{"id":"0f971a50-e2f4-4e8f-b797-2682b3e752f1","arxiv_id":"2506.02407","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A locally updated particle-in-cell scheme that maintains discrete Gauss's law exactly and avoids Poisson or Ampère solves is derived and tested on Landau damping, two-stream instability, and diocotron benchmarks.","lead":"The paper proposes a particle-in-cell plasma simulation update that changes the electric field locally at the same time particles move, so the discrete Gauss's law stays satisfied to machine precision without solving a global Poisson equation. If it works as claimed, large parallel plasma simulations could avoid the usual expensive global field solve while keeping long-term charge consistency.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Step-1 field update formulas (3.6)-(3.11) are inconsistent with the discrete Gauss law (3.3) by a factor h^2; Theorem 3.1 is not established for the printed algorithm.","rationale":"The reader's weakest assumption was convergence and linear-complexity scaling of the curl-free relaxation in Step 2. That is a legitimate gap, but the more load-bearing defect is in Step 1: the displayed equations (3.6)-(3.11) are inconsistent with the displayed proof of Theorem 3.1 by a factor of h^2. Theorem 3.1 is the central claim, and its Step-1 algebra is what actually certifies exact discrete Gauss-law preservation. A missing convergence proof can in principle be supplied from the cited preprint [32], but an unexplained h^2 in the Gauss-law update means the equations as printed do not define a Gauss-law-preserving field update. The numerical tests show machine-precision Gauss-law residuals, so the authors' implementation evidently uses a consistent normalization; the manuscript simply does not state it. This is fixable by correcting the normalization or the formulas, and it does not demonstrate that the underlying method is false. The same holds for the relaxation-convergence gap, so conditional acceptance remains the appropriate verdict. I partially agree with the reader: the normalization inconsistency was noted in the reader's rationale, but I regard it as more central than the Step-2 convergence concern.","tokens_in":18886,"tokens_out":8446,"duration_ms":80920,"concrete_test":"Run a single-particle two-cell periodic test with h ≠ 1 and λ = 1: initialize E = 0 and a uniform neutralizing density, move the particle by a fraction of a cell using (3.4), apply the printed update (3.6), and evaluate the discrete Gauss-law residual (3.3) before and after the update. If the residual changes by O(w_p S_x S_y (1 - 1/h^2)) instead of round-off, the printed update is incorrect; then repeat with δE = -(h/λ^2) w_p[-S_x(d^m_{x,i})] S_y(d^m_{y,j}) and confirm machine-precision Gauss-law preservation with the corrected normalization.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Substitute the printed x-substep update into discrete Gauss law (3.3). With shape functions normalized as in (2.8)/(2.11), Eq. (3.6) gives δE_{i+1/2,j} = -(1/(λ^2 h)) w_p[-S_x(d^m_{x,i})] S_y(d^m_{y,j}). The induced change in λ^2 times the divergence at vertex (i,j) is λ^2 δE/h = h^{-2} w_p S_x S_y, whereas the density change caused by the particle leaving the node is -w_p S_x S_y. The residual is therefore changed by (1 - 1/h^2) w_p S_x S_y, which is not zero for h ≠ 1. The proof of Theorem 3.1 instead effectively uses λ^2 δE/h = -w_p[-S_x]S_y, which corresponds to δE = -(h/λ^2) w_p[-S_x]S_y. Thus the printed equations and the proof silently use different normalizations of the shape function. Since Step 1 of Algorithm 3.1 is exactly the part certified by Theorem 3.1, the central exact-Gauss-law claim is not established for the algorithm as written. This issue is independent of the relaxation-convergence question: if E* already violates Gauss's law, Step 2, which preserves the Gauss-law residual, cannot repair it.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a structure-preserving particle-in-cell scheme for electrostatic Vlasov-Ampère/Vlasov-Poisson simulations. The electric field is updated locally as particles move, by splitting each particle displacement into x and y substeps and adjusting edge fields so that the discrete Gauss law remains satisfied; a subsequent local relaxation minimizes the electrostatic free energy over fields satisfying Gauss's law, aiming to recover a curl-free field without solving Poisson or Ampère equations. The main claims are Theorem 3.1 (exact discrete Gauss-law preservation up to round-off), linear complexity per time step, and good behavior on Landau damping, two-stream instability, and diocotron instability. The exposition is clear and the algorithm is