Searcharxiv⌕ Search

arXiv subjects

Varun Shankar

Publications and source records attributed to Varun Shankar.

At least 19 recordsLinked to original sources

Local and Sparse Kernel Operator Learning for Incompressible Flows: An Operator Lagrange Formulation

We present a local and sparse kernel method for learning the solution operators of incompressible flows. It builds on our earlier property-preserving kernel method, whose global output reconstruction and dense operator regression scale poorly. Our method replaces the global output reconstruction by local divergence-free interpolants with divergence-free polynomials of any degree, solved in parallel on the GPU. These interpolants are divergence free, periodic on periodic domains, and, with a multiscale kernel, approach Kolmogorov statistics under refinement. A compactly supported kernel, positive semidefinite in any input dimension, makes the operator regression sparse. Its support radius, and hence the sparsity, is fitted to data, and optional affine polynomials in the input capture global trends. Operator Lagrange weights join the two. They cost one sparse solve per new input and act on the training fields before the output reconstruction, so their local coefficients are never formed. We show that, unlike direct evaluation, Lagrange evaluation does not amplify solve errors by the cancellation among spatial coefficients. On ten fluid benchmarks, validation selects sparse matrices nearly everywhere, with densities that mostly decrease as the training set grows, and affine augmentation lowers the error by more than an order of magnitude for forced isotropic turbulence. These sparse models are at least as accurate as our earlier dense method and more accurate than neural operators. Lagrange evaluation is as accurate as direct evaluation with local output reconstructions, and up to nearly three orders of magnitude more accurate with global ones. We also present GPU timings and storage.

cs.LG↗

A Probabilistic, Rotation-Preserving Kernel Operator Surrogate for Continuum-Robot Design Spaces

We develop an uncertainty-aware, operator-learning-based surrogate for tendon-driven continuum robots (TDCRs) that generalizes across robot design and actuation while predicting the complete equilibrium configuration. Our central idea is a nonstationary Gaussian-process (GP) kernel whose metric is learned by a shallow neural network from mechanics-aware design coordinates. The network learns how similarity changes across design space, while the resulting kernel defines the design-to-pose operator, its posterior uncertainty, and a criterion for sequential simulator acquisition. We represent backbone position and rotational strain with spatial kernels and reconstruct orientation directly on SO(3), preserving valid material frames and tendon geometry. We use ASkotch with Nystrom preconditioning for scalable kernel operator solves and a separate Nystrom approximation for posterior covariance. Experiments show that kernel-coverage acquisition reduces mean trajectory error by approximately 15% relative to IID sampling, while posterior variance increases for increasingly unsupported designs. Compared with prior neural-operator TDCR models, the method is 30-50% more accurate using approximately 5x fewer training designs while additionally providing posterior uncertainty estimates.

cs.RO↗

KernelPack-JAX: a meshless differentiable GPU framework for node generation and forward and inverse PDE problems

We present KernelPack-JAX, a meshless, differentiable, GPU framework in JAX for solving forward and inverse PDE problems in domains whose boundaries, including internal obstacles, are given only by samples of their smooth pieces. Our node generator is based on a new stacked representation of piecewise-smooth boundaries, in which a parametric partition of unity of each smooth piece induces the supports of a curl-free implicit partition of unity of the whole boundary. This representation keeps the edges and corners of the boundary and classifies every point of space without a global fit. Boundary and interior nodes are then selected by a parallel greedy independent-set algorithm that reproduces sequential Poisson-disk sampling exactly. We discretize elliptic and parabolic problems with radial basis function-generated finite differences (RBF-FD) on stencils selected by a new visibility heuristic to avoid crossing the boundary, with ghost nodes for the boundary conditions. In addition, our framework exposes a GPU sparse direct solver to JAX as a differentiable linear solve, so that Jacobians of time-dependent problems cost a small multiple of a forward solve. We derive error estimates that separate consistency from stability and bound the parameter error by the forward error. We demonstrate high-order convergence for Darcy flow, reaction-diffusion, and advection-diffusion in 2D and 3D domains, including one with an internal obstacle, and show that the recovered parameters largely follow the forward errors. We also present timings for every stage.

