{"id":"f6532978-2164-4875-afdf-d82fe14ccce1","arxiv_id":"2412.01045","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"An adaptive-time-step explicit symplectic integrator is constructed by combining the Preto-Saha step-size control with time-transformed split Hamiltonians, and it improves energy conservation over the nonadaptive version in tests around three black hole spacetimes.","lead":"This paper adds adaptive time stepping to explicit symplectic integrators for curved spacetimes, at a cost of only two extra computational steps per integration step. Tests on particle and photon orbits near Schwarzschild, Kerr, and Schwarzschild-Melvin black holes show roughly two orders of magnitude better energy conservation, and in one case a corrected chaos classification.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Claimed symplecticity fails: F in Eq. (15) has no τ-dependence when φ=j/r, so Hamilton's equations give dΦ/ds=0, contradicting Eq. (16) that drives the adaptive step.","rationale":"I agree with the reader that Eq. (16) is the weakest point, but I would go further: it is not merely informal, it is inconsistent with Hamilton's equations for the stated Hamiltonian. The paper defines φ = j/r as a function of the coordinate r (Eq. 22) and states that r, θ, τ are independent coordinates (footnote 5). Then F in Eq. (15) does not contain τ, so ∂F/∂τ = 0 and Hamilton's equation gives dΦ/ds = 0, which contradicts the nonzero Φ-update in Eq. (23) whenever p_r ≠ 0. The derivation of Eq. (16) uses the orbit-level total derivative d ln φ/dτ, which is not a partial derivative of F on the phase space. Additionally, Steps 1 and 5 (Eq. 24) freeze r, p_r, g during the substep, so the Φ-update is not the exact flow of a Hamiltonian; the exact flow of F1 = −g ln φ would also change p_r and p_θ. Hence AS2 is not a composition of exact Hamiltonian flows, and the assertion in Point 4 that fixed h in s preserves symplecticity is unjustified. This is load-bearing because the entire adaptive mechanism and the claimed symplectic property rest on Eq. (16). The numerical results may still hold as an empirical adaptive scheme, and the paper provides useful benchmarks, but the central claim as stated is not supported. A single analytical check of ∂F/∂τ settles the issue.","tokens_in":26353,"tokens_out":19477,"duration_ms":170987,"concrete_test":"Analytical check for the Schwarzschild case (g=1, Sec. 3.1): write F explicitly as F = K(r,θ,p_r,p_θ)/Φ + ln(Φ r / j). In the extended phase space (r, θ, τ, Φ) with φ = j/r, compute −∂F/∂τ at a point with p_r ≠ 0. Since F does not contain τ, this partial derivative is exactly 0, while Eq. (23) gives dΦ/ds = −g_rr p_r / r, which is nonzero for p_r ≠ 0. These two expressions disagree at any such point, so Eq. (16) is not Hamilton's equation for F. This single inconsistency shows that the Φ-update driving the adaptive time step does not follow from the Hamiltonian F, and the claimed symplecticity of AS2 is contradicted by the paper's own equations.","verdict_should_be":"REJECT","load_bearing_attack":"The central claim is that AS2(h) (Eq. 26) is an explicit symplectic integrator with adaptive old-time steps, built from the Hamiltonian F = K/Φ + g ln(Φ/φ), φ = j/r (Eq. 15). This requires F to be a Hamiltonian on the extended phase space (r, θ, p_r, p_θ, τ, Φ) and the split maps F1, F2, S2 to be exact flows of its sub-Hamiltonians. The load-bearing step is Eq. (16): dΦ/ds = −∂F/∂τ = g d ln φ/dτ. With φ = j/r, F has no dependence on the coordinate τ: r and θ are independent coordinates, so ∂F/∂τ = 0. Hamilton's equation then gives dΦ/ds = 0, not the generally nonzero value of Eq. (23), dΦ/ds = −g g^{rr} p_r / r. The paper's derivation computes d ln φ/dτ along the orbit using dr/dτ from the geodesic equation; that is a total derivative along the solution, not a partial derivative of F at a fixed phase-space point. If φ is instead taken as an explicit function of τ, F is no longer the Hamiltonian whose K/Φ part is solved by S2(h/Φ), and the method is not a splitting of F. Moreover, the updates in Steps 1 and 5 (Eq. 24) and Step 2 (Eq. 18) freeze g, r, p_r during the substep; these are not exact Hamiltonian flows of any F1 or F2, since the exact flow of F1 = −g ln φ would also evolve p_r and p_θ through derivatives of g and φ. Thus AS2 is a composition of non-Hamiltonian maps; fixed h in s does not confer symplecticity. The numerical energy conservation is plausible empirical behavior, but the central claim of symplecticity and the correctness of the adaptive mechanism are not established; the chaos conclusion for Orbit 1 in Schwarzschild-Melvin could be an artifact of integrating a different, non-Hamiltonian system.