easy to implement, and the paper includes comparisons against Ampère-based and Poisson-based PIC updates.","tokens_in":19073,"tokens_out":16045,"duration_ms":136422,"significance":"If correct, the proposed local update would be a valuable addition to PIC methodology: it avoids global elliptic solves, is compatible with local and parallel implementations, and enforces Gauss's law intrinsically rather than by post-hoc correction. The paper contains no fitted parameters, presents an explicit algorithm, and tests standard benchmarks. However, the central theorem rests on a normalization inconsistency in the printed update formulas, and the convergence and complexity of the curl-free relaxation stage are delegated to a same-author preprint. The contribution is therefore promising but not yet established by the manuscript as written.","major_comments":[{"comment":"With the normalization S(x)=max{0,1−|x|/h}/h stated in §2.2, the products Sx Sy scale as h^{-2} and the field increments in Step 1 are too small by a factor h^2. For the boundary-crossing case of Fig. 1(c), the density at node (i,j) changes by Δn_{i,j} = −ω_p Sx(d^m_{x,i})Sy(d^m_{y,j}); equation (3.3) requires λ²δE_{i+1/2,j}/h = −Δn_{i,j}, which gives δE = +(h/λ²)ω_p Sx(d^m_{x,i})Sy(d^m_{y,j}). The printed update (3.6) instead gives δE = +(1/(λ²h))ω_p Sx(d^m_{x,i})Sy(d^m_{y,j}), which is h^2 smaller. The same factor appears in (3.7)–(3.11) and in the within-cell formulas in §3.2. The proof of Theorem 3.1 silently uses the h/λ² normalization when it writes λ²δE/h = −ω_p[−Sx]Sy; hence the theorem is not established for Algorithm 3.1 as written. Since Step 2 preserves the Gauss-law residual, it cannot repair an inconsistency created by Step 1. Please replace 1/(λ²h) by h/λ² in all Step-1 updates (or change the shape-function normalization consistently) and re-verify the numerical results in §4.","section":"§3.2, Eqs. (3.6)–(3.11); proof of Theorem 3.1"},{"comment":"The manuscript does not demonstrate the convergence of the local curl-free relaxation or the claimed linear complexity. Step 2 loops over all cells, rows, and columns; each sweep is O(N²), so the total cost per time step is O(R_s N²), where R_s is the number of iterations. Figures 4(d) and 6(d) plot R_s only versus physical time for fixed N; no study of R_s versus N or versus tolerance is provided. The text delegates the convergence and the Gauss-law-preserving property of Step 2 to the same-author preprint [32]. Since the linear-complexity advantage over a Poisson solver is a central selling point, the paper needs a convergence-rate estimate, or at least a numerical scaling study of R_s as a function of N, together with a self-contained or published proof for the relaxation stage.","section":"§3.3, Algorithm 3.1, Step 2"}],"minor_comments":[{"comment":"On the right-hand side of (3.10) and (3.11), E_{i+1/2,j} and E_{i+1/2,j+1} should be E_{l+1/2,j} and E_{l+1/2,j+1}; as printed, the update writes into the initial edge rather than the final edge.","section":"§3.2, Eqs. (3.10)–(3.11)"},{"comment":"For a periodic domain with N unique grid points, the relation h=L/(N−1) is inconsistent with the N×N Yee mesh and the loops over i,j=0,...,N−1; the conventional periodic spacing is h=L/N (or the indexing should count intervals rather than points). Please clarify.","section":"§3.1"},{"comment":"There is a typographical error: 'supscript' should be 'superscript'. Also, the minimum in (3.15) should be written with parentheses: η = −(∑_{i=0}^{N-1} E^{(q)}_{i+1/2,b})/N.","section":"§3.3"},{"comment":"The proof says the within-cell case is omitted 'for brevity'; since the within-cell update is one of the two cases covered by the theorem and is affected by the same normalization issue, it should be included or moved to an appendix.","section":"Proof of Theorem 3.1"},{"comment":"The numerical section does not state the mesh spacing h used in the runs. Given the h^2 discrepancy identified above, reporting h and the exact shape-function normalization in the implementation is necessary for reproducibility.","section":"§4"}],"recommendation":"major_revision","confidential_remarks":"To the editor: the h^2 inconsistency in §3.2 is the kind of error that may be a typographical normalization slip, but it affects the paper's main theorem and the reproducibility of §4. If the authors confirm that the code uses h/λ², the paper can likely be repaired. I would also ask whether [32] will be available in published form before this manuscript; as it stands, a load-bearing proof depends on an unpublished same-author preprint. The idea is appealing and worth a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The reader's take is close to mine, but I want to sharpen one point: the paper's central exact-Gauss-law claim is not actually established by the text as printed. The idea itself is worth engaging with. To my knowledge, no prior PIC method updates E locally along dimension-split particle paths and then relaxes to curl-free on the Gauss-law manifold, so the framework is a genuine within-field novelty. The numerical tests (Landau, two-stream, Diocotron) are appropriate, compare against Poisson and Ampère references, and show Gauss-law residuals at machine precision, which is good evidence that the intended algorithm works. The paper also does a fair job positioning itself relative to Boris/Marder/Langdon and current-conserving schemes.