math.NA↗

Localized Operator Learning with Adaptive Partition-of-Unity Mixture-of-Expert Networks

Operator learning methods such as DeepONets and FNOs often struggle with PDE families featuring sharp interfaces, heterogeneous coefficients, and localized multiscale structures. We introduce a partition-of-unity (POU) mixture-of-experts framework for localized operator learning, in which geometry-aware gating networks produce smooth spatial partitions which blend the contributions of local expert networks. Our main contribution is HiRefPOU, a residual-style hierarchical POU architecture for DeepONets that organizes localized representations through nested parent-child partitions while preserving global continuity. We also show that the same POU principle can be incorporated into Fourier Neural Operators to introduce spatial adaptivity without modifying the underlying spectral layers. On heterogeneous Darcy and reaction-diffusion benchmarks, HiRefPOU achieves substantially lower error than global DeepONet and static POU-MoE baselines, while the broader operator-learning experiments show that the benefits of localization depend on the PDE structure and the chosen neural-operator backbone. The learned partitions are interpretable and align with interfaces and regions of rapid solution variation. These results show that explicit geometric localization can improve both accuracy and interpretability in neural operator learning.

cs.LG↗

Fluids You Can Trust: Property-Preserving Operator Learning for Incompressible Flows

We present a novel property-preserving kernel-based operator learning method for incompressible flows governed by the incompressible Navier--Stokes equations. Traditional numerical solvers incur significant computational costs to respect incompressibility. Operator learning offers efficient surrogate models, but current neural operators fail to exactly enforce physical properties such as incompressibility, periodicity, and turbulence. Our kernel method maps input functions to expansion coefficients of output functions in a property-preserving kernel basis, ensuring that predicted velocity fields \emph{analytically} and \emph{simultaneously} preserve the aforementioned physical properties. We present universal approximation results and worst-case a priori convergence rates for our framework; empirically, the observed convergence rates exceed the pessimistic predictions across most benchmarks, motivating a formulation of more optimistic convergence rates. Another central contribution of this work is a novel computational framework revolving around streaming construction of kernel Gramians and recursive Schur-complement algorithms to solve the large block linear systems arising from property-preserving kernel methods. This kernel-based framework makes operator learning practical at scales up to $10{,}000$ training functions sampled at up to $10{,}000$ spatial locations. We evaluate the method on challenging 2D and 3D, laminar and turbulent, incompressible flow problems. Our method achieves up to six orders of magnitude lower relative $\ell_2$ errors upon generalization and trains up to five orders of magnitude faster compared to neural operators. Moreover, while our method enforces incompressibility analytically, neural operators exhibit large divergence errors. Our results show that our method provides an accurate and efficient surrogate for incompressible flows.

physics.flu-dyn↗

Property-Preserving Data Compression for Incompressible Flows

We develop a method for lossy compression of incompressible (i.e., divergence-free) velocity fields that preserves zero divergence and, on periodic domains, periodicity. We fit the nodal velocities with property-preserving matrix-valued kernels and compress their coefficients using standard lossy compressors. Every decompressed coefficient vector defines a field with the same constraints, independently of the compression tolerance. We present several fitting schemes designed around either low-rank representation of or sparsity in the interpolation matrix. More specifically, we show results for global Matérn interpolation, a two-level Nyström approximation, Nyström--Wendland residual interpolation, multilevel Wendland interpolation and a property-preserving (redundant) kernel frame. We derive nodal error bounds that relate the compressor tolerance to the additional velocity error. Experiments with four compressors on cavity, Taylor--Green, and species-transport flows verify the bounds under their coefficient-error assumptions and compare storage at matched velocity accuracy. The low-rank methods give compact representations when they resolve the velocity field sufficiently accurately; residual, multilevel, and frame interpolation extend the attainable accuracy when they do not. We also measure changes in turbulent velocity statistics. The resulting representations support an accuracy--compression tradeoff while preserving incompressibility and periodicity analytically.