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an explicit adaptive-time-step integrator AS2 for Hamiltonian geodesic motion in curved spacetimes. The construction combines the authors' earlier time-transformed explicit symplectic schemes with the Preto–Saha auxiliary momentum Φ: the Hamiltonian F = K/Φ + g ln(Φ/φ), with φ = j/r, is split as F = (1/Φ)Σ K_i + F1 + F2, and a symmetric composition AS2(h) = F1(h/2) F2(h/2) S2(h/Φ) F2(h/2) F1(h/2) is used with fixed step h in the new time s and adaptive steps in the old time τ. The paper claims that AS2 is symplectic in the extended phase space, costs only two extra scalar updates per step compared to S2, and typically gives two orders of magnitude smaller Hamiltonian errors. Tests are presented for Schwarzschild, Kerr, and Schwarzschild–Melvin spacetimes, including particle and photon orbits, and the method is applied to detect weak chaos in a Schwarzschild–Melvin orbit that appears regular under S2.","tokens_in":26639,"tokens_out":5178,"duration_ms":52037,"significance":"A simple, cheap, adaptive explicit symplectic integrator for curved spacetimes would be a valuable tool for long-term geodesic integrations, ray tracing, and chaos studies. The numerical demonstrations show consistent improvements in Hamiltonian-error size for the reported orbits, and the method is broad in principle because it inherits the splitting compatibility of the earlier S2 schemes. However, the central theoretical claim — that AS2 is symplectic and that the adaptive mechanism follows from Hamilton's equations — is not valid as derived, and the numerical evidence alone does not establish symplecticity. Since the title, abstract, and the main novelty of the paper rest on this claim, the contribution's central foundation needs to be reassessed.","major_comments":[{"comment":"The derivation of the adaptive mechanism is inconsistent with Hamiltonian mechanics. With φ = j/r, the Hamiltonian F in Eq. (15) has no explicit dependence on the coordinate τ because r and θ are independent phase-space coordinates; therefore ∂F/∂τ = 0 and Hamilton's equation gives dΦ/ds = 0, not the nonzero expression in Eq. (16). The manuscript obtains dΦ/ds = g d ln φ/dτ by differentiating φ along the orbit, which is a total derivative along the solution, not a partial derivative of F at fixed phase-space point. This invalidates Eqs. (19)–(21) and the crucial relation Φ/φ = 1, and hence the step-size adaptation mechanism itself is not a consequence of the Hamiltonian F.","section":"§2.2, Eq. (16)"},{"comment":"The maps labeled F1(h/2) and F2(h/2) in Eq. (26) are not exact Hamiltonian flows of the sub-Hamiltonians defined by F1 = −g ln φ and F2 = g ln Φ. The implementation in Steps 1, 2, 4, and 5 freezes r, θ, and the momenta p_r, p_θ during the substeps governed by these terms, so that (for example) the update of Φ in Eq. (24) treats g, r, and p_r as constants. The exact flow of −g(r,θ) ln φ would also evolve r and θ through derivatives of g and φ. Consequently AS2(h) is a composition of non-Hamiltonian maps and is not a symplectic splitting of F; the assertion in Point 4 that 'the integrator AS2 remains symplectic' is therefore unsupported.","section":"§2.2, Steps 1–5 and Eq. (24)"},{"comment":"The claimed old-time/new-time relation dτ = (r/j) g ds depends on the combination of Eq. (17) and Eq. (21). Since Eq. (21) relies on the invalid Eq. (16), the derivation of Eq. (27) does not follow from Hamilton's equations for F. Even if the adaptive update is retained as a time-step control heuristic, the paper does not provide a symplectic interpretation of the resulting map, and the method's good numerical energy behavior cannot be attributed to preservation of the canonical structure of F.","section":"§2.2, Eq. (27)"}],"minor_comments":[{"comment":"The phrase 'gives place to another form' should be 'gives way to another form,' and the sentence containing 'j ≥ rmax is possibly admitted' is unclear about whether j = rmax is allowed.","section":"§1, paragraph after Eq. (14)"},{"comment":"Footnote 5 acknowledges the nontrivial relation between r, θ, and τ in φ, but the coordinate dependence is exactly the point that breaks Eq. (16); the footnote does not resolve the inconsistency and should be expanded or