\n\nThe soft spots are real, and one is load-bearing. First, substituting the printed Step-1 update (3.6)-(3.11) into the discrete Gauss law (3.3) gives a factor-h^2 mismatch given the stated normalization of the shape function (S(x)=max{0,1-|x|/h}/h). The printed δE changes λ^2·div E by w_p S_x S_y / h^2, while the particle leaving the node changes the density by -w_p S_x S_y. The residual is changed by (1 - h^{-2}) w_p S_x S_y, not zero. The proof of Theorem 3.1 silently uses δE = (h/λ^2) w_p S_x S_y, which is different. So as written, Theorem 3.1 does not cover Algorithm 3.1. Since Step 2 preserves the Gauss-law residual, it cannot repair this. If the code follows the proof, the formulas need correcting; if it follows the formulas, the tests should not show machine-precision Gauss-law preservation. This is exactly the kind of thing a referee should require resolved.\n\nThe second soft spot is the reliance on the same-author preprint [32] for convergence and Gauss-law preservation of the curl-free relaxation step. That may be mathematically fine, but it means the linear-complexity claim has no proof in this paper; the iteration counts in Figures 4(d) and 6(d) are empirical. This is a smaller issue, fixable by stating assumptions and a convergence-rate bound or by including a scaling test.\n\nFor a serious referee: yes. The idea is novel enough and the tests are good enough that the paper deserves review, but the normalization inconsistency and the missing relaxation analysis must be addressed before publication.","headline":"The core idea is a genuine new direction for PIC, but the printed Step-1 field update is inconsistent with the stated shape-function normalization, so the exact-Gauss-law claim needs a correction before it can be trusted.","tokens_in":19673,"tokens_out":4578,"would_cite":false,"duration_ms":42973,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65Z05","65M75","35Q83"],"pacs":["52.65.Rr"],"model":"deepseek-v4-flash","headline":"The paper proposes GP-PIC, a particle-in-cell update that keeps the discrete Gauss's law exact to machine precision at every time step while updating the electric field locally and avoiding Poisson or Ampère solvers.","keywords":["Gauss's law preservation","particle-in-cell method","Vlasov-Ampère equations","Vlasov-Poisson equations","local relaxation","curl-free electric field","charge conservation","plasma simulation"],"falsifier":"Run the relaxation alone on an $N\\times N$ mesh with a manufactured charge distribution, starting from a field that satisfies Gauss's law but has a known nonzero circulation, and count the sweeps needed to reduce the curl residual below a tolerance $\\tau$ as both $N$ and $\\tau$ vary. If the sweep count grows faster than a modest constant independent of $N$, or if the residual stalls before reaching $\\tau$, then the linear-complexity and curl-free-accuracy premises of the method fail; separately, a long-time double-precision run in which the Gauss residual grows above round-off would refute Theorem 3.1.","tokens_in":18609,"feed_emoji":"⚡","tokens_out":13173,"duration_ms":128260,"temperature":0.7,"pith_summary":"This paper tries to establish that a particle-in-cell method can keep the discrete Gauss's law exact at every time step without solving the Poisson or Ampère equations. The field is updated locally as particles move, with particle motion split along mesh directions so the edge fields crossed by the trajectory absorb exactly the charge that moves. A separate local relaxation then pushes the field toward the curl-free state while never leaving the Gauss-law manifold. If the construction is right, the global electrostatic solve that dominates conventional PIC cost disappears, and long-time simulations do not accumulate the Gauss-law error that degrades Ampère-based schemes. Numerical tests on Landau damping, two-stream instability, and diocotron instability show the new method tracking a Poisson-based reference while an Ampère-based comparison