{"id":"36e2238b-d93f-4191-b29e-0ed00af225e4","arxiv_id":"2602.01131","paper_version":2,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":3,"one_line_summary":"Radial excitations of ρ, π, φ, D, D_s, B, B_s, ψ and Υ mesons obey d_{n,1}=β_1 n + N (universal slope, family intercept), claimed here to follow analytically from a local potential.","lead":"Meson families share a simple rule: the squared mass of each radial excitation rises in a straight line with excitation number, with a common slope. The authors say they derive this rule from a local-potential model and check it against nine quark-antiquark families.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The stability guarantee is not proven: Proposition 1 maps the Lyapunov constraint to a per-user bandwidth bound using Markov's inequality on the max-latency, but E[max_i T_i] ≤ D_req does not follow from E[T_i] ≤ D_req; a two-user counterexample can violate (17a).","rationale":"The reader's verdict is based on a summary that does not match the manuscript: the reader discusses meson radial excitations and a linear mass law, whereas the paper is about UAV-assisted low-altitude networks with a Stackelberg game and pruning-based PPO. This mismatch means the reader's weakest_assumption cannot be used. Applying the stress-test to the actual paper, the most load-bearing concern is the validity of the control-to-communication mapping in Proposition 1. The proof uses Markov's inequality on the maximum latency but then derives per-user bandwidth bounds, an inference that is mathematically unsound. The stability claim is the headline contribution, and it depends directly on this mapping. The paper does provide some independent support: a closed-form Stackelberg equilibrium and simulation comparisons of the pruning algorithm. However, those do not repair the logical gap in the stability guarantee. A concrete counterexample test can settle the issue; if it lands, the paper should be revised to either fix the mapping with a correct worst-case bound or to weaken the stability claim. Hence the verdict should move from UNVERDICTED to CONDITIONAL.","tokens_in":30647,"tokens_out":5667,"duration_ms":63597,"concrete_test":"Construct a two-user counterexample. Set e_n = 0.5 s, Γ_n = 0.8 (so D_req = 0.1 s), T_fixed = 0.02 s for both users, and S_1 = S_2 = 40 kbits. Choose SNR values 0 dB and 10 dB, and set κ_i exactly equal to κ_min^i from Eq. (18). With Rayleigh fading on the channel gains (or with a simple two-point distribution for each latency), compute the distribution of T_1 and T_2, then numerically evaluate E[max(T_1, T_2)]. If E[max(T_1, T_2)] > 0.1 s while each E[T_i] ≤ 0.1 s, the proposed constraint (23a) fails to guarantee (17a). This would falsify Proposition 1.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central claim is that the proposed SC3 scheme 'secures control loop stability.' This rests on Proposition 1 (Section III-B), which transforms the Lyapunov condition (17a) into the bandwidth lower bound (18). The proof has a load-bearing gap. In (15), the aggregate latency is defined as the maximum over connected users: T_n = max_i T_i,n^total. The Lyapunov constraint is expressed as a success probability for the whole loop: P(T_n ≤ e_n) ≥ Γ_n. The proof then uses Markov's inequality to obtain E[T_n] ≤ D_req = (1−Γ_n)e_n. From this, the authors subtract per-user fixed delays to produce T_budget and derive κ_i ≥ S_i / (T_budget log2(1+SNR_i,n)). However, the condition E[T_n] ≤ D_req is a statement about the maximum over users. Satisfying E[T_i,n^total] ≤ D_req for each individual user is necessary but not sufficient for E[max_i T_i,n^total] ≤ D_req; by Jensen, E[max] ≥ max_i E[T_i], and the gap can be arbitrarily large when the T_i are heavy-tailed or strongly fluctuating. Thus the derived per-user bandwidth allocation does not, by itself, enforce the Lyapunov descent condition. A second issue compounds this: the required success probability Γ_n in (20) depends on the current state ζ(k) through V_close and V_open, so a static bandwidth lower bound cannot guarantee a state-dependent success threshold. Together, these show that the central stability guarantee is not established as stated. The closed-form Stackelberg solution and pruning results may still be valid, but the control-stability claim needs a corrected derivation (e.g., a union bound or a scheduling rule that decouples users).","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a unified Sensing-Communication-Computing-Control (SC3) framework for UAV-assisted low-altitude economy networks. The authors use Lyapunov stability theory to derive a mapping from control-loop stability to per-user bandwidth constraints (Proposition 1), formulate the resource allocation