Computational Physics Papers

ALT Forward-flux-based weight windows give global variance reduction for duct-streaming shielding problems, but each new geometry needs an iterative Monte Carlo calculation to build them. We report preliminary results on a convolutional neural network (CNN) that learns the map from a two-dimensional duct geometry and source position to the lower window bound of such an iteration, here the MAGIC generator in OpenMC, in milliseconds. A set of 250 random concrete ducts of three curved and rectangular families on the Protvino labyrinth bounding box was run in OpenMC with a Cf-252 point source and MAGIC generation, and a 1.9-million-parameter U-Net was trained on the logarithm of the lower bound. On held-out ducts the error in the cells that carry the flux is 0.14 to 0.51 decades; replicate MAGIC runs show this part of the map to be reproducible to a few hundredths of a decade, so the residual is model error, while below 10⁻⁶ the reference itself is noise. Applied without retraining to the rect

ALT The integrated crystal orbital Hamilton population (ICOHP) is the standard measure of bond strength in a DFT calculation, but every structure needs its own DFT single point and LOBSTER projection, which takes days for a cell of a hundred atoms, such as a grain boundary or a catalytic nanoparticle. We predict ICOHP from the frozen descriptors of a MACE foundation model. A symmetric pair head reads the scalar and bond-projected vector channels of the two atoms and corrects a pair baseline built from the element embeddings of the same model. We tested the model on the high-entropy diboride (Hf,Zr,Ti,Nb,Ta)B₂, Pt clusters on defective graphene and Pt/Pd nanoparticles with adsorbed O₂. The error on bonds stronger than 0.5 eV is 1.6-4.1%, about half that of the Gaussian-process model of Ferreira et al. trained on the same data. Trained only on high-temperature HEB₂ snapshots, the model reproduces the LOBSTER ICOHP of relaxed grain boundary cells to 0.06 eV in about a second, against 5-51 hou

ALT Differential and integral control rod worth are usually obtained by differencing eigenvalue calculations at discrete rod positions, which resolves the derivative poorly and gives it no uncertainty. We build a Gaussian-process (GP) surrogate for reactivity as a function of rod position on VERA Core Physics Benchmark Problem 4, a 3×3 assembly cluster with a hybrid AIC/B₄C rod cluster control assembly, modelled in OpenMC with ENDF/B-VIII.0. A parametric sigmoidal mean carries the dynamic range, a Matern-5/2 GP carries the residual, and the reported Monte Carlo uncertainty of each run enters as known observation noise. Because differentiation is linear, the differential rod worth and its variance follow in closed form from the kernel derivatives. Trained on 43 runs at three particle counts and tested on 11 independent runs, the surrogate meets a fixed 10 pcm accuracy target above the test-run noise, and its intervals are calibrated: the standardized leave-one-out residuals over 129 indepen

ALT PhysicsNeMo-Mesh is the mesh module of the open-source PhysicsNeMo library. It provides mesh processing similar to PyVista, but on the GPU, in batches, and inside PyTorch's automatic differentiation. Like xarray for unstructured meshes, it attaches semantics to data: fields on points, cells, and whole meshes have names and shapes, and a simulation case keeps the complete PDE problem specification, including boundary conditions and their prescribed values. It addresses three problems that plague machine learning for physical simulation. First, GPU utilization is bottlenecked by I/O, because samples are very large relative to the compute per forward pass. Second, because traditional preprocessing is expensive, it is necessarily done once, during curation, rather than differently for each model and workflow at the point of use. Finally, advanced equivariant and mesh-native architectures require information that curation ordinarily discards. Fast GPU mesh processing makes it practical to p

ALT Geometric Volume-of-Fluid (VOF) on unstructured polyhedral meshes reconstructs the interface orientation from the volume fractions of a stencil of neighbours; on irregular cells and in the presence of wisps it degrades, and the curvature built on it drives the parasitic currents of capillary flows. We transport, consistently to the volume fraction of every cell, the wetted fraction of every face area. From the area and volume fractions, closed-form equations recover the interface normal and position of a cell, for the first time in VOF: the normal from one sum over the cell's own faces, the divergence theorem on the liquid sub-volume, and the plane's position from a second, without neighbours or iteration. Both identities are exact for a plane on hexahedral, perturbed and polyhedral cells. On a sphere and an ellipsoid the normal has the order of the face fractions, the position is second order for every state and mesh, and a quadric through one ring of recovered normals gives a converg

