{"id":"b355e10b-7f7f-41ce-ad05-bc0396c1c10a","arxiv_id":"2411.13175","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Fourth-order transparent boundary conditions (D4TBCs and aDTBCs) yield globally fourth-order accurate, uniquely solvable finite difference discretizations of the 1D Schrödinger-Poisson problem.","lead":"The paper develops fourth-order accurate boundary conditions for the 1D Schrödinger-Poisson equation, the workhorse model for nanoscale quantum device simulation. It proves the discrete schemes have unique solutions and demonstrates fourth-order convergence on simulations of resistors and resonant tunneling diodes.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Uniqueness theorems fail at E = V0: for V≡0, E=0 the D4TBC and aDTBC systems are singular (constant null vector), contradicting Theorems 2–3; E=0 is included in the experiments' energy integral.","rationale":"The paper's central contributions — the D4TBCs as a fourth-order, essentially non-oscillating discretization for E > max{V0, VNx}, and the aDTBC framework — are valuable and numerically supported. However, the uniqueness theorems are stated more broadly than their proofs allow. The reader's weakest point was that the condition t > max{2(E−V0), 2(E−VNx)} is violated in the experiments. Recomputing t with the paper's definition (t = (ℏ²/(2m*))·(12/Δx²)) gives t ≈ 27 eV for the RTD parameters (Δx = 0.5 nm, m* = 0.067m0), so the condition actually holds for E ≤ 0.8 eV; the reader's numerical estimate appears to use an incorrect prefactor. The genuine residual weakness is the degenerate case E = V0 (or E = VNx), where α and β become 1, Im(α⁻¹) = Im(β⁻¹) = 0, and the proof of Theorem 2 collapses. A concrete constant solution shows the uniqueness claim is false as stated. This is a real mathematical gap, but it is confined to a measure-zero set in energy and is easily fixed by requiring E > max{V0, VNx} and handling the endpoint either by a limiting procedure or by excluding it from the quadrature. The numerical experiments, which use adaptive Simpson sampling, may not be materially affected if the E = 0 contribution is negligible, but this should be checked. Thus the appropriate verdict remains CONDITIONAL, with the required revision being an explicit exclusion of the degenerate equality case and a discussion of its treatment in the energy integral. The reader's verdict is therefore unchanged, though the justification differs.","tokens_in":18076,"tokens_out":21029,"duration_ms":193654,"concrete_test":"Set V ≡ 0, E = 0, Nx = 2, and assemble the (Nx+1)×(Nx+1) matrix for the homogeneous D4TBC system (46)–(48), i.e., interior equation (24) at j = 1 plus boundary equations (27)–(28). Compute its determinant or null space; it will be singular with the constant vector [1,1,1] in the kernel. This directly contradicts Theorem 2 as stated. Additionally, rerun the energy integral using a small positive lower cutoff (e.g., ε = 10⁻⁶ eV instead of 0) and compare the computed I-V curves; a material difference would indicate that the E = 0 singularity affects the reported device characteristics.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 2 states that, under t > max{2(E−V0), 2(E−VNx)}, (24) equipped with the D4TBCs has a unique solution for arbitrary V(x). This is false at the degenerate point E = V0. Take V ≡ 0 and E = 0, so t > 0 is satisfied (in the RTD experiments t ≈ 27 eV for Δx = 0.5 nm and m* = 0.067m0, using the correct prefactor t = (ℏ²/(2m*))·(12/Δx²)). Then α = β = 1, and the D4TBCs become ψ−1 = ψ0 and ψNx+1 = ψNx, while the interior equation (24) reduces to the discrete Laplacian. On a two-point mesh (Nx = 2) the system (24)+(27)+(28) has every constant vector as a solution, so the homogeneous problem has a nonzero null vector and the inhomogeneous problem has infinitely many solutions. The proof of Theorem 2 fails exactly here: for α = 1, Im(α⁻¹) = 0, so equation (53) carries no information and cannot force ψ−1 = ψNx+1 = 0. The same degeneracy affects the aDTBCs, whose Theorem 3 proof is omitted. Since the energy integral (11) is evaluated on [0, 0.8] eV, the endpoint E = 0 lies in the computational range, and the discrete Schrödinger solve at that energy is singular. The reader's concern about t being too small is not supported: with the correct t the condition t > 2(E−V0) holds comfortably for all energies used. The real gap is the equality case E = V0, which the theorems do not exclude.