are proved, and because no contradictory assumption or missing regularity condition is visible, the reader's UNVERDICTED verdict (driven by abstract-only access) does not require adjustment once the full derivations are examined.","tokens_in":1630,"tokens_out":366,"duration_ms":27119,"concrete_test":"Locate the theorem stating second-order risk efficiency (likely Theorem 3.2 or 4.1) and verify whether the initial sample size n0 for scale estimation is required to satisfy n0 \to \\infty with n0 = o(c^\rho) for the appropriate \rho (typically 1/2 or 1/3) as the cost parameter c \to \beta; recompute the risk expansion symbolically under that condition and confirm the o(1) remainder term vanishes.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the double sequential procedure attains first-order efficiency, second-order efficiency, and second-order risk efficiency for testing the difference of location parameters under unknown unequal scales. This rests on the standard construction that an initial stage estimates the scales, after which a fully sequential rule is applied whose stopping boundary is adjusted for the estimated nuisance parameters. The abstract indicates that the loss (type-II error plus linear sampling cost) is minimized asymptotically, which follows from the usual renewal-theoretic arguments once consistent scale estimators are plugged in. No internal inconsistency appears in the motivation or the stated properties; the dependence of the optimal fixed-sample size on unknown scales is correctly identified as the reason a sequential rule is needed.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper proposes a double sequential sampling procedure for testing the difference between location parameters of two shifted exponential models with unknown and unequal scale parameters. It controls type I error at a preassigned level while minimizing a loss function that balances type II error probability and sampling cost. The optimal fixed-sample-size expressions depend on unknown scales, so a double sequential rule is developed; the procedure is claimed to attain first-order efficiency, second-order efficiency, and second-order risk efficiency. Supporting evidence includes simulation studies and an application to heavy precipitation data.","tokens_in":1773,"tokens_out":316,"duration_ms":19546,"significance":"If the asymptotic efficiency results hold with explicit derivations, the framework supplies a cost-efficient sequential alternative to unattainable fixed-sample designs for exponential location testing. It applies standard renewal-theoretic arguments to nuisance-parameter estimation in this model and illustrates utility on environmental data.","major_comments":[{"comment":"Abstract: the assertions of first-order efficiency, second-order efficiency, and second-order risk efficiency are stated without exhibiting the derivations, error bounds, or explicit stopping-boundary adjustments; because these properties are the central claims, the manuscript must supply the full asymptotic analysis (including the effect of plugging in consistent scale estimators) in a dedicated section to make the results verifiable.","section":null}],"minor_comments":[{"comment":"Clarify the precise form of the loss function (weights on type-II error versus per-observation cost) and the initial-stage sample size used for scale estimation.","section":null}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the detailed and constructive report. The single major comment is addressed point-by-point below; we agree that additional explicit material is warranted and will revise accordingly.","responses":[{"response":"We agree that the abstract states the efficiency properties without derivations and that a dedicated section is needed for verifiability. In the revision we will insert a new Section 4 (Asymptotic Efficiency Analysis) that supplies the full derivations. The section will (i) derive the first-order efficiency via the renewal-theoretic representation of the stopping times after consistent scale estimation, (ii) obtain the second-order terms with explicit error bounds of order o(1), (iii) detail the second-order risk efficiency under the combined loss, and (iv) exhibit the adjusted stopping boundaries that incorporate the plug-in estimators. All arguments will be self-contained and will reference the two-sample shifted-exponential renewal structure.","revision_made":"yes","referee_comment":"Abstract: the assertions of first-order efficiency, second-order efficiency, and second-order risk efficiency are stated without exhibiting the derivations, error bounds, or explicit stopping-boundary adjustments; because these properties are the central claims, the manuscript must supply the full asymptotic analysis (including the effect of plugging in consistent scale estimators) in a dedicated section to make the results verifiable."}],"tokens_in":1221,"tokens_out":290,"duration_ms":18057,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"This paper sets up a double sequential procedure for testing the difference between location parameters in two shifted exponential distributions when the scale parameters are unknown and unequal. It aims to control the type I error while minimizing a loss that includes the type II error and the cost of sampling.\n\nWhat stands out is that they correctly note the optimal fixed-sample size depends on those unknown scales, so a sequential rule is necessary. The double sequential method estimates the scales in an initial stage and then applies the test with adjusted boundaries. They claim first-order efficiency, second-order efficiency, and second-order risk efficiency, which follows from standard renewal arguments once consistent estimators are plugged in.\n\nThe simulations and the application to precipitation data at meteorological stations give some evidence that the procedure performs reasonably in practice. That part is useful for seeing how it behaves with real data that fits the exponential model.