ALT The hybrid finite element method stores a dense inverse of the nodal even-parity operator for every node and energy group, and that storage limits the spatial and angular expansion orders, more tightly still on a GPU, whose memory is smaller than that of a host. This work introduces fast diagonalization for the nodal even-parity operator of the hybrid finite element SPN method: the stored dense inverse is replaced by the eigendecompositions of a planar, an axial and an angular factor, and the inverse is applied without being formed explicitly. Every even-parity moment of the SPN equations obeys an equation of the same diffusion form, with no mixed derivatives, so the angular coupling is the same in every spatial direction. On a node extruded along the axis, the operator is then an exact Kronecker sum of the three factors. The factorization merely recasts the computation, so the discretization and the solution are unchanged, and the nodes are not required to be identical to one another.

ALT QMOMKit is an open-source C++ library for the phase-space evolution of population balance equations with quadrature-based moment methods. The present version implements the quadrature method of moments (QMOM) and the conditional quadrature method of moments (CQMOM). QMOMKit is designed to be independent of any particular application and can be used either as a standalone solver for spatially homogeneous population models or as a local phase-space solver coupled to a host application. It handles the numerical aspects of the evolution of a population in phase space, including quadrature reconstruction, advection of the quadrature abscissas, and changes to the population through moment sources and phase-space boundary contributions. The number of quadrature nodes is adapted to what the supplied moments support, subject to prescribed bounds of the phase-space coordinates, and the CQMOM implementation supports an arbitrary number of phase-space dimensions as well as any conditioning order.

ALT Many computational models exhibit multiple operating regimes separated by abrupt or non-smooth transitions as the input parameters vary. Such regime transitions complicate the construction of global parametric reduced-order models, often necessitating large training datasets to capture the distinct system behaviors. Accurate regime boundary detection instead enables localized modeling strategies by partitioning the parameter space into regions of qualitatively distinct behavior. In many cases, regime switching corresponds to the excursion set of a simulator output. When such a quantity is discontinuous or non-smooth, regression-based algorithms for level set estimation can struggle to correctly approximate the boundary. In this work, we show that solving a resulting binary classification problem can improve sample efficiency and we further propose a novel adaptive sampling strategy for support vector machines. Unlike uncertainty-based sampling schemes, the proposed approach uses a cont

ALT This paper focuses on the numerical and analytical determination of the mechanical parameters of aluminum-matrix nanocomposites using atomistic-scale simulations within a multiscale modeling framework. Molecular statics simulations were employed to evaluate the tensile and compressive behavior of Al based composites reinforced with nanosized alumina particles. Two classes of atomistic models were investigated. First, a unit-cell model containing a single Al2O3 monocrystal embedded in an aluminum matrix was used to examine the influence of the volume fraction of the ceramic inclusion on the effective mechanical properties. Second, a polycrystalline RVE model comprising an alumina particle was analyzed to assess the effect of the number of Al grains on effective properties. For each case, the mechanical response was characterized by stress-strain curves, enabling the identification of upper-scale material parameters associated with the elastic regime (bulk, shear and Young's modulus) and

ALT Monte Carlo PDE solvers enable mesh-free pointwise queries, but repeatedly traversing heterogeneous media is costly, particularly in scenes with reusable or dynamically changing materials. We introduce Walk on Multiscale Media (WoMM), which compiles fine-scale transport into reusable, editable and composable transport prefabs. A probabilistic interpretation of multiscale finite-element bases and condensed operators yields coarse transitions retaining fine-scale material effects. Each prefab combines port-level operators, entrance-dependent exit distributions, and Poisson-marked trajectory records, enabling incremental updates for overlapping absorption fields, gradient-field steering through path reweighting and transient diffusion. Integrated with conventional Walk-on-Spheres in a hybrid continuous–discrete solver, these prefabs accelerate repeated pointwise PDE queries and support interactive material edits without repeatedly solving the fine-scale problem.