math.NA↗

Decoding Extrahepatic Targeting of Lipid Nanoparticles with Interpretable Machine Learning

Lipid nanoparticles (LNPs) have transformed RNA medicine, yet their clinical utility remains constrained by predominant hepatic accumulation after systemic administration. Redirecting LNPs to extrahepatic tissues requires understanding of how lipid chemistry and formulation composition jointly govern in vivo biodistribution. Here, we develop an interpretable machine learning framework to predict hepatic versus extrahepatic LNP accumulation and identify molecular design rules for extrahepatic RNA delivery. A literature-derived dataset of 476 intravenous LNP formulations was curated from 81 studies, integrating formulation composition, lipid chemical structures, and IVIS-based biodistribution profiles. Standardized SMILES representations of ionizable lipids, helper lipids, sterols, PEGylated or polymer-conjugated lipids, additional lipids, and polymer repeat units were converted into RDKit Expert descriptors and combined with formulation-level variables to generate an 808-dimensional feature representation. Logistic regression, random forest, and XGBoost achieved ROC-AUC values of 0.839, 0.866, and 0.874, respectively. SHAP-based interpretation and consensus feature ranking revealed that ionizable-lipid descriptors dominate biodistribution prediction, while formulation composition, particularly ionizable lipid, sterol, and PEGylated/polymer-conjugated lipid fractions, contributes substantially. The top 20 consensus features retained nearly all predictive information in tree-based models. The most informative features implicated electrotopological surface properties, charge- and hydrophobicity-weighted surface areas, molecular topology, and amide/alkyl structural motifs as drivers of extrahepatic accumulation. This study establishes an interpretable, data-driven strategy for decoding LNP biodistribution and provides actionable design principles for engineering LNPs beyond the liver.

q-bio.QM↗

The Frame Kernel Method for Multiscale Operator Learning

We present a natively multiscale operator learning method for the surrogate modeling of (numerical solvers for) multiscale partial differential equations (PDEs). The primary novelty of our method lies in a novel multiscale kernel frame function approximation technique. Leveraging this new kernel frame technique, we cast the operator learning problem as one of learning frame coefficients of output functions as a function of frame coefficients of input functions. The generalization step then automatically allows for a multiscale decomposition of the output functions. Our method is applicable to both tensor-product grids and point clouds. We present interpolation proofs, error estimates, and numerical convergence rates for our frame approximation. We the demonstrate the applicability of our method for the surrogate modeling of inherently multiscale PDEs. The new multiscale frame kernel method is significantly more accurate than popular neural operators on challenging problems from the literature, while simultaneously admitting an a posteriori multiscale decomposition upon generalization.

cs.LG↗

Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation

We present polynomial-augmented neural networks (PANNs), a novel machine learning architecture that combines deep neural networks (DNNs) with polynomial expansions. PANNs combine the strengths of DNNs (flexibility and efficiency in higher-dimensional approximation) with those of polynomial approximation (rapid convergence rates for smooth functions). To aid in both stable training and enhanced accuracy over a variety of problems, we present (1) a family of orthogonality constraints that impose mutual orthogonality between the polynomial and the DNN within a PANN; (2) a simple basis pruning approach to combat the curse of dimensionality introduced by the polynomial component; and (3) an adaptation of a polynomial preconditioning strategy to both DNNs and polynomials. We test the resulting architecture for its polynomial reproduction properties, ability to approximate both smooth functions and functions of limited smoothness, and as a method for the solution of partial differential equations (PDEs). Through these experiments, we demonstrate that PANNs offer superior approximation properties to DNNs for both regression and the numerical solution of PDEs, while also offering enhanced accuracy over both polynomial and DNN-based regression (each) when regressing functions with limited smoothness.

