Search Results (348)

In physical model inversion of leaf area index (LAI) over complex vegetation canopies, variations in canopy structure, background moisture, and background brightness can cause spectrally similar lookup table (LUT) candidates to represent different combinations of canopy and background parameters, reducing retrieval stability. This study proposes a PROSAIL lookup table inversion method for GaoFen-5A (GF-5A) hyperspectral imagery based on physically informed parameter range refinement. A small set of key GF-5A feature bands was selected to preserve information sensitive to canopy structure and background conditions while reducing spectral redundancy. Vegetation type information was used to determine the LIDFa range, while the normalized difference infrared index (NDII) and brightness index (BI) were used to guide the refinement of psoil and rsoil ranges according to moisture related and brightness related spectral responses, respectively. Combining compact spectral inputs with refined PROSAIL parameter ranges improved LUT candidate selection under variable canopy and background conditions. Validation using 62 field LAI values aggregated at the GF-5A pixel scale yielded a coefficient of determination (R²) of 0.6000, a root mean square error (RMSE) of 1.1461, a mean absolute error (MAE) of 0.8712, and a Bias of −0.0092, outperforming the comparison schemes. These results indicate improved physical consistency and stability in LAI inversion under complex canopy and background conditions. Full article

Retailers of non-instantaneously deteriorating items must jointly set inventory, preservation technology, and payment decisions. Preservation technology reduces deterioration, but excessive investment increases operational costs, making the determination of an optimal preservation level essential for maximizing profit. Although preservation technology, hybrid payment schemes, and prepayment discounts have been studied individually, their joint treatment alongside partially backlogged shortages remains largely unexplored. To address this gap, this study develops an inventory model that simultaneously incorporates preservation technology investment, a hybrid payment structure, advance payment combined with trade credit, optionally supplemented by a prepayment discount, and partially backlogged shortages for non-instantaneously deteriorating items. A classical optimization approach is employed, yielding quasi-closed-form solutions for the shortage and replenishment timing across four trade credit scenarios, while the profit-maximizing preservation investment level is identified through sensitivity analysis. Numerical examples and sensitivity analysis show that increasing the number of prepayment installments lowers the discount rate offered by the supplier; because this forgone discount outweighs the benefit of retaining capital longer, the retailer’s profit falls. Profit responds most strongly to purchasing cost, the advance payment period, and lead time. These results give retailers a practical basis for balancing preservation investment, payment structure, and shortage policy to maximize profitability. In the sensitivity analysis, profit varies by more than 45% over the tested range of the purchasing cost and by up to 21% depending on the number of prepayment installments negotiated with the supplier. That gives retailers a concrete ranked basis for prioritizing which contract terms to negotiate first. Full article

Low Earth orbit (LEO) satellite downlinks broadcast over wide ground footprints, exposing Earth-observation and remote-sensing sensor data to passive eavesdropping. Affine frequency division multiplexing (AFDM) is a candidate waveform for the doubly dispersive LEO channel and a natural integrated sensing and communication (ISAC) waveform whose delay–Doppler structure supports target parameter estimation; yet existing secure-AFDM schemes act only in the discrete affine Fourier transform (DAFT) parameter domain, leaving the transmitted waveform structurally recognizable. To address this gap, this paper applies time-domain waveform obfuscation to AFDM as physical layer encryption. Using a secret key, the transmitter permutes the inverse-DAFT samples and applies a phase rotation before chirp-periodic-prefix generation; the mask is unitary, so the peak-to-average power ratio is preserved exactly, and the key-holding receiver retains AFDM’s full delay–Doppler sensing capability, while a no-key receiver obtains a dense composite response that destroys target localization (sensing concentration drops from 0 dB to −16.6 dB). Secret pilot phases enable channel estimation at the legitimate receiver while blocking a naive composite-channel attack. Simulations at $N=64$ and 128 show that a wrong-key eavesdropper achieves uncoded BER within $0.01$ of $0.5$ across 0–20 dB and that blind Viterbi–Viterbi phase recovery is no more effective under QPSK (BER $0.46$–$0.48$), while the legitimate SNR penalty stays below $0.5$ dB. The mask also suppresses AFDM’s internal structure to the AWGN level under AFDM-aware processing. Time-domain obfuscation offers a complementary physical-layer security layer for confidential LEO remote-sensing data downlink and ISAC waveforms. Full article