problem as a Stackelberg game with UAVs as leaders and users as followers, and solve the game with a pruning-based PPO algorithm that dynamically prunes neural network structures during training. Simulation results are presented to show that the proposed scheme keeps the allocated bandwidth above the derived stability threshold and improves system utility compared to baselines.","tokens_in":31089,"tokens_out":9295,"duration_ms":88897,"significance":"If the stability-to-bandwidth mapping were rigorously established, the paper would offer a valuable cross-layer framework that ties physical control stability to communication resource allocation in UAV networks. The Stackelberg game formulation with a closed-form equilibrium and the lightweight pruning-based PPO are practical and relevant contributions for edge deployment. However, the central theoretical claim—that the proposed scheme 'secures control loop stability'—rests on Proposition 1, whose proof has load-bearing gaps: the state dependence of the success threshold is not handled, and the per-user bandwidth bound is at best a sufficient condition rather than the claimed minimum. The simulations also do not directly verify the Lyapunov descent condition. The framework is promising, but the current version does not adequately support its central stability guarantee.","major_comments":[{"comment":"The success threshold Γ_n in Eq. (20) depends on the current control state ζ(k) through V_close and V_open, making D_req and T_budget state-dependent. The paper nevertheless treats κ_min_i as a fixed constraint in Problem 1 (23a) and in the Stackelberg price cap (28a). No mechanism is provided for resolving this state dependence (e.g., worst-case bound, state augmentation, or online estimation). Thus the derived bandwidth lower bound does not, by itself, enforce the state-dependent Lyapunov condition (17a).","section":"§III-B, Proposition 1 (Eqs. (19)–(22))"},{"comment":"The aggregate latency T_n in Eq. (15) is the maximum over connected users. The proof expands the Lyapunov expectation with a single-user success probability ξ_i,n and derives a per-user bound from E[T_n] ≤ D_req. For a maximum, satisfying E[T_i,n^total] ≤ D_req for each user does not imply E[max_i T_i,n^total] ≤ D_req; conversely, the per-user deterministic bound S_i/R_i ≤ T_budget is sufficient but not necessary. The claim that κ_min_i is the 'minimum bandwidth required' is therefore not established. The proposition should be restated as a sufficient condition and proved accordingly.","section":"§III-B, Proposition 1 (Eqs. (15), (19), (22))"},{"comment":"The follower's best response κ*_i = ϑ_i/ϱ_n − 1/H_i,n can be negative when ϱ_n > ϑ_i H_i,n; the non-negativity constraint is not imposed. Additionally, the stability price cap in (28a) should be ϱ_n = min_i ϑ_i/(κ_min_i + 1/H_i,n) over all connected users, but the paper defines a single ϱ_n and drops the user index in (30). Without these corrections, the closed-form equilibrium may be infeasible and the stability constraint may be violated for some users.","section":"§III-B, Eq. (27)–(30)"},{"comment":"The simulations verify that the allocated bandwidth exceeds the derived 'min requirement,' but they do not directly validate the Lyapunov descent condition (6) or the probabilistic stability guarantee. Since the central claim is securing control-loop stability, the paper should report a direct stability metric (e.g., empirical decay of V(ζ(k)), or empirical success probability P(T_n ≤ e_n) compared with the threshold Γ_n). Without such evidence, the stability claim is not empirically supported.","section":"§V, Fig. 2"}],"minor_comments":[{"comment":"Typo: 'Stakelberg' should be 'Stackelberg' in the figure caption.","section":"Fig. 1"},{"comment":"The notation ϱ_n is used both for the price variable and for the upper bound; please clarify, and explicitly define the min over i for the stability cap.","section":"§III-B, Eq. (28a)"},{"comment":"The Lyapunov decay rate ρ_n = 0.95 is given without justification or sensitivity analysis; stability guarantees depend critically on this parameter.","section":"Table I"},{"comment":"The explanation of the non-monotonic effect of pruning start epoch is speculative; no error bars, confidence intervals, or multiple seeds are reported.","section":"§V, Fig. 4"},{"comment":"Line 22: 'Updatemask' should be 'Update mask'. Also, the pruning threshold in Eq. (36) is not clearly defined in terms of the importance score n_j^(h); please formalize.","section":"Algorithm 1"}],"recommendation":"major_revision","confidential_remarks":"The paper's weakest link is the Lyapunov-to-communication mapping in Proposition 1. The proof currently does not support the strong claim of a 'minimum bandwidth' that guarantees stability; the state dependence of Γ_n and the max-over-users aggregation are both unresolved. The authors should either repair the proof to be a valid sufficient condition with a fixed (conservative) threshold, or present the stability guarantee as conditional on a worst-case state bound. The pruning-based PPO component is largely incremental over the authors' prior work [8]. The paper may be suitable for a major revision if the theoretical core is rewritten and the simulations are augmented with direct stability metrics."