cs.LG↗

A high-order, meshless, Lagrangian--Eulerian RBF-FD method for advection--diffusion--reaction on moving manifolds

We present a high-order radial basis function-generated finite difference (RBF-FD) method for partial differential equations on moving manifolds $\mathcal M(t)\subset\mathbb R^3$ of co-dimension one. Our method builds on the tangent-plane formulation of surface RBF-FD and combines Lagrangian and Eulerian treatments: the manifold and material derivative are evolved in a Lagrangian fashion, while the remaining surface differential operators are reconstructed on the instantaneous point cloud and stabilized, when necessary, by a quasi-analytical hyperviscosity formulation. Lagrangian marker drift is handled by adaptive rearrangement using a compact global parametric model of the moving surface, which also supplies accurate normals and geometry-based quadrature. After rearrangement, we reconstruct the multistep history by backward semi-Lagrangian tracing and interpolation before resuming the Lagrangian time discretization. We exploit temporal coherence through three update strategies: defect correction for the local RBF-FD weights, a curvature-based update of the hyperviscosity coefficients between spectral recomputations, and a global defect-correction iteration that reuses an incomplete LU (ILU) factorization before preconditioned generalized minimal residual (GMRES) iterations. Finally, we enforce the prescribed global mass balance through a scalar projection based on the evolving surface quadrature. Numerical experiments demonstrate high-order convergence, conservation to roundoff in source-free problems, stable long-time integration, and substantial savings from the proposed update strategies.

math.NA↗

Anisotropic representations for E(3)-equivariant machine learning coarse-grained potentials

Coarse-graining (CG) lowers the computational cost of atomistic simulations by representing groups of atoms as effective interaction sites, reducing the degrees of freedom of the system but often compromising structural fidelity or requiring system-specific parameterization. Here, we introduce a novel anisotropic machine learning CG potential that extends the point particle representation of atomic nuclei to massive ellipsoidal beads with orientation-dependent features, enabling the learning of energies, forces, and torques directly from atomistic data. The anisotropic representation is physically motivated for polar and asymmetric molecules, where directional interactions and shape anisotropy play important roles in determining structure and dynamics. Using an equivariant message-passing neural network, the model accurately reproduces radial and angular distribution functions as well as relative orientation correlations in liquid water, demonstrating that both translational and rotational dynamics are well captured. Comparison with an isotropic baseline reveals that the lack of orientation information leads to systematic errors in short and long range order and degradation of angular correlations, proving orientation features are essential for accurate coarse-graining. The anisotropic model also exposes rotational structural observables fundamentally inaccessible to isotropic representations, with minimal computational overhead. Even for coarse-graining just three degrees of freedom, CG simulations achieve 7-27$\times$ speedups while preserving structural fidelity, highlighting the efficiency gains of this systemic reduction. This framework establishes the feasibility and necessity of learned equivariant representations for anisotropic CG modeling and provides a path towards accurate and efficient mesoscopic simulations of complex molecular liquids, polymers, and biomolecular systems.

cond-mat.mtrl-sci↗

Kernel Neural Operators (KNOs) for Scalable, Memory-efficient, Geometrically-flexible Operator Learning

This paper introduces the Kernel Neural Operator (KNO), a provably convergent operator-learning architecture that utilizes compositions of deep kernel-based integral operators for function-space approximation of operators (maps from functions to functions). The KNO decouples the choice of kernel from the numerical integration scheme (quadrature), thereby naturally allowing for operator learning with explicitly-chosen trainable kernels on irregular geometries. On irregular domains, this allows the KNO to utilize domain-specific quadrature rules. To help ameliorate the curse of dimensionality, we also leverage an efficient dimension-wise factorization algorithm on regular domains. More importantly, the ability to explicitly specify kernels also allows the use of highly expressive, non-stationary, neural anisotropic kernels whose parameters are computed by training neural networks. We present universal approximation theorems showing that both the continuous and fully discretized KNO are universal approximators on operator learning problems. Numerical results demonstrate that on existing benchmarks the training and test accuracy of KNOs is closely comparable to or higher than that of popular neural operators while typically using an order of magnitude fewer trainable parameters, with the more expressive kernels proving important to attaining high accuracy. KNOs thus facilitate low-memory, geometrically-flexible, deep operator learning, while retaining the implementation simplicity and transparency of traditional kernel methods from both scientific computing and machine learning.

