(1) Strong Structural Priors and Benchmark-Wrapper Alignment ("White-Box" Coupling Geometry):
PACT's dramatic sample efficiency and parameter compactness (r+1=4 scalars) rely on two strong domain assumptions: (i) exact knowledge of the infrastructure coupling operator W (e.g., exact linearized DistFlow reactances Sm[i,j] in MAPDN and exact spatial decay scale λ and hardware class gains recvcls(i) in VMAS), and (ii) immediate, noiseless access to all peers' executed actions ajex(t−1) and an explicit, uncontaminated per-step sensor for relative excess yi(t) (e.g., observing both aisent and aiex separately). Crucially, because the authors also designed the interaction-mediated harm injections on the stock URB, MAPDN, and VMAS benchmarks (Section 3.1), the benchmark harm definitions (Eqs. 7–10) and PACT's basis construction Λi(⋅;W) share the exact same operator W and channel structure below saturation (Proposition C.2(i)). While URLS tests removing W, URLS must fit O(n) raw parameters; there is no evaluation of PACT under misspecified or approximate operators W^=W (e.g., inaccurate line impedances or unmodeled cross-class leakage) or noisy excess measurements.
(2) Marginal Empirical Impact on URB and Lack of Environment Details for Multi-Agent MuJoCo:
On the URB routing benchmark (Table 1), where the true harm (a max over link volume-to-capacity ratios, Eq. 5–6) lies outside PACT's linear additive surrogate class, the harm itself only degrades host learners by 1.3%−3.8%. PACT (fixed g) reduces travel time from 3.401 to 3.369 (−0.032), but 0.024 (75%) of this improvement already occurs at σ=0 simply because Eq. (16) biases logits toward shorter free-flow travel times fftk. Moreover, simple vanilla IQL (3.160) and Greedy (3.044) substantially outperform the IPPO host and PACT (3.369) at σ=3, and learned g (3.397) provides virtually no gain over g≡0 (3.401).
Section 5.4 introduces a fourth experiment on Multi-Agent MuJoCo (Ant-4x2 at σ=3, Fig. 3), but the paper never defines what the shared medium (M), coupling operator (W), exogenous driver (D), or harm map (H) actually are for Ant-4x2, nor is it mentioned in Section 3.1 or Appendix A.
(3) Breakdown of Unbiased Identification and Exact Compensation under Physical Saturation:
PACT's core theoretical guarantees (Proposition C.2, Theorem 4.1, and Corollary C.10) explicitly assume that the system operates strictly below saturation—meaning neither the disturbance clip (qsat, δmax) nor the actuator capability box (qˉi, clipbox) is active[cite: 4, 5, 6, 21, 26]. Furthermore, Section 4 claims that because the regression target yi is measured against the sent action aisent, compensation does not bias identification[cite: 5]. However, at high severity (σ=3), compensating via aisent=aicmd/(1−c^i) or qisent=qicmd−gisˉiy^i actively pushes aisent toward the actuator saturation limits[cite: 6]. Once the outer actuator clip activates, the observed excess yi conflates peer coupling harm with hardware saturation, introducing severe non-zero bias (bˉ>0 in Theorem C.4) that is driven by the agent's own compensation[cite: 4, 5, 22]. The paper does not analyze how frequently saturation occurs at σ=3 or how much estimation bias it induces.
(4)Monotonicity with respect to severity in Proposition C.1(v):
The proposition claims that the magnitude of interaction-mediated harm is non-decreasing in the severity parameter σ. However, the proof treats the coupling load xi(t) as fixed, whereas this load depends on other agents' previously executed actions, which themselves depend on σ.Consequently, the claimed monotonicity need not hold when comparing complete closed-loop trajectories under different severity levels. For example, consider a two-agent unsaturated feeder model with qiex(t)=0.5−σqjex(t−1) and initial executed actions equal to 0.5. At t=2, the reaction magnitude is 0$.5\sigma|1-\sigma|$, which is not monotonic in σ. The proposition should explicitly restrict monotonicity to fixed coupling loads and driver values, or provide additional assumptions guaranteeing closed-loop monotonicity.
We thank all reviewers for their valuable feedback and believe that all their concerns can be addressed. However, we have withdrawn the paper to undertake additional work that may not be feasible to complete within the rebuttal period.