The foundation of modern vehicle control relies on Electronic Control Units (ECUs) communicating via the Controller Area Network (CAN). However, CAN was not designed with security in mind. Limited bandwidth and lack of security make CAN vulnerable, while centralized solutions like AUTOSAR SecOC suffer from high latency. To solve this problem, we propose a novel security protocol named Self-Updating Hash Chain Layered Security Protocol (SUHC-LSP), which uses a “space-time coupled” frame structure to fit a robust authentication code into the limited CAN data field. Unlike conventional schemes that require explicit freshness negotiation during routine operation, SUHC-LSP adopts a self-updating hash-chain mechanism in which chain evolution proceeds autonomously under normal conditions. In addition, SUHC-LSP introduces a self-updating hash chain mechanism that enables freshness iteration during steady-state operation. Traditional AUTOSAR SecOC configuration schemes require additional synchronization messages and freshness counter management, whereas the approach proposed in this paper eliminates the resulting bandwidth overhead while adhering to the 8-byte CAN payload limit and ensuring message authenticity and integrity. In addition, a risk-adaptive two-layer architecture is designed to balance fast speed for local messages and strong encryption for cross-domain messages. BAN-logic and ProVerif verification show that the protocol preserves authentication, freshness, secrecy, and event-correspondence properties under the stated assumptions. Experiments on an STM32 platform show that, in a 4-device prototype, the computational overheads are 1.61 ms for intra-domain communication and 0.83 ms for inter-domain communication. Moreover, analytical overhead comparison indicates that the proposed protocol has lower sensitivity to network scale than the compared schemes. Full article

Ensuring continuous load supply under N-1 contingencies poses a significant challenge for active distribution networks. Traditional fault recovery relies on mechanical tie switches for network reconfiguration; however, such rigid interconnections often fail due to severe terminal voltage drops during long-distance load transfers, leading to forced load shedding. To address this, this paper proposes an optimal planning framework for Electric Energy Routers (EERs) to secure load supply, aiming to maximize load preservation within investment budget constraints. First, a bi-level robust optimization model is constructed: the upper level determines the optimal EER configuration to minimize load shedding in the worst-case N-1 scenarios, while the lower level evaluates the optimal restoration strategies via mixed-integer second-order cone programming. To handle the inherent complexity of discrete–continuous variable coupling and the computational burden of the bi-level structure, a novel topological manifold evolution algorithm based on Wasserstein geometric flow is developed. By embedding the electrical topology into a sensitivity-driven Riemannian metric field, the algorithm effectively enhances global optimization capabilities. Case studies on the IEEE 33-node system demonstrate that the integration of EERs facilitates a transition from rigid to flexible interconnection, circumventing power transfer bottlenecks caused by low voltage through controllable power flow. Compared with baseline schemes, the optimal EER scheme reduces load shedding in the worst-case scenario by 60.3%, providing a robust and economical solution for high-resilience distribution network planning. Full article

Hyperspectral infrared sounders provide valuable observations for numerical weather prediction (NWP), but their native nadir spatial resolution of approximately 12–16 km is coarser than the approximately 4 km grid spacing commonly used in convection-permitting regional forecasting systems. To enhance the spatial resolution of these observations toward this scale, we propose the Channel-Decoupling and Geographic-Prior Fusion Network (CDGP-Net), an unsupervised hyperspectral–multispectral fusion framework that reconstructs 4 km high-spatial-resolution hyperspectral radiances by fusing the FengYun-3D (FY-3D) Hyperspectral Infrared Atmospheric Sounder (HIRAS) data with co-platform Medium Resolution Spectral Imager II (MERSI-II) imagery while preserving the original spectral sampling. To adapt hyperspectral–multispectral fusion to infrared sounder data, CDGP-Net incorporates two components: a self-reconstruction and spectral-degradation channel-decoupling (SDCD) design, which allows physically related non-overlapping MERSI-II infrared information to be used as an auxiliary input while keeping the spectral degradation physically consistent; and a reconstruction-domain geographic-prior regularization (RGPR) scheme, which constrains the reconstructed radiances in both geographic space and spectral shape. Because true high-resolution observations are unavailable, we further introduce a radiative-transfer-anchored evaluation (RTAE) scheme that uses the line-by-line radiative transfer model (LBLRTM) simulations driven by reanalysis and forecast atmospheric fields as independent physical references. For the selected FY-3D overpass cases, the evaluation using Ref-HR as the high-resolution physical reference shows that CDGP-Net improves the peak signal-to-noise ratio (PSNR) by 3.8 dB and reduces the spectral angle mapper (SAM) and erreur relative globale adimensionnelle de synthèse (ERGAS) by 64.1% and 54.6%, respectively, compared with the unmixing baseline. Under the same evaluation conditions, relative to geographic interpolation, it improves the structural similarity index measure (SSIM) by 16.4% and reduces ERGAS by 8.8%, with the clearest advantages in partial-cloud and coastal transition scenes. Full article

