{"id":"384c0c45-90bd-4393-89d8-be696565485f","arxiv_id":"2603.16424","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.5,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Douglas–Rachford iteration in scattering coordinates yields an anytime energy-safe coupling for partitioned port-Hamiltonian systems with finite-iteration passivity and hard-coupling convergence.","lead":"The paper gives a parallel coupling method for split robot dynamics that stays energy-safe even when the coupling loop is stopped early. It turns power-consistent interconnection into a Douglas–Rachford iteration in wave variables so finite budgets still certify discrete passivity and recover the monolithic step as the budget grows.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified beyond the reader's already-flagged FNE fragility; the central anytime-passivity argument holds under its stated conditions.","rationale":"The strongest claim is precisely Theorem 1 (augmented-storage discrete passivity for any finite Kn under Conditions 1–2) plus the hard-coupling limit of Theorem 2. The only place that claim can fail while the rest of the construction remains intact is loss of firm nonexpansiveness of Sn. The reader already identified this, noted that the Duffing case is only trajectory-checked, and correctly assigned CONDITIONAL / medium correctness risk. My re-examination of the DR reduction (24)–(25), the Fejér argument in the proof of Theorem 1, and the numerical diagnostics (Figs. 5–6) finds no additional hidden assumption, algebraic error, or over-claim. The concrete test above simply makes the already-flagged dependence falsifiable in one controlled experiment. Consequently the verdict stays CONDITIONAL and no adjustment is warranted.","tokens_in":14737,"tokens_out":604,"duration_ms":6464,"concrete_test":"Re-run the two-oscillator benchmark with a deliberately detuned γ that violates the empirical FNE margin (e.g., γ larger than the observed min λmin of the incremental discrete impedance along the trajectory) and report the positive-part of the augmented-storage residual of (28). If residuals remain at roundoff, the numerical certificate is more robust than the theory suggests; if they become systematically positive, the FNE dependence is confirmed as load-bearing exactly as claimed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's weakest_assumption correctly isolates the load-bearing point: Theorem 1 rests on Condition 2 (firm nonexpansiveness of the frozen wave maps Sn_i), which supplies the maximal monotone operator An so that the DR operator is firmly nonexpansive and Fejér-monotone. Proposition 1 gives a sufficient (and, for linear symmetric impedance, necessary) γ-rule Zn_i,d ⪰ γI, but for the stiff Duffing subsystem this is only verified numerically along realized trajectories (Fig. 5a, finite test pairs, tol=1e-12). If FNE fails for some state, the Fejér residual term no longer certifies the augmented-storage inequality (28). That said, the paper does not claim an a-priori global FNE proof for nonlinear ports; it states the condition, supplies a practical monitor (the empirical margin Δ), and shows non-negative margins plus roundoff-level residuals on the chosen benchmark. The proofs themselves (Fejér decrease of the DR residual + Condition 1) are standard and free of circularity. Thus the concern is real but already accurately scoped by the reader; it does not introduce a new internal inconsistency or derivation gap that would overturn the conditional acceptance of the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper proposes an early-terminable (anytime) iterative coupling interface for parallel simulation of partitioned port-Hamiltonian systems. At each macro-step, subsystem integrators induce frozen scattering port maps; a Douglas–Rachford inner iteration in wave coordinates reconciles these maps with a lossless (orthogonal) interconnection constraint. Under discrete passivity of the subsystem one-step maps (Condition 1) and firm nonexpansiveness of the frozen wave maps (Condition 2, enforced by an impedance-tuning rule on γ), Theorem 1 proves an augmented-storage inequality that certifies discrete passivity of the coupled macro-step for any finite inner-iteration budget Kn, with residual mismatch captured by Fejér decrease of the DR fixed-point residual. Theorem 2 shows that as Kn→∞ the partitioned update recovers the monolithic discrete-time update induced by the same integrators. A two-oscillator (linear–Duffing) benchmark reports FNE margins, passivity residuals at numerical roundoff (~1e-14), and monotonic RMS state-error decay with Kn.","tokens_in":15021,"tokens_out":1325,"duration_ms":12048,"significance":"If the claims hold under the stated conditions, the work supplies a principled, certificate-driven alternative to ad-hoc co-simulation interfaces: an anytime energy-safe contract that remains valid under finite inner budgets while converging to the monolithic discretization. That combination is directly relevant to real-time parallel simulation and co-simulation of robotic systems under varying compute budgets. Strengths include a clean reduction of power consistency to wave-domain orthogonality and norm inequalities, an explicit link between discrete passivity and monotone-operator resolvents, and a transparent Fejér argument for finite-iteration safety. The experimental residuals at double-precision roundoff and the monotonic hard-coupling error decay support the theory on the chosen benchmark. The main practical caveat is that Condition 2 is only a