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper proposes a closed-loop sensing-communication-computing-control framework for UAV networks, mapping Lyapunov stability to bandwidth constraints via a Stackelberg game and solving it with pruning-based PPO. The integration is well motivated, and the game formulation is clean. The user and leader problems each have unique solutions, the closed-form price is sensible, and the pruning ablations are competent.\n\nThe problem is Proposition 1, which is the load-bearing bridge from stability to bandwidth. The proof uses Markov's inequality on T_n, the maximum per-step latency across users, to get E[T_n] < D_req. Then it subtracts fixed delays and derives a per-user bandwidth lower bound. That step is invalid: ensuring each user's expected latency stays below D_req does not ensure the expected maximum is; Jensen goes the other way. The bound you get is necessary, not sufficient, for the Lyapunov condition. On top of that, the threshold Γ_n in (20) is state-dependent through V_open and V_close, so a static bandwidth allocation based on one state cannot guarantee the descent condition for all states. The paper also conflates transmission success probability with a latency deadline probability without justification. So the headline claim that the scheme \"secures control loop stability\" is not established.\n\nWhat survives is the Stackelberg game and the pruning-based PPO, which are reasonably evaluated and may be empirically reasonable. But the theory needs a corrected derivation—for instance, a union bound over users or a scheduling rule that decouples the max, plus a time-varying budget that tracks the Lyapunov state. Without that, the central contribution is overstated.\n\nOne note: the attached reader's report is about meson mass spectra, not this paper; I disregarded it. The stress-test note is accurate.\n\nThis is a serious paper worth sending to peer review, but it needs major revision. A careful referee should fix Proposition 1 and the state-dependence issue before the stability claim can stand.","headline":"Useful systems integration with a real proof gap: the Lyapunov-to-bandwidth mapping in Proposition 1 conflates per-user expected latency with the expected max, and the state-dependent threshold is not handled, so the stability guarantee is not established as stated.","tokens_in":31573,"tokens_out":3396,"would_cite":false,"duration_ms":35993,"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":"UAV control stability can be converted into a bandwidth price floor, and a pruning-based reinforcement learning algorithm can reach the resulting market equilibrium on resource-limited hardware.","keywords":["low-altitude economy","UAV networks","Lyapunov stability","control-communication co-design","Stackelberg game","resource pricing","deep reinforcement learning","structured pruning"],"falsifier":"Configure a hardware-in-the-loop UAV tracking test, give every link exactly the derived κ_min, and count how often the Lyapunov function fails to decrease at the required rate; if the empirical violation frequency exceeds 1−Γ under the assumed packet-loss model, the closed-form mapping under-provisions bandwidth.","tokens_in":30569,"feed_emoji":"🚁","tokens_out":6409,"duration_ms":72177,"temperature":0.7,"pith_summary":"This paper argues that in low-altitude UAV networks, communication resources should be allocated with an explicit eye on the physical stability of the control loops they serve, not just on throughput. It derives a closed-form mapping from a Lyapunov stability condition to a minimum bandwidth per user, then embeds that floor in a Stackelberg game where UAVs set bandwidth prices and users choose how much to request. To make the game tractable on battery-powered UAVs, the authors propose a structured-pruning version of Proximal Policy Optimization that shrinks the neural network during training. Simulations show the resulting prices keep every link above its stability bandwidth while raising system utility, and that pruning can improve final reward by acting as a regularizer.","feed_headline":"Lyapunov stability sets a bandwidth floor for every UAV link","feed_subtitle":"A control-to-communication