cs.LG↗

Neural Operators for Design-Space Surrogate Modeling of Tendon-Actuated Continuum Robots

Continuum robots enable dexterous manipulation in constrained environments, but require accurate and efficient models for real-time manipulation and control. Traditional physics-based models can be computationally expensive and may suffer from inaccuracies due to unmodeled effects, while current learning-based methods often generalize poorly beyond the specific robot on which they are trained. We present a formulation of surrogate modeling for tendon-driven continuum robots as an operator learning problem that maps robot design parameters and tendon actuation inputs to resulting configurations. This formulation enables a single trained model to generalize across a large class of robot designs. We develop four novel neural operator architectures--two based on Deep Operator Networks (DeepONets) and two based on Fourier Neural Operators (FNOs)--and train them on simulation data to predict robot configurations. All architectures achieve good accuracy while allowing for fast and accurate generalization across designs. Our results demonstrate that operator learning provides an effective and generalizable surrogate for continuum robot mechanics in the design space, enabling fast modeling for control, planning, and design optimization in surgical and industrial applications.

cs.RO↗

JetSCI: A Hybrid JAX-PETSc Framework for Scalable Differentiable Simulation

The rapid rise of scientific machine learning (SciML) has expanded the role of differentiable modeling, surrogate modeling, and data-driven constitutive laws in large-scale simulation. The JAX framework provides an attractive environment for these workflows through automatically differentiable programs, vectorization, GPU acceleration, and while enabling seamless learning of surrogate models. However, large-scale simulation still relies on mature HPC infrastructure. Libraries, such as PETSc, provide scalable MPI-based parallelism, robust linear and nonlinear solvers, and advanced preconditioning capabilities that remain difficult to reproduce in JAX-only workflows. We present JetSCI, a hybrid JAX-PETSc framework that unifies these complementary strengths. JetSCI uses JAX for GPU-parallel differentiable discretizations and PETSc for robust, scalable solution of the resulting systems on distributed-memory architectures, exposing multilevel parallelism through GPU acceleration within nodes and MPI parallelism across nodes. For finite element discretizations of heterogeneous micromechanics problems, JetSCI outperforms JAX-only implementations in efficiency and accuracy.

cs.MS↗

Deep Gaussian Processes for Functional Maps

Learning mappings between functional spaces, also known as function-on-function regression, is a fundamental problem in functional data analysis with broad applications, including spatiotemporal forecasting, curve prediction, and climate modeling. Existing approaches often struggle to capture complex nonlinear relationships and/or provide reliable uncertainty quantification when data are noisy, sparse, or irregularly sampled. To address these challenges, we propose Deep Gaussian Processes for Functional Maps (DGPFM). Our method constructs a sequence of GP-based linear and nonlinear transformations directly in function space, leveraging kernel integral transforms, GP conditional means, and nonlinear activations sampled from Gaussian processes. A key insight enables a simplified and flexible implementation: under fixed evaluation locations, discrete approximations of kernel integral transforms reduce to direct functional integral transforms, allowing seamless integration of diverse transform designs. To support scalable probabilistic inference, we adopt inducing points and whitening transformations within a variational learning framework. Empirical results on both real-world and synthetic benchmark datasets demonstrate the advantages of DGPFM in terms of predictive accuracy and uncertainty calibration.