priori guaranteed for linear symmetric impedances (Proposition 1) and is checked numerically for the nonlinear Duffing subsystem; the contribution remains useful as a certified interface design pattern with an explicit monitor for the FNE margin.","major_comments":[{"comment":"Condition 2 / Proposition 1 and §V (Fig. 5a): Theorem 1’s Fejér argument requires firm nonexpansiveness of every frozen wave map Sn_i. Proposition 1 gives a sufficient (and, for linear symmetric Z, necessary) rule Zn_i,d ⪰ γI, but for the stiff Duffing subsystem the paper only verifies non-negative empirical margins Δ along realized trajectories on a finite set of test pairs (tol=1e-12). This is consistent with the stated claims and does not create circularity, but it is load-bearing for the nonlinear case: if FNE fails off the tested pairs or under other (γ,Δt,amplitude) regimes, the augmented-storage certificate (28) no longer holds. The manuscript should either (i) strengthen the a-priori guarantee for the discrete-gradient Duffing map under the chosen γ, or (ii) more prominently frame Condition 2 as a runtime-monitored hypothesis and report the empirical margin as part of the certifi","section":"Condition 2, Proposition 1, §V Fig. 5a"},{"comment":"§V benchmark scope: The only numerical support is a two-port linear–Duffing oscillator pair (Table I, Figs. 4–6). The theory is written for general N and multiport Dirac interconnections (stacked waves, orthogonal P), yet there is no multiport, multi-subsystem, or stiff DAE-style example that would stress parallel evaluation, non-swap P, or more severe state-dependent impedance. A second experiment (or a clear limitation statement) is needed before the “parallel simulation of robotic systems” framing can be taken as demonstrated rather than illustrated.","section":"§V Experiments"}],"minor_comments":[{"comment":"Abstract and §I claim “1e-14” / “10e-14” residuals; Fig. 5b positive-part summaries are consistent with roundoff but the exact residual definition (how the RHS of (28) is moved) should be stated once in the caption or text so the number is reproducible.","section":"Abstract, §V Fig. 5b"},{"comment":"Notation: z is used for effort/flow pairs in §II and ζ for wave pairs in §IV; a short reminder when ζ is introduced would help. Also, “10e-14” in the abstract should be “10^{-14}”.","section":"§II–IV, Abstract"},{"comment":"Algorithm 1 and (18)/(25): the lifted DR form and the reduced form are stated to be equivalent; a one-line pointer that the shadow ˆb used for the macro-step is the same object in both presentations would reduce reader friction.","section":"§III-C, §IV-B"},{"comment":"Remark 6 (nonuniqueness) is appropriately cautious; a sentence on how the experiments select the monolithic reference (same init / same discrete-gradient solver) would close the loop with Theorem 2.","section":"Remark 6, §V"},{"comment":"Related work on TLM and energy-leak correction is cited; a brief contrast on delay-free vs. delay-based passivity certificates would help position the contribution for co-simulation readers.","section":"§I"}],"recommendation":"minor_revision","confidential_remarks":"The central anytime-passivity argument is sound under its stated conditions; the FNE fragility for nonlinear ports is real but already scoped by the authors via the empirical margin. I would not block on that alone. The main editorial risk is overclaim relative to a single two-oscillator benchmark: if the journal expects multi-body or multiport robotic demos for cs.RO letters, a second experiment or a tighter title/abstract framing may be needed. Otherwise minor revision is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The useful thing here is a concrete parallel interface (Alg. 1) that stays energy-safe for any finite inner budget Kn, not only at convergence. They put Douglas–Rachford in scattering coordinates so the lossless coupling is an orthogonal constraint and the DR residual becomes an explicit algorithmic-dissipation term in an augmented storage. Theorem 1 is short and standard once you grant discrete passivity of the integrators (Condition 1) and firm nonexpansiveness of the frozen wave maps (Condition 2): Fejér decrease of the DR residual plus the subsystem energy inequality gives the macro-step certificate. Theorem 2 is the usual hard-coupling limit under continuity. That combination—iterate-level certificate plus recovery of the monolithic discrete update—is the actual novelty; the ingredients (wave variables, TLM-style thinking, pH integrators, DR) are all cited and known.\n\nWhat they do well: the derivation is not circular. γ is a design parameter with a clear sufficient rule (Prop. 1) for linear symmetric impedance, and they monitor the FNE margin numerically rather than hide it. On the linear–Duffing pair the passivity residuals sit at double-precision roundoff (~1e-14) and RMS state error falls monotonically with Kn, which matches the claims. Citations look appropriate for co-simulation, pH structure preservation, and monotone operators.\n\nSoft spots, in proportion: Condition 2 is load-bearing. For the stiff Duffing subsystem they only check FNE along realized trajectories with finite test pairs; if it fails off-trajectory the Fejér argument collapses. That is already scoped honestly in the paper (they treat γ as tuning and report the empirical margin), but it is still the main fragility for stiff nonlinear ports. The experiment is minimal—one two-oscillator case, no code release, no multi-port or asynchronous stress. Those are real limits on how far you can trust the method today, not holes in the stated theorems.