As Vision Language Models (VLMs) become increasingly accessible to farmers and agricultural experts, there is a growing need to evaluate their potential in specialized tasks. We present AgEval, a comprehensive benchmark for assessing VLMs' capabilities in plant stress phenotyping, offering a solution to the challenge of limited annotated data in agriculture. Our study explores how general-purpose VLMs can be leveraged for domain-specific tasks with only a few annotated examples, providing insights into their behavior and adaptability. AgEval encompasses 12 diverse plant stress phenotyping tasks, evaluating zero-shot and few-shot in-context learning performance of state-of-the-art models including Claude, GPT, Gemini, and LLaVA. Our results demonstrate VLMs' rapid adaptability to specialized tasks, with the best-performing model showing an increase in F1 scores from 46.24% to 73.37% in 8-shot identification. To quantify performance disparities across classes, we introduce metrics such as the coefficient of variation (CV), revealing that VLMs' training impacts classes differently, with CV ranging from 26.02% to 58.03%. We also find that strategic example selection enhances model reliability, with exact category examples improving F1 scores by 15.38% on average. AgEval establishes a framework for assessing VLMs in agricultural applications, offering valuable benchmarks for future evaluations. Our findings suggest that VLMs, with minimal few-shot examples, show promise as a viable alternative to traditional specialized models in plant stress phenotyping, while also highlighting areas for further refinement. Results and benchmark details are available at: https://github.com/arbab-ml/AgEval

G-code (Geometric code) or RS-274 is the most widely used computer numerical control (CNC) and 3D printing programming language. G-code provides machine instructions for the movement of the 3D printer, especially for the nozzle, stage, and extrusion of material for extrusion-based additive manufacturing. Currently, there does not exist a large repository of curated CAD models along with their corresponding G-code files for additive manufacturing. To address this issue, we present Slice-100K, a first-of-its-kind dataset of over 100,000 G-code files, along with their tessellated CAD model, LVIS (Large Vocabulary Instance Segmentation) categories, geometric properties, and renderings. We build our dataset from triangulated meshes derived from Objaverse-XL and Thingi10K datasets. We demonstrate the utility of this dataset by finetuning GPT-2 on a subset of the dataset for G-code translation from a legacy G-code format (Sailfish) to a more modern, widely used format (Marlin). Our dataset can be found here. Slice-100K will be the first step in developing a multimodal foundation model for digital manufacturing.

In patients with dyssynchronous heart failure (DHF), cardiac conduction abnormalities cause the regional distribution of myocardial work to be non-homogeneous. Cardiac resynchronization therapy (CRT) using an implantable, programmed biventricular pacemaker/defibrillator, can improve the synchrony of contraction between the right and left ventricles in DHF, resulting in reduced morbidity and mortality and increased quality of life. Since regional work depends on wall stress, which cannot be measured in patients, we used computational methods to investigate regional work distributions and their changes after CRT. We used three-dimensional multi-scale patient-specific computational models parameterized by anatomic, functional, hemodynamic, and electrophysiological measurements in eight patients with heart failure and left bundle branch block (LBBB) who received CRT. To increase clinical translatability, we also explored whether streamlined computational methods provide accurate estimates of regional myocardial work. We found that CRT increased global myocardial work efficiency with significant improvements in non-responders. Reverse ventricular remodeling after CRT was greatest in patients with the highest heterogeneity of regional work at baseline, however the efficacy of CRT was not related to the decrease in overall work heterogeneity or to the reduction in late-activated regions of high myocardial work. Rather, decreases in early-activated regions of myocardium performing negative myocardial work following CRT best explained patient variations in reverse remodeling. These findings were also observed when regional myocardial work was estimated using ventricular pressure as a surrogate for myocardial stress and changes in endocardial surface area as a surrogate for strain. These new findings suggest that CRT promotes reverse ventricular remodeling in human dyssynchronous heart failure by increasing regional myocardial work in early-activated regions of the ventricles, where dyssynchrony is specifically associated with hypoperfusion, late systolic stretch, and altered metabolic activity and that measurement of these changes can be performed using streamlined approaches.

