Publications
Publications by the Baskar Group, in reverse chronological order.
2026
- SAGE: Scalable Agentic Grounded Evaluation for Crop Disease DiagnosisIowa State University Digital Repository (Iowa State University), 2026Preprint
Summary
We build an AI system that uses vision-language models as reasoning agents to diagnose crop diseases across many crops and field conditions, even where labeled disease images are scarce. - Microstructure inference of organic thin films via light modulated photocurrent characterizationOrganic Electronics, 2026
- Cryogenic transmission electron microscopy reveals assembly and nanostructure of PEDOT:PSSNature Communications, 2026
Summary
We use cryogenic electron microscopy to watch how a common conducting polymer assembles into fibers in solution and in the finished film. Seeing this nanostructure directly helps explain and improve flexible, wearable electronics. - Neural Geometry for PDEs: Regularity, Stability, and Convergence GuaranteesOpen MIND, 2026
Summary
We study when neural network representations of shapes are smooth enough to trust in physics simulations, and prove the conditions under which they give reliable, convergent results. - HS-3D-NeRF: 3D Surface and Hyperspectral Reconstruction From Stationary Hyperspectral Images Using Multi-Channel NeRFsOpen MIND, 2026
Summary
We reconstruct produce and plants in 3D and across the full color spectrum from a fixed hyperspectral camera and a rotating sample, giving high-throughput measurements of quality and traits without moving the camera. - Procedural Volumetric Modeling of Plant Branching Structures for Finite Element AnalysisarXiv (Cornell University), 2026Preprint
Summary
We generate detailed 3D models of plant branching that are ready for engineering simulation, providing the geometry needed to study water and nutrient transport, light capture, and structural loads. - A High Throughput Framework for Large Scale Building Energy Simulation: From Real-Time Alerts to AI-Ready SurrogatesIn , 2026
Summary
We build a portable high-throughput system that runs large ensembles of building energy simulations at scale, turning a slow tool into one usable for optimization, city-scale modeling, and training AI surrogates. - Neural-Network-based Viscosity Closure for Non-Newtonian Multiphase FlowsIowa State University Digital Repository (Iowa State University), 2026Preprint
Summary
We train a neural network to represent the complex viscosity of printing materials, so simulations of non-Newtonian multiphase flows can handle new materials without hand-tuning a rheology model. - ADKO: Agentic Decentralized Knowledge OptimizationIowa State University Digital Repository (Iowa State University), 2026Preprint
Summary
We introduce a framework where independent agents jointly optimize a shared goal without sharing their private data, combining their Gaussian process models efficiently. - From Simulation to Discovery: AI Enabled Probabilistic Emulation of Mechanistic Crop SystemsIowa State University Digital Repository (Iowa State University), 2026Preprint
Summary
We build a fast neural emulator of a detailed crop model that predicts how maize responds to climate and genetics, making large-scale exploration of genotype and environment practical. - Field conserving adaptive mesh refinement (AMR) scheme on massively parallel adaptive octree meshesOpen MIND, 2026
Summary
We develop an adaptive mesh method that keeps conserved quantities from drifting over long simulations, so large parallel runs stay physically accurate as the mesh refines and coarsens. - Data-driven optimal control with neural network modeling of gradient flowsEngineering With Computers, 2026
- AI-integrated models for assessing agricultural resiliencearXiv (Cornell University), 2026Preprint
Summary
We build an AI tool that links economic and crop models so policy makers and markets can ask how shocks ripple through agricultural supply chains. - MolGen-Transformer: A molecule language model for the generation and latent space exploration of organic moleculesComputational Materials Science, 2026
2025
- Leveraging Vision Language Models for Specialized Agricultural TasksIn , 2025
Summary
We create a benchmark, AgEval, that measures how well vision-language models handle specialized plant-stress tasks, guiding their use in agriculture where labeled data is limited. - 3D multiphase heterogeneous microstructure generation using conditional latent diffusion modelsDigital Discovery, 2025
Summary
We use a conditional diffusion model to generate 3D multiphase material microstructures on demand with targeted properties, speeding up the design of advanced materials. - Accelerating space-time methods using physics-informed neural networksJournal of Computational Physics, 2025