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript develops two families of discrete transparent boundary conditions for the one-dimensional Schrödinger equation on a bounded interval, both paired with a compact fourth-order interior finite-difference scheme. The D4TBCs are derived from the discrete dispersion relation of the interior scheme and are shown to have fourth-order consistency and to avoid the spurious oscillations seen with the simpler C4TBCs when the potential vanishes. The aDTBCs are obtained by using the exact plane-wave form of the solution outside the device, so they introduce no additional discretization error and can in principle be combined with arbitrarily high-order interior schemes. The authors prove uniqueness for the D4TBC discrete system, state an analogous theorem for the aDTBC system without proof, formulate a fourth-order discretization of the coupled Schrödinger–Poisson problem, and validate the schemes on a resistor and two resonant-tunneling diodes, comparing against NEGF benchmarks. The central numerical claims are plausible, but the uniqueness theorem for the D4TBCs is false at the degenerate point E = V0, which is included in the energy range used in the experiments; this requires a substantive revision.","tokens_in":18511,"tokens_out":9712,"duration_ms":96512,"significance":"If the technical gaps are repaired, the paper would be a useful contribution to high-order quantum-transport simulation: the D4TBC construction is parameter-free and derived in-paper from the discrete dispersion relation, the fourth-order consistency proof is explicit, and the numerical experiments against NEGF benchmarks support the practical value of the schemes. The data availability statement is a further positive feature. However, the current manuscript's headline claim that uniqueness of both discrete Schrödinger systems is \"rigorously proved\" is undermined by a concrete counterexample at E = V0, and the omitted proof of Theorem 3 leaves the aDTBC system in the same regime unsupported. Because the energy quadrature in Section 4 samples E = 0, the issue is not merely cosmetic; it affects the actual computational procedure.","major_comments":[{"comment":"Theorem 2 is false as stated at E = V0. For V ≡ 0 and E = 0, the formulas for α± and β± on page 7 give α = β = 1, so the D4TBCs (38)–(39) reduce to ψ−1 = ψ0 and ψNx+1 = ψNx, while the interior equation (24) reduces to the discrete Laplacian. On a mesh with Nx = 2 every constant vector solves the homogeneous system, so the solution is not unique. The hypothesis t > max{2(E−V0), 2(E−VNx)} is satisfied in this example (it only requires t > 0), so the counterexample lies inside the stated hypotheses. In the proof, g0 and gNx are real and α^{-1} = β^{-1} = 1, so equation (53) becomes identically 0 and cannot imply ψ−1 = ψNx+1 = 0. The theorem should be restricted to E > V0 and E > VNx, or the degenerate case must be analyzed separately. I note that the earlier concern that t is too small is not supported: with Δx = 0.5 nm and m* = 0.067m0 one has t ≈ 27 eV, comfortably above 2(E−V0) for E ≤ 0.8 eV; the actual gap is the equality E = V0.","section":"§3.2, Theorem 2"},{"comment":"Theorem 3 is stated without proof, and the omitted proof cannot be merely \"similar\" to that of Theorem 2 in the degenerate case. At E = V0 the aDTBCs (55)–(56) give e^{ik1Δx} = e^{ik2Δx} = 1, producing the same constant null vector as in the D4TBC counterexample above. Since the energy integral (11) is truncated to [0, 0.8] eV and evaluated with an adaptive Simpson rule, the endpoint E = 0 lies in the computational range; for vanishing contact potential the discrete system at that energy is singular. The authors should either provide a complete proof covering all admissible E, or explicitly restrict the uniqueness theorem to E > max{V0, VNx} and adjust the numerical quadrature so that no singular energy is sampled.","section":"§3.3, Theorem 3"},{"comment":"The numerical procedure evaluates the density integral (11) on [0, 0.8] eV using the adaptive Simpson routine, which typically samples the endpoints. At E = 0 with V(0) = V(L) = 0, the discrete Schrödinger system with either D4TBCs or aDTBCs is singular, so the linear solve at that energy is not well defined. The paper does not discuss how this energy point is handled. Please state explicitly whether an open quadrature rule, a small positive cutoff, or a limiting argument is used, and confirm numerically that the reported densities and I-V curves are insensitive to that choice.","section":"§4, Algorithm 1 and density evaluation"}],"minor_comments":[{"comment":"The nomenclature