Standard pushers for the relativistic equations of motion of a charged particle in an electromagnetic field—Boris, Vay, Higuera–Cary—do not, in general, preserve the full symplectic structure of the underlying Hamiltonian system, while high-order non-symplectic schemes such as Runge–Kutta accumulate secular error over long times. We propose a two-stage, symmetry-preserving framework (SP-PINN) for the 3+1-dimensional relativistic dynamics of a charged particle in a prescribed field, including a focused Gaussian laser pulse, that pairs a physics-informed neural network with an explicit symplectic integrator: the network learns a surrogate relativistic Hamiltonian, while the integrator—which is not itself learned—advances it. In Stage 1, an unsupervised physics-informed neural network learns the surrogate from the covariant equations of motion using a Lorentz-invariant loss that enforces the mass-shell constraint $\mathcal{H}=m{c}^2\gamma$; in Stage 2, the surrogate is advanced with an explicit symplectic map built on Tao’s extended phase space, valid for the non-separable relativistic Hamiltonian. To isolate the geometric integrator from neural-network approximation error, every benchmark figure advances the analytic relativistic Hamiltonian through Stage 2, the learned Stage-1 surrogate being assessed separately. We benchmark against the Boris pusher and Runge–Kutta on three core test problems (adding the Higuera–Cary pusher in the symplecticity diagnostic), supplemented by plane-wave, ensemble, and pulse-family studies, and we measure the first Poincaré–Cartan loop invariant directly as a quantitative diagnostic of symplecticity. The magnetic-field test illustrates the contrast between bounded and secular error growth: Runge–Kutta drifts secularly, the Boris pusher conserves the invariants to machine precision as a volume-preserving gyro-integrator, and the symplectic map keeps the error bounded for all time; on a non-integrable magnetic trap, where no exact volume-preserving rotation exists, the symplectic map alone keeps the energy error bounded. The learned surrogate is the current accuracy bottleneck—not yet competitive with the conventional pushers for the static cases—but for the demanding laser case, a vector-potential light-cone reformulation reduces this surrogate error to $(3.0\pm 0.1)\times {10}^{-4}$ (three seeds) and yields learned trajectories that remain phase-coherent over essentially the whole interaction. The framework targets laser–plasma acceleration, synchrotron-radiation modeling, and particle tracking. Full article

High-precision 3D reconstruction of on-orbit non-cooperative targets is essential for space situational awareness. However, extreme space environments induce severe imaging degradations, including high-dynamic-range (HDR) illumination, rapid motion blur, and platform jitter. Traditional 3D Gaussian Splatting (3DGS) conflates these optical distortions with geometric optimization, leading to pathological structural inflation and the loss of thin appendages like solar panels. To overcome this, we propose OrbitGS, a physically decoupled 3DGS framework. OrbitGS integrates physical imaging priors via a Kinematics-Driven Degradation Synthesizer (KDDS) to deterministically extract view-specific degradation kernels. Furthermore, a blur-decoupled rendering strategy with intensity-aware weighting mitigates HDR variations, while a semantic-aware densification scheme mathematically penalizes abnormal primitive expansion. Evaluations on the SPE3R dataset demonstrate that OrbitGS effectively disentangles optical degradations from the geometric representation. Quantitatively, evaluated across seven space targets under moderate (200-view) and extreme sparse (50-view) settings, our framework achieves state-of-the-art robustness against extreme degradations. Notably, it avoids the catastrophic structural blow-ups observed in baseline methods, yielding an average geometric F1-score of 0.80 and a Chamfer Distance of 0.818, alongside a rendering Structural Similarity Index (SSIM) of 0.82 and a Learned Perceptual Image Patch Similarity (LPIPS) of 0.16. By preserving delicate structures under severe degradation, OrbitGS provides a robust, high-fidelity 3D reconstruction solution for complex orbital environments. Full article