drifts.","feed_headline":"Keep Gauss's law exact in plasma runs without Poisson solves","feed_subtitle":"A local field update tracks particle charge cell by cell, so long-term plasma runs stay accurate at round-off precision.","key_machinery":"The central object is the discrete Gauss-law manifold on a Yee mesh, the set of edge electric fields whose vertex divergences match the deposited charge. The method stays on it through two local operations. First, the trajectory flux update splits particle motion into mesh-parallel sub-steps and modifies exactly those edge fields crossed by the trajectory, using the partition-of-unity identity for the shape function so every vertex's charge balance changes consistently. Second, the free-energy relaxation applies cell rotations, adding $\\eta$ to two opposite edges and $-\\eta$ to the other two with $\\eta = -(E^{(q)}_{i+1/2,j} - E^{(q)}_{i+1/2,j+1} + E^{(q)}_{i+1,j+1/2} - E^{(q)}_{i,j+1/2})/2$, and row/column line shifts with $\\eta$ equal to the negative mean field. Each operation preserves all vertex fluxes, so Gauss's law holds after every sweep, and iterating them minimizes $\\lambda^2 \\int E^2/2\\, dx$ on the manifold, whose unique minimizer is the curl-free field. All of this is local, so one pass over the mesh costs linear work.","core_discovery":"For the electrostatic Vlasov system on a Yee mesh, the authors construct Algorithm 3.1 (GP-PIC). In Step 1, each particle's displacement is split into an x-move and a y-move; after each move the electric-field components on the cell edges crossed by the particle are updated, with increments fixed by the shape function, so that the discrete divergence at every affected vertex changes exactly by the deposited charge. Step 2 relaxes the resulting field, which already satisfies Gauss's law, by sweeping cells and row/column lines with increments chosen to minimize the electrostatic free energy $\\lambda^2 \\int E^2/2\\, dx$; each sweep preserves the flux balance at every vertex. Theorem 3.1 asserts that if Gauss's law held at the start of the step, it holds at the end up to round-off error, while the curl-free condition is approached only to the relaxation tolerance $\\tau$. This is the paper's central claim: Gauss's law is intrinsic to the update, not restored by a correction or engineered current deposition.","pith_inferences":["The paper does not test how the number of relaxation sweeps scales with mesh size and tolerance; a scaling study would determine whether the per-step cost stays linear in practice, since low-frequency modes on large meshes typically slow iterative relaxations.","In the two-stream experiment, the results at tolerances $10^{-4}$ and $10^{-7}$ suggest that Gauss-law compliance matters more for long-time phase-space fidelity than a tightly curl-free field; a targeted test with a very loose tolerance would make that dependence explicit.","The Step 1 update is a discrete path sum of shape-function differences, which suggests the same charge-tracking idea could be carried to higher-order or non-product shape functions and to unstructured meshes, if a Gauss-preserving local smoother replaces the cell/line relaxation."],"forward_implications":["PIC simulations can maintain Gauss's-law residuals at machine precision over long runs without any Poisson solve, divergence cleaning, or charge-conserving current-deposition scheme.","Because each time step uses only local operations and linear work per sweep, the method removes the global field solve that typically dominates PIC cost and is a natural candidate for massively parallel electrostatic simulations.","The split-motion field update is not tied to the leapfrog integrator; the paper argues it can be combined with other temporal discretizations, with energy-conserving velocity corrections, and with dimension splitting in three dimensions.","The relaxation tolerance $\\tau$ controls only how close the field is to curl-free; the Gauss-law guarantee holds at any tolerance, so users can trade curl-free accuracy for speed without violating the charge constraint.","In the reported benchmarks the method reproduces the Poisson-reference damping and growth rates, whereas the Ampère-with-interpolated-current comparison shows growing Gauss-law residuals, total-energy drift, and distorted phase-space vortices."],"supporting_citations":[{"why":"Supplies the finite-difference local relaxation algorithm and the proof that its sweeps stay on the Gauss-law manifold and converge to the curl-free minimizer; Theorem 3.1's Step 2 depends on this result.","marker":"[32]"},{"why":"Introduced the local free-energy relaxation strategy for Coulomb interactions that the paper adapts to obtain a linear-complexity, Poisson-free