removed.","section":"§2.2, footnote 5"},{"comment":"The orbits referred to as Orbit 1 through Orbit 7 are not explicitly defined in the text or figure; the description of which curves correspond to which initial radii would help reproducibility.","section":"§3.3.1, Figure 5a"},{"comment":"The tables list only orders of magnitude for the Hamiltonian error, without stating the norm or the exact time at which the error is measured (except for the final time); a precise definition would make the comparisons more reproducible.","section":"Tables 1 and 2"},{"comment":"The reference list contains formatting inconsistencies, such as 'Virbhadra1, K. S.' and 'Kop ´aˇcek, O.', and several entries lack complete page numbers or article numbers; these should be corrected before publication.","section":"References"}],"recommendation":"reject","confidential_remarks":"The central mathematical claim of the paper is not valid: the update equation for Φ that drives the adaptive time step is not a Hamilton equation for F because F has no τ-dependence in the extended phase space. This is not a minor technical gap but the mechanism that makes the method adaptive and the basis for the symplecticity claim. Unless the authors can place the method on a correct symplectic footing — or explicitly resubmit as a non-symplectic adaptive integrator with a different theoretical justification — the paper's central contribution is not established. The numerical experiments appear internally consistent but do not compensate for this formal issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. The method is a neat combination of Preto-Saha's Phi-device with the authors' earlier time-transformed splits for curved spacetimes, and the numerical tests are thorough and internally consistent. But the symplecticity claim does not hold as written: with phi = j/r, the F in Eq. (15) has no explicit tau-dependence, so Hamilton's equation gives dPhi/ds = 0, not Eq. (16).\n\nWhat is genuinely new: applying the Preto-Saha step-size control to time-transformed Hamiltonians for black-hole spacetimes, with phi = j/r acting as a scaling factor that adjusts the old-time step. The insight that r/j should only rescale time and stay decoupled from the equations of motion is sensible, and the implementation cost is genuinely low: two extra scalar updates per step. The energy-error results are also clean: errors bounded over 10^4-10^5 steps, about two orders of magnitude better than S2 at fixed step count, with transparent reporting of the j-dependence.\n\nThe load-bearing problem is Eq. (16). In the extended phase space, tau is an independent coordinate and r and theta are separate coordinates. phi = j/r is a function of r, so the partial derivative dF/dtau vanishes and Phi would remain constant. The paper obtains dPhi/ds from dr/dtau along the geodesic, which is a total derivative along the solution, not a partial derivative at a fixed phase-space point. If phi is instead treated as an explicit function of tau, then F is no longer the Hamiltonian whose K/Phi part is solved by S2(h/Phi), and the F1 flow would also evolve p_r and p_theta, which Steps 1 and 5 do not do. So AS2 is a composition of non-Hamiltonian maps; a fixed step h in the new time does not make it symplectic. The numerical energy behavior is encouraging empirical behavior, but the central claim is not established. Consequently, the chaos classification for Orbit 1 in Schwarzschild-Melvin rests on a method whose symplectic credentials are in doubt, though the agreement with Li and Wu (2019) is some comfort.\n\nWho is this for: people who integrate geodesics in non-separable spacetimes and care about long-term conservation. The method might be repairable by deriving a proper time-dependent Hamiltonian or by honestly presenting it as an adaptive non-symplectic scheme with strong numerical evidence. As written, the main theoretical claim is not backed by the math. I would still send it to peer review: the flaw is subtle and the empirical part is worth refereeing, but the referee should be explicitly asked to check Eq. (16) and the claimed splitting. If the authors can fix the symplecticity argument, the method could be genuinely useful.","headline":"The adaptive scheme is clever and the numerics look good, but the symplecticity argument fails: Eq. (16) treats an orbit-level total derivative as a Hamilton partial derivative, so the central claim is