The paradigm of differentiable programming has significantly enhanced the scope of machine learning via the judicious use of gradient-based optimization. However, standard differentiable programming methods (such as autodiff) typically require that the machine learning models be differentiable, limiting their applicability. Our goal in this paper is to use a new, principled approach to extend gradient-based optimization to functions well modeled by splines, which encompass a large family of piecewise polynomial models. We derive the form of the (weak) Jacobian of such functions and show that it exhibits a block-sparse structure that can be computed implicitly and efficiently. Overall, we show that leveraging this redesigned Jacobian in the form of a differentiable "layer'' in predictive models leads to improved performance in diverse applications such as image segmentation, 3D point cloud reconstruction, and finite element analysis. We also open-source the code at \url{https://github.com/idealab-isu/DSA}.

We consider the distributed training of large scale neural networks that serve as PDE (partial differential equation) solvers producing full field outputs. We specifically consider neural solvers for the generalized 3D Poisson equation over megavoxel domains. A scalable framework is presented that integrates two distinct advances. First, we accelerate training a large model via a method analogous to the multigrid technique used in numerical linear algebra. Here, the network is trained using a hierarchy of increasing resolution inputs in sequence, analogous to the `V', `W', `F' and `Half-V' cycles used in multigrid approaches. In conjunction with the multi-grid approach, we implement a distributed deep learning framework which significantly reduces the time to solve. We show scalability of this approach on both GPU (Azure VMs on Cloud) and CPU clusters (PSC Bridges2). This approach is deployed to train a generalized 3D Poisson solver that scales well to predict output full field solutions up to the resolution of 512 X 512 X 512 for a high dimensional family of inputs. This strategy opens up the possibility of fast and scalable training of neural PDE solvers on heterogeneous clusters.

Efficiently and accurately simulating partial differential equations (PDEs) in and around arbitrarily defined geometries, especially with high levels of adaptivity, has significant implications for different application domains. A key bottleneck in the above process is the fast construction of a `good' adaptively-refined mesh. In this work, we present an efficient novel octree-based adaptive discretization approach capable of carving out arbitrarily shaped void regions from the parent domain: an essential requirement for fluid simulations around complex objects. Carving out objects produces an incomplete octree. We develop efficient top-down and bottom-up traversal methods to perform finite element computations on incomplete octrees. We validate the framework by (a) showing appropriate convergence analysis and (b) computing the drag coefficient for flow past a sphere for a wide range of Reynolds numbers (O(1 - 106)) encompassing the drag crisis regime. Finally, we deploy the framework on a realistic geometry on a current project to evaluate COVID-19 transmission risk in classrooms.

We present practical algorithms for accelerating distance queries on models made of trimmed NURBS surfaces using programmable Graphics Processing Units (GPUs). We provide a generalized framework for using GPUs as coprocessors in accelerating CAD operations. By supplementing surface data with a surface bounding-box hierarchy on the GPU, we answer distance queries such as finding the closest point on a curved NURBS surface given any point in space and evaluating the clearance between two solid models constructed using multiple NURBS surfaces. We simultaneously output the parameter values corresponding to the solution of these queries along with the model space values. Though our algorithms make use of the programmable fragment processor, the accuracy is based on the model space precision, unlike earlier graphics algorithms that were based only on image space precision. In addition, we provide theoretical bounds for both the computed minimum distance values as well as the location of the closest point. Our algorithms are at least an order of magnitude faster and about two orders of magnitude more accurate than the commercial solid modeling kernel ACIS.