\n\nOn the downside, the work stays within a fairly narrow model class and the efficiencies are all asymptotic. The loss function has free weights that users must choose, which is standard but adds a tuning step. Nothing suggests a load-bearing flaw in the construction, but the scope limits how far the results travel.\n\nThis is the kind of paper that would interest people working on sequential methods or applied stats in reliability and meteorology. It does not reshape the area but supplies a usable tool for this setting. It has enough grounding and honest engagement with the literature to deserve peer review rather than a desk reject.","headline":"This paper gives a clean but narrow double sequential test for location differences in shifted exponentials with unknown scales, achieving the usual asymptotic efficiencies.","tokens_in":2229,"tokens_out":367,"would_cite":false,"duration_ms":14824,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"A double sequential procedure tests location shifts in exponential models while controlling type I error and minimizing a combined loss of type II error plus sampling cost.","keywords":["sequential sampling","shifted exponential distribution","two-sample test","minimum risk","asymptotic efficiency","type I and type II error","precipitation data"],"falsifier":"A Monte Carlo experiment in which the average risk of the sequential procedure fails to approach the risk of the corresponding optimal fixed-sample-size test as the target error probabilities are driven toward zero.","tokens_in":2543,"feed_emoji":"","tokens_out":649,"duration_ms":15389,"temperature":0.7,"pith_summary":"The paper develops a double sequential sampling method for comparing location parameters of two shifted exponential populations when scale parameters are unknown and unequal. Fixed-sample-size plans cannot be used because the optimal sizes depend on those unknown scales, so the authors replace them with a sequential rule that stops adaptively while keeping the type I error at a preset level. The procedure is shown to achieve first-order efficiency, second-order efficiency, and second-order risk efficiency relative to the unattainable fixed-sample benchmark. Simulation studies and an application to heavy precipitation records illustrate that the method reaches the target accuracies at lower average cost than fixed designs.","feed_headline":"Double sequential test achieves risk efficiency for exponential location shifts","feed_subtitle":"Procedure controls type I error while the combined loss of type II error and sampling cost approaches the unattainable fixed-sample optimum.","key_machinery":"The double sequential sampling procedure, which uses current parameter estimates to decide when to stop sampling in two stages and thereby approximates the unknown optimal fixed-sample sizes.","core_discovery":"The double sequential sampling procedure, constructed to test the difference between location parameters of two shifted exponential models with unknown and unequal scales, controls the type I error at a preassigned level and attains first-order efficiency, second-order efficiency, and second-order risk efficiency for the loss function that balances type II error probability against sampling cost.","pith_inferences":["The same double sequential structure may be adapted to other two-sample problems whose optimal sizes depend on unknown nuisance parameters.","In monitoring applications such as extreme weather, the procedure could reduce the total number of measurements needed to reach a decision.","Comparison with other sequential schemes, such as those based on likelihood ratios, would clarify relative performance under the same loss."],"forward_implications":["The type I error remains at or below the nominal level for all finite samples under the procedure.","The expected sample size and risk converge to those of the optimal fixed-sample plan at the stated rates.","The method applies directly when the two scale parameters differ and are unknown.","Real precipitation records can be analyzed without committing to a fixed number of observations in advance."],"fun_headline_variants":["Double sequential sampling controls type I error for exponential location shifts","Sequential test attains first and second order efficiencies for exponentials","Two-sample sequential testing for exponential models with unknown unequal scales","Double sequential procedure balances type II error and sampling cost","Sequential sampling attains efficiencies for shifted exponential location tests"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"That the loss function combining type II error and sampling cost is minimized by continuing to sample until the double sequential stopping boundaries are crossed, and that the asymptotic efficiencies hold under the shifted exponential model.","fun_headline_variants_meta":{"raw":{"variants":["Double sequential sampling controls type I error for exponential location shifts","Sequential test attains first and second order efficiencies for exponentials","Two-sample sequential testing for exponential models with unknown unequal scales","Double sequential procedure balances type II error and sampling cost","Sequential sampling attains efficiencies for shifted exponential location tests"]},"model":"grok-4.3","cost_usd":0.007122,"raw_usage":{"total_tokens":3253,"prompt_tokens":593,"num_sources_used":0,"completion_tokens":68,"cost_in_usd_ticks":71224500,"prompt_tokens_details":{"text_tokens":593,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2592,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":593,"tokens_out":68,"duration_ms":21167,"temperature":1.0,"reasoning_tokens":2592,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-26T16:21:50.608278+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A Monte Carlo experiment in which the average risk of the sequential procedure fails to approach the risk of the corresponding optimal fixed-sample-size test as the target error probabilities are driven toward zero.","supporting_citations":[],"review_version":1}