unproven.","tokens_in":27370,"tokens_out":4698,"would_cite":false,"duration_ms":46044,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65P10","37M15","83C57"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that a small symmetric modification of a time-transformed explicit symplectic integrator makes its physical step size adaptive while preserving the scheme's symplectic structure.","keywords":["explicit symplectic integrator","adaptive time steps","time-transformed Hamiltonian","curved spacetimes","black hole geodesics","Schwarzschild-Melvin","Hamiltonian splitting","backward ray tracing"],"falsifier":"Integrate a simple Schwarzschild geodesic with an independent high-accuracy solver and compare $d\\Phi/ds$ from Eq. (24) with $-\\partial F/\\partial\\tau = g\\, d\\ln(j/r)/d\\tau$ evaluated from that independently computed $r(\\tau)$; a systematic discrepancy at first order in the step would settle that the adaptive flow is not the Hamiltonian flow of $F$.","tokens_in":25981,"feed_emoji":"🕳️","tokens_out":7707,"duration_ms":66808,"temperature":0.7,"pith_summary":"The paper claims that a small symmetric modification of an existing time-transformed explicit symplectic integrator makes its step size adapt to the local spacetime geometry while the symplectic character of the method is preserved. The modification introduces an auxiliary momentum $\\Phi$ conjugate to the old time $\\tau$, so the effective physical step becomes $d\\tau = (r/j) g\\, ds$, shrinking near a black-hole horizon and stretching at large radius. Compared with the nonadaptive scheme $S_2$, the new scheme $AS_2$ costs only two extra scalar updates per step and, in the reported tests, produces Hamiltonian errors typically two orders of magnitude smaller at the same step count. If the construction holds, adaptive stepping becomes available for every curved spacetime whose Hamiltonian, or time-transformed Hamiltonian, splits into explicitly integrable parts, with direct uses in long-term geodesic studies and black-hole ray tracing.","feed_headline":"Adaptive steps shrink black-hole orbit errors by 100x","feed_subtitle":"Two extra scalar updates per step preserve the integrator's structure and reveal chaos S2 hides.","key_machinery":"The construction is carried by the split Hamiltonian $F = K/\\Phi + g\\ln(\\Phi/\\varphi)$ with $\\varphi = j/r$, together with the composition $AS_2(h) = F_1(h/2)\\, F_2(h/2)\\, S_2(h/\\Phi)\\, F_2(h/2)\\, F_1(h/2)$. Here $\\Phi$ is the momentum conjugate to the old time $\\tau$, treated as an additional coordinate; $F_1$ and $F_2$ are the flows of $-g\\ln\\varphi$ and $g\\ln\\Phi$, and $S_2$ is the existing second-order symmetric integrator built from explicitly integrable sub-Hamiltonians of $K$. The two logarithmic terms isolate the step-size control in two scalar flows, so the adaptive integrator needs only two extra updates per step, and the frozen-$\\Phi$ solve of $S_2(h/\\Phi)$ yields the time-step law $d\\tau = (r/j) g\\, ds$. This split is what makes adaptive stepping compatible with explicit symplectic integration instead of requiring an implicit solve.","core_discovery":"The paper's central claim is that the extended-phase-space Hamiltonian $F = K/\\Phi + g\\ln(\\Phi/\\varphi)$, with $\\varphi = j/r$, supports an explicit second-order symplectic integrator with adaptive steps in the original time. Writing $F_1 = -g\\ln\\varphi$ and $F_2 = g\\ln\\Phi$, the symmetric composition $AS_2(h) = F_1(h/2)\\, F_2(h/2)\\, S_2(h/\\Phi)\\, F_2(h/2)\\, F_1(h/2)$ advances the spatial coordinates with the existing integrator $S_2$ while $\\Phi$ and $\\tau$ receive half-step scalar updates at the edges. $\\Phi$ is frozen during the spatial solve and advanced by $\\Phi(s) = \\Phi_0 - s\\, g\\, g^{rr} p_r / r$, acting only as a rescaling of the time step, which keeps the implementation cheap. Because the step $h$ in the new time $s$ is fixed, the paper argues that the symplectic structure is preserved, while the old-time step $d\\tau = (r/j) g\\, ds$ varies with the orbit. Tests on Schwarzschild, Kerr, and Schwarzschild-Melvin spacetimes, for both particles and photons, show Hamiltonian errors about two orders smaller than $S_2$, and the two methods disagree on one Schwarzschild-Melvin orbit that $AS_2$ labels weakly chaotic.","pith_inferences":["If the derivation of Eq. (16) is legitimate, the same frozen-$\\Phi$ trick should compose with higher-order symmetric integrators, giving adaptive versions of fourth- and sixth-order explicit symplectic schemes.","A practical extension would be an automatic or adaptive