is potentially confusing: equations (27)–(28) are first introduced as C4TBCs, while equations (38)–(39) are later called D4TBCs. Please clarify at first use that these are two different boundary closures and explain why C4TBCs are abandoned in favor of D4TBCs.","section":"§3.2, page 7"},{"comment":"The notation L = max{l, s} in the discussion of (57)–(58) refers to stencil parameters l and s introduced in (17)–(18), but this connection is not restated; a brief reminder would improve readability.","section":"§3.3, page 14"},{"comment":"The abstract's phrase \"essentially non-oscillating\" is stronger than what is proved: the manuscript proves uniqueness for vanishing potential and shows in Fig. 1 that oscillations disappear in that case, while the authors themselves note in §3.3 that it is not clear whether spurious oscillations still exist when V ≠ 0. Please soften the wording accordingly.","section":"Abstract and §3.3"},{"comment":"There are several grammatical and typographical issues, e.g., \"the dispersion relation brought numerical error, so does the D4TBCs consequently\" in §3.3, and \"V /nequivalence0\" in the same section. A careful proofreading pass is needed.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The counterexample to Theorem 2 at E = V0 is decisive and should be fixed before publication. The rest of the derivation—D4TBC consistency, the aDTBC framework, and the numerical comparisons—appears sound and likely repairable, so I do not recommend rejection. Please also ask the authors to check whether the adaptive Simpson quadrature actually evaluates E = 0 and, if so, to justify the treatment of the singular endpoint. The scope of the paper is appropriate for a numerical analysis journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper deserves a serious look. The D4TBCs are a genuine fourth-order extension of Arnold's discrete transparent boundary conditions, and the aDTBC framework — using exact plane-wave relations at ghost points — is elegant and does let you push the boundary closure to arbitrary order without solving high-degree algebraic equations. The derivation of the D4TBCs is careful, Theorem 1 (global fourth-order accuracy) is supported by the Taylor analysis, and the numerical work is honest: convergence rates are reported, NEGF benchmarks are used as reference, and the data are on figshare.\n\nThe main soft spot is not the one the reader flagged. The condition t > max{2(E−V0), 2(E−VNx)} holds comfortably in the RTD experiments; the central issue is the equality case E = V0. With V ≡ 0 and E = 0 you get α = β = 1, the D4TBCs reduce to ψ−1 = ψ0 and ψNx+1 = ψNx, and the homogeneous problem has nonzero constant solutions. The proof of Theorem 2 fails exactly there: equation (53) hinges on Im(α^{-1}) and Im(β^{-1}) being nonzero, and at α = 1 they vanish. Since E = 0 is the endpoint of the energy integral (11) used in the experiments, this is not a purely academic edge case. Theorem 3 inherits the problem, and its proof is omitted, so the aDTBC uniqueness claim is doubly unsupported.\n\nOther, smaller concerns: the abstract says aDTBCs introduce no discretization error, but Fig. 1 shows spurious oscillations (smaller than C4TBCs, but visible), so that claim should be read narrowly as \"no error from the boundary closure itself.\" And the claim about improving accuracy to arbitrarily high order is conditional on having an interior scheme of matching order; the paper only demonstrates fourth order. The self-citation to Arnold is appropriate and properly contextualized.\n\nWho is this for? Numerical analysts working on transparent boundary conditions for quantum transport. It is a solid, citable contribution in that niche, and the flaws are repairable. I would send it to peer review with a request for major revision: state and prove a corrected uniqueness theorem excluding E = V0 (or handle the degenerate case), write out the aDTBC uniqueness proof, and soften the no-error wording. Not a desk reject.","headline":"Useful fourth-order TBC construction, but the uniqueness theorems have a real gap at E=V0 that lands inside your own energy integral.","tokens_in":18985,"tokens_out":2762,"would_cite":false,"duration_ms":28728,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["34L40","65L10","65L20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a compact fourth-order interior scheme combined with its discrete fourth-order transparent boundary conditions gives a globally fourth-order discretization of the 1D Schrödinger problem, and that