field update.","marker":"[38]"},{"why":"Establishes the Maxwell-Ampère Nernst-Planck setting in which constrained free-energy minimization with local relaxation yields the curl-free field, the basis for Step 2.","marker":"[49]"},{"why":"Introduced the row/column line-shift correction used in Step 2 to further reduce the free energy while preserving all vertex flux balances.","marker":"[26]"},{"why":"Supplies the energy- and charge-conserving implicit electrostatic PIC framework that motivates splitting particle motion into sub-steps for exact charge conservation, the template for Step 1.","marker":"[8]"},{"why":"Cited together with [8] for the sub-step splitting strategy that keeps discrete charge conservation while particles cross cells.","marker":"[6]"},{"why":"Developed the local algorithm for Coulomb interactions whose per-step linear complexity the proposed GP-PIC update inherits.","marker":"[37]"}],"fun_headline_variants":["PIC that keeps Gauss's law exactly, no Poisson solve needed","Local PIC update enforces Gauss's law at round-off precision","Charge-conserving PIC without Poisson or Ampère solves","Intrinsic Gauss's law in PIC: local updates, no correction steps","Exact Gauss's law in PIC via split-motion edge updates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the local relaxation in Step 2 converges to the unique curl-free minimizer on the Gauss-law manifold with a bounded number of sweeps per time step; the convergence proof is delegated to the companion preprint, and the paper does not analyze how the sweep count scales with mesh size or tolerance.","fun_headline_variants_meta":{"raw":{"variants":["PIC that keeps Gauss's law exactly, no Poisson solve needed","Local PIC update enforces Gauss's law at round-off precision","Charge-conserving PIC without Poisson or Ampère solves","Intrinsic Gauss's law in PIC: local updates, no correction steps","Exact Gauss's law in PIC via split-motion edge updates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000451,"raw_usage":{"total_tokens":2299,"prompt_tokens":999,"completion_tokens":1300,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":615,"completion_tokens_details":{"reasoning_tokens":1211}},"tokens_in":615,"tokens_out":1300,"duration_ms":9307,"temperature":1.0,"reasoning_tokens":1211,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T11:25:24.862573+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the relaxation alone on an $N\\times N$ mesh with a manufactured charge distribution, starting from a field that satisfies Gauss's law but has a known nonzero circulation, and count the sweeps needed to reduce the curl residual below a tolerance $\\tau$ as both $N$ and $\\tau$ vary. If the sweep count grows faster than a modest constant independent of $N$, or if the residual stalls before reaching $\\tau$, then the linear-complexity and curl-free-accuracy premises of the method fail; separately, a long-time double-precision run in which the Gauss residual grows above round-off would refute Theorem 3.1.","supporting_citations":[{"cited_title":"Finite-Difference Approximations and Local Algorithm for the Poisson and Poisson-Boltzmann Electrostatics","cited_arxiv_id":"2409.15796","evidence_quote":"Supplies the finite-difference local relaxation algorithm and the proof that its sweeps stay on the Gauss-law manifold and converge to the curl-free minimizer; Theorem 3.1's Step 2 depends on this result."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduced the local free-energy relaxation strategy for Coulomb interactions that the paper adapts to obtain a linear-complexity, Poisson-free field update."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the Maxwell-Ampère Nernst-Planck setting in which constrained free-energy minimization with local relaxation yields the curl-free field, the basis for Step 2."},{"cited_title":"Jiang and Z","cited_arxiv_id":null,"evidence_quote":"Introduced the row/column line-shift correction used in Step 2 to further reduce the free energy while preserving all vertex flux balances."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the energy- and charge-conserving implicit electrostatic PIC framework that motivates splitting particle motion into sub-steps for exact charge conservation, the template for Step 1."},{"cited_title":"Chac´ on and G","cited_arxiv_id":null,"evidence_quote":"Cited together with [8] for the sub-step splitting strategy that keeps discrete charge conservation while particles cross cells."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Developed the local algorithm for Coulomb interactions whose per-step linear complexity the proposed GP-PIC update inherits."}],"review_version":1}