choice of $j$, since the paper's recommended range $(r_{\\min}+r_{\\max})/2 \\le j \\le r_{\\max}$ is orbit-dependent and trading accuracy against the $\\tau$-$s$ drift is currently a manual decision.","The $S_2$-versus-$AS_2$ disagreement on Orbit 1 is a testable warning: re-running earlier chaotic-transition scans near Schwarzschild-Melvin horizons with an adaptive scheme may shift apparent critical magnetic-field strengths."],"forward_implications":["Long-term integrations of particle and photon geodesics in non-integrable spacetimes can use variable physical step sizes without giving up an explicit symplectic integrator.","For the Schwarzschild-Melvin parameters tested, $AS_2$ changes the inferred dynamics of one orbit from regular to weakly chaotic, so published portraits of chaos computed with nonadaptive $S_2$ may need re-examination near black-hole horizons.","The method inherits the applicability of $S_2$: any spacetime whose Hamiltonian, or time-transformed Hamiltonian, splits into explicitly integrable terms can use $AS_2$ with no structural changes.","Ray-tracing codes can choose $j$ near the observer distance, obtaining small steps near the photon sphere where shadow structure is decided and larger steps far away."],"supporting_citations":[{"why":"Supplies the auxiliary momentum $\\Phi$ and the step-size-control Hamiltonian that the adaptive scheme adapts to curved spacetimes.","marker":"Preto & Saha 2009"},{"why":"Gives the time-transformation function and five-term split that make $S_2$ explicit for Kerr-type Hamiltonians.","marker":"Wu et al. 2021"},{"why":"Provides time-transformation functions and splits for the family of spacetimes to which $AS_2$ is said to apply.","marker":"Wu et al. 2022"},{"why":"Supplies the direct four-term split for the magnetized Schwarzschild test problem and the baseline nonadaptive $S_2$ performance.","marker":"Wang et al. 2021a"},{"why":"Establishes the time-transformation idea for explicit symplectic schemes on which the split Hamiltonian $K$ is built.","marker":"Mikkola 1997"},{"why":"Derives the adaptive-stepsize Hamiltonian formalism that motivates the $\\Phi$ construction.","marker":"Emel'yanenko 2007"}],"fun_headline_variants":["Adaptive symplectic steps tame black-hole orbits","Cheap adaptive integrator exposes hidden chaos","Two extra steps cut spacetime orbit errors","Adaptive integrator for curved spacetimes, no fuss","Time-transformed symplectic methods go adaptive"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the orbit-level change of $\\varphi = j/r$ can be treated as an explicit partial derivative of the Hamiltonian with respect to the time coordinate, as Eq. (16) does; if that step is not a legitimate Hamiltonian equation, the method's symplectic property is unproved.","fun_headline_variants_meta":{"raw":{"variants":["Adaptive symplectic steps tame black-hole orbits","Cheap adaptive integrator exposes hidden chaos","Two extra steps cut spacetime orbit errors","Adaptive integrator for curved spacetimes, no fuss","Time-transformed symplectic methods go adaptive"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000184,"raw_usage":{"total_tokens":1400,"prompt_tokens":1108,"completion_tokens":292,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":724,"completion_tokens_details":{"reasoning_tokens":220}},"tokens_in":724,"tokens_out":292,"duration_ms":3440,"temperature":1.0,"reasoning_tokens":220,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:47:54.460173+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Integrate a simple Schwarzschild geodesic with an independent high-accuracy solver and compare $d\\Phi/ds$ from Eq. (24) with $-\\partial F/\\partial\\tau = g\\, d\\ln(j/r)/d\\tau$ evaluated from that independently computed $r(\\tau)$; a systematic discrepancy at first order in the step would settle that the adaptive flow is not the Hamiltonian flow of $F$.","supporting_citations":[{"cited_title":"2009, ApJ, 703, 1743","cited_arxiv_id":null,"evidence_quote":"Supplies the auxiliary momentum $\\Phi$ and the step-size-control Hamiltonian that the adaptive scheme adapts to curved spacetimes."},{"cited_title":"1997, Celest","cited_arxiv_id":null,"evidence_quote":"Establishes the time-transformation idea for explicit symplectic schemes on which the split Hamiltonian $K$ is built."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Derives the adaptive-stepsize Hamiltonian formalism that motivates the $\\Phi$ construction."}],"review_version":1}