ALT Finite-volume and discontinuous Galerkin methods update cells using the space-time flux integral over their faces, which typically depends on the solutions to local Riemann problems. In the multidimensional Riemann problem setting, this integral is difficult to evaluate exactly because of complex wave interactions. In this work, we show that the exact space-time flux integral is a polynomial in the time step and face dimensions. This motivates a learning-based approach, as only the mapping from the initial states to the polynomial coefficients needs to be learned. We propose a structure-preserving, data-informed Riemann solver framework that uses a neural network trained on high-resolution simulation data to learn this mapping. Key structural properties, including consistency and symmetry, are enforced by construction. Provided that the high-resolution simulations converge to the exact flux integrals and the network accurately predicts the coefficients, this framework yields near-exact

ALT Algorithm design tailored to graphics processing units (GPUs) can substantially improve the efficiency of computational fluid dynamics (CFD) simulations. Efforts to adapt CFD solvers to GPUs often focus on spatial discretization, while time advancement commonly relies on multistage Runge–Kutta (RK) methods. However, these RK methods require repeated residual evaluations and full-domain operations, increasing the cost of compressible-flow direct numerical simulation (DNS). To reduce this overhead, we develop a local arbitrary high-order derivatives (ADER) finite-difference method that couples a compact ten-point quadratic predictor to a conservative central corrector through operator matching. With two quadrature nodes, one Picard correction, and selective dissipation, smooth convective tests recover third-order convergence under fixed-CFL refinement and near-seventh-order spatial behavior when background spatial error dominates. Temporal states remain kernel-local, and halo communicati

ALT The stress tensor resolved in wavevector qvec and time t, σαβ(qvec,t), gives access to the elastic moduli of amorphous solids as a function of the probed length scale, and to the long-range stress correlations that distinguish an amorphous solid from a liquid. Its evaluation requires the contribution of every interacting pair and therefore an interface to the force loop of the molecular dynamics code. In the current work, we present: i) an optimized implementation of the LAMMPS module for calculating σαβ(qvec,t), in which trigonometric functions are calculated only for the basic wavevectors and, in addition, all the relevant wavevectors are supplied as an explicit list rather than generated as a full grid, so that only the wavevectors of interest are computed and stored, and ii) a validation protocol for the code illustrated for a two-dimensional polydisperse Lennard-Jones (2DpLJ) glass-forming liquid. Based on these technical achievements we elucidated how the affine moduli for the 2D

ALT This paper presents a novel framework for simulating load-transfer mechanisms in nanocomposites reinforced by thin, stiff, and prestressed reinforcements, such as, e.g., nanosheets or nanoplatelets. The reinforcements are modeled as open material surfaces treated as membranes endowed with their own elastic properties and residual surface tension; the matrix is assumed to be isotropic elastic. In the two-dimensional plane-strain setting considered here, each material surface is represented by a sufficiently smooth curve of otherwise arbitrary shape. The theoretical part of the framework is based on the Gurtin–Murdoch theory in which the surfaces are represented by traction jumps across them. The elastic fields in the matrix are exactly expressed by the integral representations written in terms of the unknown traction jumps that are functions of the second-order derivatives of the displacements. The computational part is based on the isogeometric boundary element method (IGABEM) that uti

ALT This work develops a differentiable finite element framework based on an intrinsic cohesive-zone formulation for identifying traction-separation laws from global force-displacement histories and spatially sparse strain measurements. The framework employs physics-constrained neural networks to represent flexible, thermodynamically structured cohesive laws, with sensitivities computed through automatic differentiation and a discrete adjoint that accounts for the history-dependent evolution of interfacial damage. To enable identification of spatially heterogeneous cohesive behavior, we extend learned constitutive priors to irreversible interfacial damage by constructing a shared, low-dimensional family of cohesive laws from representative traction-separation histories. This representation exploits common constitutive structure and parameter sharing to reduce the number of unknowns and regularize the inverse problem, while allowing local cohesive behavior to vary spatially through latent c