In applications pertaining to sensor network, the Internet of Things, industrial monitoring, and cyber–physical security, graphs are increasingly being outsourced to clouds, where similarity search should be supported without exposing graph content, query graphs, update contents, or database evolution. Existing privacy-preserving graph similarity schemes mainly target static encrypted databases, and as such struggle to handle insertions, deletions, label updates, and long-running index maintenance. This paper proposes DFB-PPGSQ, a dynamic forward-private and epochal backward-private graph similarity matching scheme that moves branch-based lower-bound filtering into a structured encryption framework. DFB-PPGSQ uses epoch-local feature tokens, per-record occurrence handles, one-time update labels, update buffers, deletion tombstones, and shuffle-based branch–tree re-randomization to preserve pruning efficiency while making same-epoch tombstone, traversal, size, timing, and refresh leakage explicit. We formalize the system model, leakage functions, algorithms, and security interpretation, then implement a reproducible Python prototype with HMAC-SHA256 token generation and multi-profile dynamic sensor-topology workloads. Across five random seeds, DFB-PPGSQ keeps server-side filtering latency close to the static branch–tree baseline (46.60 ms versus 44.72 ms at 4000 graphs), avoids immediate full-rebuild updates (0.091 ms insertion and 0.092 ms label update), and keeps metadata-assisted cross-epoch token linkage below 5.6% attack success after refresh in additional industrial and campus IoT stress workloads. Storage, communication, exact GED refinement, side-channel hardening, and verifiable-result protection are treated as deployment costs and limitations rather than being included in the headline latency. Full article

Background/Objectives: Antibiotic-resistant Pseudomonas aeruginosa and limited experimental validation capacity motivate efficient prioritization of antibacterial candidates from large chemical libraries. Quantitative structure–activity relationship (QSAR) benchmarks often binarize disk diffusion inhibition-zone (IZ) measurements, obscuring activity gradients and imposing threshold dependence. We examined whether continuous-IZ modeling provides complementary retrospective prioritization relative to calibrated binary classification in a highly imbalanced dataset. Methods: In this retrospective matched-data evaluation, we revisited a published ChEMBL-derived P. aeruginosa disk diffusion dataset using preserved training and locked external validation partitions. A calibrated support vector classifier using Molecular ACCess System keys (SVC/MACCS) provided conservative binary active calls. RegressionStack combined source-descriptor extreme gradient boosting (XGBoost) and ElasticNet regressors, Morgan-fingerprint random forest and gradient-boosting regressors, and a MACCS-key XGBoost regressor through an XGBoost meta-regressor to predict continuous IZ. Results: On the locked external set (n = 1130; 87 actives), SVC/MACCS achieved a positive predictive value (PPV) = 0.619, receiver operating characteristic area under the curve (ROC-AUC) = 0.857, precision–recall area under the curve (PR-AUC) = 0.479, and enrichment factor at 1% (EF@1%) = 7.58. RegressionStack achieved a mean absolute error (MAE) = 3.20 mm, ROC-AUC = 0.896, PR-AUC = 0.545, and EF@1% = 9.74. Neither paired permutation tests (ROC-AUC, p = 0.501; PR-AUC, p = 0.442) nor paired bootstrap confidence intervals resolved these differences. The y-randomization analyses supported non-random signals; scaffold-grouped validation retained early enrichment but showed reduced broader performance. The consensus-positive tier contained 27 actives among 35 nominations (PPV = 0.771). At measured IZ ≥ 30 mm, MAE increased to 9.46 mm, and all 30 compounds were underpredicted. Conclusions: Continuous-target modeling retained the IZ scale during training and generated a threshold-flexible predicted-IZ prioritization coordinate complementary to, but not statistically superior to, calibrated binary classification. The methodological contribution is a matched-data evaluation of complete workflows and their nomination behavior under the same partitions, yielding a retrospective compound-tiering scheme. Because the pipelines differed in architecture and molecular representation, their differences cannot be attributed solely to endpoint formulation. The workflows were not prospectively evaluated on compounds lacking pre-existing IZ measurements, and whether retrospective enrichment improves experimental hit discovery or reduces screening workload remains to be established. Full article