cs.LG↗

Triplet Envelope Functions for increasing machine learning interatomic potential efficiency and stability

Central to interatomic potential efficiency is the radial envelope function that enables linear scaling with computational cost by defining a local neighborhood of atoms. This has enabled MLIPs to revolutionize materials science over the past decade by providing DFT accuracy with linear scaling computational cost in molecular dynamics workflows. However, MLIPs still have a relatively high computational cost compared to empirical interatomic potentials, preventing them from transforming molecular dynamics workflows. A central issue is that MLIPs use relatively large cutoff radii, converging to 6A over the last few years. The large cutoffs prioritize accuracy of any material over efficiency in any particular region of phase space, capturing dispersion effects and low density materials at the expense of increased computational cost in higher density materials. Past work has aimed to address this with KNN graph sparsification, which, while significantly reducing cost, has the drawback of breaking energy conservation. In this work, we propose higher-order envelope functions that prune local atomic neighborhoods through physically inspired geometric functions to provide the memory and efficiency benefits of KNN graph sparsification while eliminating non-conservative energy dynamics. Through numerical experiments on solids and liquids with 5-8A cutoffs, we show that triplet envelope functions complement radial envelope functions by doubling training and inference speed, tripling memory efficiency, and increasing simulation stability while not impacting accuracy or data efficiency for the most common 6A cutoff. Moreover, experiments with 8A radial cutoffs show triplet envelope functions create a pathway to larger cutoff radii for efficiently and accurately modeling open structures with large interatomic distances, showing a promising new direction for engineering MLIP efficiency.

cond-mat.mtrl-sci↗

HyResPINNs: A Hybrid Residual Physics-Informed Neural Network Architecture Designed to Balance Expressiveness and Trainability

Physics-informed neural networks (PINNs) have emerged as a powerful approach for solving partial differential equations (PDEs) by training neural networks with loss functions that incorporate physical constraints. In this work, we introduce HyResPINNs, a two-level convex-gated architecture designed to maximize approximation expressiveness for a fixed number of degrees of freedom (DoF). The first level involves a trainable, per-block combination of smooth basis functions with trainable sparsity, and deep neural networks; the second involves the ability to gate entire blocks (much like in ResNets or Highway Nets), allowing for expressivity along the depth dimension of the architecture. Our empirical evaluation on a diverse set of challenging PDE problems demonstrates that HyResPINNs consistently achieve superior accuracy to baseline methods while remaining competitive relative to training times. These results highlight the potential of HyResPINNs to combine desirable features from traditional scientific computing methods and modern machine learning, paving the way for more robust and expressive approaches to physics-informed modeling.

cs.LG↗

An Optimal Weighted Least-Squares Method for Operator Learning

We consider the problem of learning an unknown, possibly nonlinear operator between separable Hilbert spaces from supervised data. Inputs are drawn from a prescribed probability measure on the input space, and outputs are (possibly noisy) evaluations of the target operator. We regard admissible operators as square-integrable maps with respect to a fixed approximation measure, and we measure reconstruction error in the corresponding Bochner norm. For a finite-dimensional approximation space $V$ of dimension $N$, we study weighted least squares estimators in $V$ and establish probabilistic stability and accuracy bounds in the Bochner norm. We show that there exist sampling measures and weights - defined via an operator-level Christoffel function - that yield uniformly well-conditioned Gram matrices and near-optimal sample complexity, with a number of training samples $M$ on the order of $N \log N$. We complement the analysis by constructing explicit operator approximation spaces in cases of interest: rank-one linear operators that are dense in the class of bounded linear operators, and rank-one polynomial operators that are dense in the Bochner space under mild assumptions on the approximation measure. For both families we describe implementable procedures for sampling from the associated optimal measures. Finally, we demonstrate the effectiveness of this framework on several benchmark problems, including learning solution operators for the Poisson equation, viscous Burgers' equation, and the incompressible Navier-Stokes equations.

math.NA↗