an analytic variant…","keywords":["1D Schrödinger–Poisson problem","transparent boundary condition","compact finite difference scheme","discrete transparent boundary condition","resonant tunneling diode","fourth-order convergence","quantum transport"],"falsifier":"Run the paper's zero-potential test with $\\hbar=m_*=1$, $E=0.5$, $L=10$, and a coarse grid $\\Delta x=2.5$, so that $t=6/\\Delta x^2=0.96<2E=1$, solving (24) with the D4TBCs for a unit-amplitude incoming wave. If the numerical wave reproduces $e^{ikx}$ and the homogeneous version of the discrete system has only the zero solution, then the condition $t>2(E-V)$ is not necessary; if not, that hypothesis is doing essential work.","tokens_in":17904,"feed_emoji":"⚛️","tokens_out":9072,"duration_ms":82869,"temperature":0.7,"pith_summary":"Simulating nanoscale semiconductor devices with the 1D Schrödinger–Poisson model requires artificial boundary conditions that let electron waves leave the computational box without reflection. The paper introduces two boundary-condition discretizations: D4TBCs, fourth-order discrete transparent boundary conditions that match the accuracy of the compact interior scheme and avoid spurious oscillations when the potential vanishes, and aDTBCs, analytic discretizations that introduce no boundary error at all. It proves that the interior scheme closed with D4TBCs is globally fourth order, proves uniqueness for both discrete Schrödinger problems under a grid-dependent condition, and demonstrates fourth-order convergence and realistic current–voltage curves for a resistor and two resonant tunneling diodes.","feed_headline":"Boundary treatment yields fourth-order Schrödinger accuracy","feed_subtitle":"Discrete and analytic boundary schemes match the fourth-order interior solver and reproduce diode current-voltage curves.","key_machinery":"The load-bearing objects are the two boundary-condition families. The D4TBCs are fourth-order discrete transparent boundary conditions obtained by inserting the ansatz $\\psi_j=\\alpha^j$ on the left and $\\psi_j=\\beta^j$ on the right into the compact fourth-order stencil, solving the resulting discrete dispersion relations for $\\alpha=e^{i\\tilde k_1\\Delta x}$ and $\\beta=e^{i\\tilde k_2\\Delta x}$, and using these roots in the one-sided relations (38)–(39). The aDTBCs are analytic discrete transparent boundary conditions obtained by expanding $\\psi(\\pm j\\Delta x;k)$ in infinite Taylor series around the boundary and using the exact dispersion relation $k_1=\\sqrt{2m_*(E-V_0)}/\\hbar$; this yields relations (55)–(56) and their multi-point generalizations (57)–(58). The interior engine is the compact fourth-order approximation (21) for $u_{xx}=f$, together with the fourth-order one-sided derivative formula (23) used to discretize the transparent boundary conditions (7)–(8).","core_discovery":"The paper's central claim is that transparent boundary conditions should be discretized together with the interior stencil, not as separate low-order add-ons. Theorem 1 states that the compact fourth-order interior scheme (24) equipped with the D4TBCs forms a globally fourth-order discretization of the 1D Schrödinger problem. The D4TBCs are constructed so that, when the potential is constant outside the device, the exterior discrete waves $\\alpha^{\\pm j}$ and $\\beta^{j}$ are unimodular, which prevents the spurious oscillations seen when the boundary conditions are discretized naively. The aDTBCs go further: because the wave function is analytic near the boundaries, the paper writes the exterior relations as exact Taylor series, so the boundary introduces no discretization error and the overall order becomes whatever order the interior scheme delivers, even arbitrarily high order.","pith_inferences":["Beyond the paper: if aDTBCs are truly error-free, then for smooth potentials the practical convergence rate of Schrödinger–Poisson codes should track the interior stencil order exactly; a clean test would be to compare measured convergence orders for sixth- and eighth-order interior stencils against the predicted rates.","Beyond the paper: the violation of the uniqueness condition at high energies in the resonant-tunneling diode runs suggests a concrete numerical experiment, namely checking whether spurious oscillations or non-uniqueness appear in a zero-potential test as $\\Delta x$ is increased until $t\\leq 2E$, which would map the boundary of the proved regime.","Beyond the paper: the aDTBC construction relies only on analyticity of the wave function at the