The DR Code (Data Restorable Code), originally proposed by Sriphum in 2013, is a two-dimensional barcode that achieves a 33% data-recovery rate against six distinct cases of strip-shaped data loss using simple XOR-based parity. The original work presented a single 3 × 3 arrangement of nine logical blocks (A0, A1, A2, B0, B1, B2, C0, C1, C2) in which each row and each column contain exactly one element from each of the three data classes (A, B, C). This paper systematically enumerates every 3 × 3 arrangement that satisfies this recoverability property and proves, by exhaustive search released as an open-source program (DR15.py), that exactly 2592 such arrangements exist. We then introduce five structural theorems—mirror reflection, vertical flipping, Tetris-style rotation, cyclic column rotation, and cyclic row rotation—and prove that each preserves recoverability. We show that these five generators, viewed as a group action, produce a finite group of order 72 isomorphic to the semi-direct product (C₃ × C₃) ⋊ D₄, which partitions the 2592 patterns into exactly 36 absolute equivalence classes. We further explore the partial-quotient structure under D₄ alone (yielding 324 classes, the case the practitioner is most likely to encounter) and under cyclic-only quotient (yielding 288 classes). As practical contributions, we propose (i) a compact equivalence-class encoding that reduces the storage cost of one DR Code template from a naïve 36 bits to 13 bits, and (ii) a canonical-form deduplication scheme suitable for cloud and embedded storage systems. Additionally, we propose two further contributions: a fast pattern-validity oracle based on canonical lookup, and a randomization-friendly DR Code variant for security-aware barcode applications. Empirical results confirm all theorems on the full enumeration of 2592 patterns. Full article

Machine learning techniques for network-based intrusion detection systems (NIDS) have advanced considerably over the past decade. Still, improvements are inhibited by handcrafted feature pipelines, isolated public benchmark data, and evaluation procedures that do not reflect real-life deployment. AutoML, a branch of ML automating model selection, automated architecture search, and the creation of model pipelines, may help overcome these shortcomings. While numerous NIDS applications employing automated ML techniques have been proposed, and recent surveys have mapped the AutoML framework landscape for network intrusion detection, no existing review critically audits the evaluation practice of this literature: the quality of its benchmark datasets, the reproducibility of its reported results, and the realism of its deployment assumptions. This paper critically reviews 26 research works published between January 2023 and June 2026, collected via a two-phase structured search: a documented keyword search across five databases (Scopus, IEEE Xplore, Web of Science, ACM Digital Library, and Google Scholar), followed by full-text eligibility screening, citation chaining, and expert evaluation. Findings drawn from this collection capture trends observed among the selected studies, rather than reflecting the broader state of the field. Analysis of the corpus reveals that 88% of dataset-verified studies evaluate exclusively or partly on the legacy benchmark family (KDD-derived, CICIDS, UNSW-NB15, CIDDS), 21% evaluate on a single dataset only, and among attribute-verified studies only 32% release source code, 40% report statistical significance testing, and 36% include variance analysis, findings that collectively motivate the four contributions of this study. First, a recommended evaluation framework is proposed, addressing baseline parity, transparent search-space and budget reporting, nested cross-validation for selection-bias control, and stability reporting across multiple random seeds. Second, a dataset quality scoring framework is introduced, assessing five dimensions: overlap rate, duplication rate, label correctness, attack-type representativeness, and coverage of benign, IoT, and IIoT traffic. Third, a cross-domain justification is provided for neural architecture search (NAS) and meta-learning in NIDS, grounded in advances in federated NAS, out-of-distribution robustness, edge-constrained search cost reduction, and few-shot adaptation. Fourth, a structured research roadmap is outlined, targeting real-world validation, standardized benchmarks, curated datasets, resource-aware AutoML, and privacy-preserving federated NAS. In contrast to prior surveys of AutoML for network intrusion detection, which map frameworks and computational paradigms, this review contributes a formalized evaluation checklist, an explicit and partially empirically validated dataset quality scoring scheme, and evidence-based methodological guidance grounded in a transparent, fully enumerated study corpus. Full article