We present algorithms for computing accurate moments of solid models that are represented using multiple trimmed NURBS surfaces. Our algorithms make use of programmable Graphics Processing Units (GPUs) to accelerate the computations. We evaluate the surface coordinates and normals accurately, with theoretical bounds, using our GPU NURBS evaluator. We have developed a framework that makes use of this data to evaluate surface integrals of trimmed NURBS surfaces in real time. With our framework, we can compute volume and moments of solid models with theoretical guarantees. The framework also supports local geometry changes, which is useful for providing interactive feedback to the designer while the solid model is being designed. We can compute the center of mass and check for stability of the solid model interactively. Applications of such real-time moment computation include deformation modeling, animation, and physically based simulations.

We present practical algorithms for accelerating geometric queries on models made of NURBS surfaces using programmable Graphics Processing Units (GPUs). We provide a generalized framework for using GPUs as co-processors in accelerating CAD operations. By attaching the data corresponding to surface-normals to a surface bounding-box structure, we can calculate view-dependent geometric features such as silhouette curves in real time. We make use of additional surface data linked to surface bounding-box hierarchies on the GPU to answer queries such as finding the closest point on a curved NURBS surface given any point in space and evaluating the clearance between two solid models constructed using multiple NURBS surfaces. We simultaneously output the parameter values corresponding to the solution of these queries along with the model space values. Though our algorithms make use of the programmable fragment processor, the accuracy is based on the model space precision, unlike earlier graphics algorithms that were based only on image space precision. In addition, we provide theoretical bounds for both the computed minimum distance values as well as the location of the closest point. Our algorithms are at least an order of magnitude faster than the commercial solid modeling kernel ACIS.

We present algorithms for evaluating and performing modeling operations on NURBS surfaces using the programmable fragment processor on the Graphics Processing Unit (GPU). We extend our GPU-based NURBS evaluator that evaluates NURBS surfaces to compute exact normals for either standard or rational B-spline surfaces for use in rendering and geometric modeling. We build on these calculations in our new GPU algorithms to perform standard modeling operations such as inverse evaluations, ray intersections, and surface-surface intersections on the GPU. Our modeling algorithms run in real time, enabling the user to sketch on the actual surface to create new features. In addition, the designer can edit the surface by interactively trimming it without the need for retessellation. Our GPU-accelerated algorithm to perform surface-surface intersection operations with NURBS surfaces can output intersection curves in the model space as well as in the parametric spaces of both the intersecting surfaces at interactive rates. We also extend our surface-surface intersection algorithm to evaluate self-intersections in NURBS surfaces.

We present algorithms for evaluating and performing modeling operatyons on NURBS surfaces using the programmable fragment processor on the Graphics Processing Unit (GPU). We extend our GPU-based NURBS evaluator that evaluates NURBS surfaces to compute exact normals for either standard or rational B-spline surfaces for use in rendering and geometric modeling. We build on these calculations in our new GPU algorithms to perform standard modeling operations such as inverse evaluations, ray intersections, and surface-surface intersections on the GPU. Our modeling algorithms run in real time, enabling the user to sketch on the actual surface to create new features. In addition, the designer can edit the surface by interactively trimming it without the need for re-tessellation. We also present a GPU-accelerated algorithm to perform surface-surface intersection operations with NURBS surfaces that can output intersection curves in the model space as well as in the parametric spaces of both the intersecting surfaces at interactive rates.

This paper presents a new method to evaluate and display trimmed NURBS surfaces using the Graphics Processing Unit (GPU). Trimmed NURBS surfaces, the de facto standard in commercial 3D CAD modeling packages, are currently tessellated into triangles before being sent to the graphics card for display since there is no native hardware support for NURBS. Previous GPU-based NURBS display methods relied on first approximating the NURBS patches with lower degree Bezier patches before evaluation. Our method uses a GPU fragment program to evaluate the surface point coordinates of the original NURBS patch directly, from the control points and knot vectors stored as textures in graphics memory. This evaluated surface is trimmed during display using a dynamically generated trim-texture calculated via alpha blending. The implementation incorporates dynamic Level of Detail (LOD) for real-time interaction at different resolutions of the NURBS surfaces. We obtain rendering speeds at least one order of magnitude faster than evaluation using the CPU.