boundary and the constant-exterior-potential dispersion relation, so the same Taylor-series idea should transfer to time-dependent Schrödinger equations or to higher-dimensional settings where the exterior solution is known in closed form, though the paper does not explore these cases."],"forward_implications":["A Schrödinger solver using the compact fourth-order interior scheme and D4TBCs converges as $O(\\Delta x^4)$ throughout the computational box, including the boundary region.","Because aDTBCs introduce no boundary error, the global accuracy of a Schrödinger–Poisson calculation is limited only by the interior finite-difference scheme.","With aDTBCs, the overall accuracy can be raised to arbitrarily high order by widening the interior stencil, whereas DTBCs would require solving algebraic equations of degree at least six, which cannot be done explicitly.","The two discrete Schrödinger–Poisson models have unique solutions under the stated condition, so the iterative device-simulation algorithm is well posed at the discrete level.","The fourth-order schemes reproduce benchmark resistor and resonant-tunneling-diode current–voltage curves, including negative differential resistance, on meshes as coarse as 0.5 nm."],"supporting_citations":[{"why":"Introduces discrete transparent boundary conditions and the spurious-oscillation phenomenon that D4TBCs are designed to cure.","marker":"[1]"},{"why":"Establishes the transparent boundary conditions and the 1D Schrödinger–Poisson scattering model that the paper discretizes.","marker":"[2]"},{"why":"Supplies the compact finite-difference and Taylor-matching methodology used to build the fourth-order interior scheme.","marker":"[11]"},{"why":"Provides the central compact-scheme framework that motivates the one-sided derivative approximation used at the boundaries.","marker":"[12]"}],"fun_headline_variants":["Fourth-order transparent boundaries for Schrödinger-Poisson","Boundary discretization matches interior order exactly","No discretization error at transparent boundaries","Boundary conditions to arbitrary order for 1D Schrödinger-Poisson"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The uniqueness and non-oscillation theorems require a grid-dependent quantity $t$, proportional to the inverse square of the grid spacing, to exceed $2(E-V)$ at both contacts for every energy $E$; the resonant-tunneling diode simulations use energies up to $0.8$ eV with $\\Delta x=0.5$ nm, where this condition is violated for part of the range, so the proved regime is a subset of the tested regime.","fun_headline_variants_meta":{"raw":{"variants":["Fourth-order transparent boundaries for Schrödinger-Poisson","Boundary discretization matches interior order exactly","No discretization error at transparent boundaries","Boundary conditions to arbitrary order for 1D Schrödinger-Poisson"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000658,"raw_usage":{"total_tokens":3034,"prompt_tokens":989,"completion_tokens":2045,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":605,"completion_tokens_details":{"reasoning_tokens":1982}},"tokens_in":605,"tokens_out":2045,"duration_ms":15312,"temperature":1.0,"reasoning_tokens":1982,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:45:16.959622+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the paper's zero-potential test with $\\hbar=m_*=1$, $E=0.5$, $L=10$, and a coarse grid $\\Delta x=2.5$, so that $t=6/\\Delta x^2=0.96<2E=1$, solving (24) with the D4TBCs for a unit-amplitude incoming wave. If the numerical wave reproduces $e^{ikx}$ and the homogeneous version of the discrete system has only the zero solution, then the condition $t>2(E-V)$ is not necessary; if not, that hypothesis is doing essential work.","supporting_citations":[{"cited_title":"Mathematical concepts of open quantum boun dary conditions","cited_arxiv_id":null,"evidence_quote":"Introduces discrete transparent boundary conditions and the spurious-oscillation phenomenon that D4TBCs are designed to cure."},{"cited_title":"On a one-dimensional Schr¨ odinger-Poisson scattering model","cited_arxiv_id":null,"evidence_quote":"Establishes the transparent boundary conditions and the 1D Schrödinger–Poisson scattering model that the paper discretizes."},{"cited_title":"A new class of central compact schemes with spectral-like resolution I: Linear schemes","cited_arxiv_id":null,"evidence_quote":"Provides the central compact-scheme framework that motivates the one-sided derivative approximation used at the boundaries."}],"review_version":1}