\n\nThis is for people who care about certified co-simulation, partitioned multibody, or real-time parallel robot simulation under budget. It is not a continuum-mechanics rewrite. I would send it to peer review; the formal core is solid enough to deserve referee time, with the usual requests for broader validation and clearer FNE control. Worth engaging if that interface problem is on your desk.","headline":"Clean anytime energy-safe DR interface in wave coordinates; the math holds under its conditions, the main soft spot is FNE for nonlinear ports and a thin benchmark.","tokens_in":15654,"tokens_out":569,"would_cite":true,"duration_ms":6169,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A wave-domain Douglas–Rachford interface keeps partitioned port-Hamiltonian simulations energy-safe at any finite iteration budget.","keywords":["port-Hamiltonian systems","co-simulation","Douglas–Rachford splitting","scattering variables","discrete passivity","parallel simulation","wave-domain coupling"],"falsifier":"On the Duffing–linear benchmark, choose an impedance γ that drives the measured firm-nonexpansiveness margin of the frozen port maps negative; if the positive part of the augmented-storage residual then rises well above roundoff while the same integrators remain passive, Theorem 1 fails.","tokens_in":15613,"feed_emoji":"⚙️","tokens_out":630,"duration_ms":5993,"temperature":0.7,"pith_summary":"Parallel robotic simulation often splits a large model into coupled subsystems, but stopping the interface iteration early can inject fake energy even when every piece is passive. This paper builds the interface in scattering (wave) coordinates and reconciles the subsystems with a Douglas–Rachford inner loop. Passivity becomes a norm inequality and lossless coupling becomes an orthogonal swap, so the iteration’s Fejér decrease supplies algorithmic dissipation that can be booked in an augmented storage. Under a passivity condition on each subsystem integrator and an impedance-tuning condition that makes the frozen port maps firmly nonexpansive, the coupled macro-step stays discrete-passive for any finite inner budget and converges to the monolithic discrete update as that budget grows. On a linear–Duffing oscillator benchmark the energy residual sits at numerical roundoff and the state error falls as more inner iterations are allowed, giving a certified accuracy–compute knob for real-time parallel co-simulation.","feed_headline":"Wave-domain iteration keeps split robot sims energy-safe anytime","feed_subtitle":"Finite Douglas–Rachford budgets still certify discrete passivity and approach the monolithic update","key_machinery":"The lifted Douglas–Rachford operator in wave coordinates: coupling is the orthogonal projection onto the subspace a = P b, subsystem maps are firmly nonexpansive resolvents, and Fejér monotonicity of the DR residual supplies the algorithmic dissipation term in the augmented storage Vn.","core_discovery":"Under passivity-preserving subsystem integrators and firm nonexpansiveness of the frozen wave port maps (secured by impedance tuning), a Douglas–Rachford iteration in scattering coordinates yields an augmented-storage inequality that certifies discrete passivity of the coupled macro-step for every finite inner-iteration budget, while the same iteration recovers the monolithic discrete-time update in the hard-coupling limit.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Wave-domain DR keeps partitioned robot sims energy-safe for any finite budget","Finite Douglas–Rachford iterations certify discrete passivity in split PH systems","Early-stoppable energy-safe coupling for parallel port-Hamiltonian simulations","Scattering DR yields anytime passivity for partitioned robot dynamics","Impedance-tuned wave iteration preserves energy safety under finite coupling steps"],"cache_read_input_tokens":128,"weakest_assumption_plain":"Each frozen discrete port map must be firmly nonexpansive, which the paper obtains only by tuning the scattering impedance and, for the nonlinear oscillator, only verifies numerically along the realized trajectory rather than proving it for all relevant states.","fun_headline_variants_meta":{"raw":{"variants":["Wave-domain DR keeps partitioned robot sims energy-safe for any finite budget","Finite Douglas–Rachford iterations certify discrete passivity in split PH systems","Early-stoppable energy-safe coupling for parallel port-Hamiltonian simulations","Scattering DR yields anytime passivity for partitioned robot dynamics","Impedance-tuned wave iteration preserves energy safety under finite coupling steps"]},"model":"grok-4.5","effort":"low","cost_usd":0.005318,"raw_usage":{"total_tokens":1447,"prompt_tokens":745,"num_sources_used":0,"completion_tokens":98,"cost_in_usd_ticks":53180000,"prompt_tokens_details":{"text_tokens":745,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":604,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":745,"tokens_out":98,"duration_ms":4656,"temperature":1.0,"reasoning_tokens":604,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-13T23:46:57.520813+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"On the Duffing–linear benchmark, choose an impedance γ that drives the measured firm-nonexpansiveness margin of the frozen port maps negative; if the positive part of the augmented-storage residual then rises well above roundoff while the same integrators remain passive, Theorem 1 fails.","supporting_citations":[],"review_version":1}