The article presents the results of research on the justification of the working unit parameters and the structural scheme of the chisel for loosening stubble fields. It was established that, according to the criterion of minimum draft resistance, the optimal installation angles are 20–24 degrees to the bottom of the furrow at a working share width of 40–50 mm. The calculated values of the share installation angles were verified on a laboratory setup, and the results confirmed the theoretical calculations. Based on the quality of loosening of the worked layer and the preservation of stubble, the shank width should be 20–30 mm, the share width 40–50 mm, and the share installation angle 20–24 degrees. The shank should be equipped with a safety mechanism to prevent breaking or bending. The results of research on different leveling device options are presented, and the structural–technological scheme, main parameters, and operating modes of the tool for chiseling compacted stubble fields are substantiated. Dependencies of the treatment quality on the parameters of the loosening working units, the leveling device, and the operating speed of the implement were obtained. Based on the justified parameters, an experimental prototype of the soil chiseling machine was manufactured and field tests were conducted under production conditions on stubble fields. Full article

This study proposes an intelligent speed-loop controller for a brushless DC motor (BLDCM) drive implemented under a field-oriented control (FOC) scheme. The controller is constructed by embedding the coati optimization algorithm (COA) into a composite sliding mode theory (CSMT) control structure. In sliding mode control (SMC), the use of only one reaching law normally produces a design trade off: increasing the reaching speed tends to aggravate overshoot or chattering, whereas reducing switching activity often slows convergence. To mitigate this compromise, the proposed controller adopts a composite reaching law (CRL) formed by an exponential component and a power component. When the system trajectory is distant from the sliding surface, the exponential component strengthens the reaching action and shortens the transient interval. When the trajectory moves close to the sliding surface, the power component decreases the effective switching intensity, thereby attenuating high-frequency chattering and reducing the overshoot associated with an aggressive exponential action. For adaptive gain selection, the COA search variables are chosen as four controller parameters: the sliding mode gain, the exponential reaching gain, the power reaching gain, and the power exponent. The fitness index is established from the rotor-speed tracking error and the time variation in that error. By imitating the hunting and predator-avoidance behaviors of coatis, the optimization process updates candidate solutions and selects the parameter combination that best matches the current operating condition. The resulting gains are supplied to the composite sliding mode controller (CSMC) so that the BLDCM can follow speed commands rapidly while preserving stable regulation. Because the proposed method performs online optimization of controller gains rather than data-driven training, it can be realized without a large training dataset. MATLAB/Simulink simulations are carried out to examine the effectiveness of the proposed strategy. The controller is compared with four benchmark methods, namely power reaching law (PRL)-based SMC, exponential reaching law (ERL)-based SMC, non-optimized composite reaching law SMC, and zebra optimization algorithm (ZOA)-assisted ERL-based SMC. The simulation results demonstrate that the proposed COA-based composite sliding mode controller improves both speed command-tracking and load-disturbance rejection relative to the comparative controllers. Full article

Wave propagation modeling in heterogeneous media requires numerical methods that can simultaneously handle complex geometries, artificial boundary reflections, and spatially varying resolution demands. In this study, we present a Discontinuous Galerkin (DG) forward modeling method for wave propagation with split-field absorbing boundary conditions and gradient-based adaptive meshes. The wave equation is formulated as a first-order hyperbolic system and discretized by the DG method, which preserves local conservation and is well suited for explicit Runge–Kutta time integration on unstructured meshes. To reduce spurious reflections from truncated computational boundaries, a split-field absorbing boundary treatment is introduced in the absorbing layer through directional damping terms, maintaining the first-order structure and local update form of the DG scheme. In addition, a physics-based mesh metric is constructed from the local velocity-gradient length scale, allowing the mesh to be automatically coarsened in smooth regions and refined near strong velocity contrasts, interfaces, and discontinuities. Numerical convergence tests show that the quadratic DG scheme achieves the expected third-order accuracy in the L2 norm. Quantitative PML evaluation gives a reflection coefficient of approximately $1.65\times {10}^{-5}$, indicating effective suppression of artificial boundary reflections. For the three-dimensional Marmousi model, the proposed adaptive mesh reduces the number of tetrahedral elements from 158,492 to 88,681, decreases the CPU time from 241.4453 s to 139.1328 s, and reduces memory consumption from 3788.67 MB to 2109.55 MB compared with the uniform mesh. These results demonstrate that the proposed method can improve the balance between computational efficiency and solution accuracy while maintaining stable and physically interpretable wavefield modeling in heterogeneous media. Full article