ALT The Discrete Element Method (DEM) is an indispensable tool for modeling and understanding the basic mechanical behaviors of (dry) granular media, a class of disordered material systems of significant interest to multiple scientific domains from (geo-)physics to (chemical) engineering. At the core of all present DEM frameworks is a phenomenological friction law–commonly chosen to be of the Amontons-Coulomb type–that models the frictional interaction among constituent grains. Here, inspired by research in the tribology community, we present an alternative, fundamentally different framework, termed surface-roughness-based DEM (SR-DEM), in which contact friction is handled in a more physics-based manner. In this framework, instead of using a friction law, we explicitly resolve the surface topography of each grain and let friction emerge from collective frictionless interactions of surface asperities. In essence, we trade phenomenology for physics at the cost of an increased computational d

ALT We develop an uncertainty-aware surrogate modeling approach for coupling an atomistic kinetic Monte Carlo (KMC) surface-chemistry solver with a continuum multi-species fluid dynamics solver across disparate physical length and time scales. The surrogate model predicts boundary mass fluxes for six species and their uncertainties, which the continuum solver monitors to automatically trigger additional KMC simulations when new training data are needed. This enables efficient multiscale coupling and maintains a user-specified level of accuracy. We demonstrate the approach for reactive halide-perovskite vapor deposition, which involves a multi-species vapor jet impinging on a reactive substrate. The KMC solver uses a DFT-calibrated energetics model that we construct as part of this work and provides net deposition and desorption rates for six species. With a conservative uncertainty threshold, the surrogate-coupled approach reproduces cumulative mass fluxes of the directly coupled approach

ALT This paper presents an efficient quantum spectral-Galerkin method for constant-coefficient second-order linear elliptic equations. The method employs a boundary-adapted Jacobi tensor-product basis that incorporates the boundary condition into the approximation space and produces Galerkin systems with polynomially bounded condition numbers and sparsity. For analytic solutions satisfying a suitable uniform regularity condition, spectral convergence allows the normalized Galerkin coefficient state to be prepared with complexity poly(d,log(1/ε)), where d is the spatial dimension and ε is the error tolerance, under standard sparse-access and state-preparation assumptions. At the state-preparation level, this improves the precision dependence over fixed-order finite-difference and finite-element discretizations. We further show how to estimate prescribed linear functionals of the solution, with an additional O(1/ε) factor arising from norm and inner-product estimation.

ALT High-fidelity Reynolds-averaged Navier–Stokes (RANS) simulations remain a standard tool for aerodynamic analysis, yet each solve on an unstructured mesh is computationally expensive, making many-query tasks such as design exploration, optimization, and multi-condition evaluation costly. Neural operators provide mesh-flexible surrogates for mapping geometry and operating conditions to flow fields; however, in many existing architectures, global parameters such as the freestream state and design variables are concatenated with pointwise inputs, leaving their influence on intermediate features implicit. This paper proposes CoPiT, a Conditional Position-induced Transformer that builds on the Position-induced Transformer (PiT) as its geometric backbone and injects a global condition vector through multi-stage feature-wise linear modulation (FiLM), thereby making parametric modulation explicit while preserving geometry-based attention weights. The model is evaluated on two-dimensional AirfRA

ALT We present a coherent axion-electrodynamics extension of the OSIRIS particle-in-cell code and use it to identify and benchmark a parametric axion-photon instability supported by two equal counter-propagating electromagnetic pumps. The module evolves a cell-centered axion field together with a separately stored first-order electromagnetic response on the native staggered OSIRIS grid. Re-examining the previously proposed strictly co-linear branch shows that it is removed when the axion-Maxwell equations are reduced consistently. A closely related standing-wave geometry, however, supports a distinct Floquet instability. We derive the coupled Floquet problem for an equal-amplitude counter-propagating pump pair and obtain a reduced analytic growth rate and cutoff for the dominant long-wavelength envelope branch. For an axion mass-to-pump-frequency ratio of 0.06 and a per-beam vector-potential amplitude of 0.3, simulations at coupling strengths of 0.60, 0.80, and 1.00 reproduce both the grow