mapping lets UAVs price bandwidth so each control loop gets the latency it needs to stay stable.","key_machinery":"The load-bearing object is the Lyapunov descent inequality for the quadratic function V=ζ^T P ζ on the augmented state (position, velocity, previous control). With the packet-success model it gives the required success probability Γ; Markov's inequality converts Γ into a latency budget and then into the bandwidth floor κ_min. In the Stackelberg game, the follower's utility is strictly concave, yielding κ*_i=ϑ_i/ϱ_n−1/H_i,n; the leader's utility is concave too, so the optimal price is the unconstrained optimum clipped by a stability cap and a capacity floor. Dynamic structured pruning (binary masks with a cubic sparsity schedule) compresses the actor–critic network during training.","core_discovery":"The central claim is that UAV control stability can be converted, in closed form, into communication constraints. The paper models the UAV–user loop as a sensing–communication–computing–control system with fixed sampling, derives a Lyapunov descent condition for the augmented tracking state, and uses Markov's inequality to turn the required success probability into a latency budget. That budget yields a per-user bandwidth floor κ_min = S_i/(T_budget log2(1+SNR)). These floors are embedded in a Stackelberg game: UAVs lead by pricing bandwidth, users follow by requesting bandwidth, and backward induction gives a unique optimal price in one of three regimes—unconstrained, stability-capped, or c","pith_inferences":["Because the derivation only needs a monotone link between a resource and latency, the same stability-floor idea could be applied to transmit power, computation frequency, or schedule slots, not just bandwidth.","The Markov-inequality step is sufficient but not necessary, so the computed κ_min may be conservative; in deployments with well-characterized delay tails, a tighter probabilistic bound could lower the floor without endangering stability.","The pruning-as-regularization result suggests sparse policies might generalize better in other control-oriented DRL tasks, independent of the pricing game.","A natural stress test is to replace the fixed sampling period with a jittered or event-triggered sampler; the Lyapunov argument would need a different treatment, but the price-floor architecture could remain."],"forward_implications":["Stability becomes a hard lower bound on allocated bandwidth, so control requirements can be checked at the resource allocation layer.","The optimal price is the unconstrained profit-maximizing price capped when stability is at risk and floored when capacity is at risk, making pricing a distributed stability safeguard.","Users with higher urgency weights receive larger bandwidth at the equilibrium, so the game inherently prioritizes control-critical links.","Dynamic structured pruning produces a smaller policy network with no loss—and sometimes a gain—in test reward, suggesting DRL controllers can run on energy-limited UAVs.","If the equilibrium is reached, no UAV or user can improve utility by changing price or request alone, giving a stable operating point for the closed loop."],"fun_headline_variants":["Lyapunov maps stability to a per-link bandwidth floor","Pruned PPO keeps UAV control loops stable under latency","Stability-aware pricing in a UAV Stackelberg game"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole mapping rests on treating the control loop as synchronous and linear, with latency appearing only as a shortened actuation window and packet drops as the only disturbance; if real delays cause missed, reordered, or asynchronous updates, the derived bandwidth floor is not a stability guarantee.","fun_headline_variants_meta":{"raw":{"variants":["Lyapunov maps stability to a per-link bandwidth floor","Pruned PPO keeps UAV control loops stable under latency","Stability-aware pricing in a UAV Stackelberg game"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001562,"raw_usage":{"total_tokens":6106,"prompt_tokens":807,"completion_tokens":5299,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":551,"completion_tokens_details":{"reasoning_tokens":5246}},"tokens_in":551,"tokens_out":5299,"duration_ms":40213,"temperature":1.0,"reasoning_tokens":5246,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T05:46:41.747347+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Configure a hardware-in-the-loop UAV tracking test, give every link exactly the derived κ_min, and count how often the Lyapunov function fails to decrease at the required rate; if the empirical violation frequency exceeds 1−Γ under the assumed packet-loss model, the closed-form mapping under-provisions bandwidth.","supporting_citations":[],"review_version":1}