Summary
We use physics-informed neural networks to speed up space-time simulation methods, reaching accurate solutions with less computational effort. - AI-assisted Image-Based Phenotyping Reveals Genetic Architecture of Pod Traits in Mungbean ( Vigna radiata L.)bioRxiv (Cold Spring Harbor Laboratory), 2025Preprint
Summary
We use AI image analysis to measure mungbean pod traits across hundreds of varieties and map the genetics behind them, supporting breeding for plant-based protein. - Time Series GWAS for Iron Deficiency Chlorosis Tolerance in Soybean using Aerial ImagerybioRxiv (Cold Spring Harbor Laboratory), 2025Preprint
Summary
We use drone imagery collected over time to score iron deficiency chlorosis in soybean and map the genetics of tolerance, replacing slow visual ratings. - TerraIncognita: A Dynamic Benchmark for Species Discovery Using Frontier ModelsarXiv (Cornell University), 2025Preprint
Summary
We introduce TerraIncognita, a benchmark that tests how well frontier AI can discover unknown insect species from images, a task central to tracking biodiversity loss. - Evaluating Molecular Similarity Measures: Do Similarity Measures Reflect Electronic Structure Properties?Journal of Chemical Information and Modeling, 2025
Summary
We test whether common molecular similarity measures actually reflect the electronic properties that matter for chemistry, an important check for machine learning in molecular design. - Robust soybean seed yield estimation using high-throughput ground robot videosFrontiers in Plant Science, 2025
Summary
We estimate soybean seed yield by counting seeds in video from a ground robot using computer vision, a faster and cheaper alternative to traditional yield measurement. - GRATEv2: computational tools for real-time analysis of high-throughput high-resolution TEM (HRTEM) images of conjugated polymersMaterials Advances, 2025
Summary
We build a near real-time pipeline that extracts structural measurements from high-resolution electron microscopy of polymers, auto-tuning itself and signaling when enough data has been collected. - High-resolution thermal simulation framework for extrusion-based additive manufacturing of complex geometriesFinite Elements in Analysis and Design, 2025
- Digital twins for the plant sciencesTrends in Plant Science, 2025
Summary
We lay out how digital twins, virtual models kept in step with real plants, can change how the plant sciences predict, design, and manage crops. - Assessing the cybersecurity of connected 3D printers using large language models (LLMs)Manufacturing Letters, 2025
Summary
We use large language models to probe the cybersecurity of networked 3D printers, showing how tampered instructions can compromise additive manufacturing and how such attacks might be detected. - FloraForge: LLM-Assisted Procedural Generation of Editable and Analysis-Ready 3D Plant Geometric Models For Agricultural ApplicationsarXiv (Cornell University), 2025Preprint
Summary
We introduce FloraForge, which uses a language model to generate editable, analysis-ready 3D plant models from natural language, avoiding the heavy training data and expertise other methods need. - Procedural generation of 3D maize plant architecture from LiDAR dataComputers and Electronics in Agriculture, 2025
- Assessing phenotypic diversity and sensor‐based metrics for drought response in soybeanCrop Science, 2025
Summary
We screen a diverse soybean panel for drought response using sensor-based field measurements, identifying traits and metrics that help breed for drought resilience. - Direct flow simulations with implicit neural representation of complex geometryComputer Methods in Applied Mechanics and Engineering, 2025
Summary
We simulate fluid flow around complicated shapes by representing the geometry with a neural network instead of a traditional mesh. This makes direct simulation practical for objects that are hard to mesh. - Mechanics simulation with Implicit Neural Representations of complex geometriesComputer-Aided Design, 2025
- Mesh-Free Mechanics Simulations Using Implicit Neural Representations of Complex GeometriesSSRN Electronic Journal, 2025Preprint
- A Semi-Implicit Variational Multiscale Formulation for the Incompressible Navier-Stokes Equations via Exact Adjoint LinearizationarXiv (Cornell University), 2025Preprint
Summary
We develop a fast, stable finite element method for incompressible flow that linearizes the equations at each step, making large flow simulations more efficient. - Solving fluid flow problems in space-time with multiscale stabilization: Formulation and examplesComputers & Mathematics with Applications, 2025
- Space-time finite element analysis of the advection-diffusion equation using Galerkin/least-square stabilizationComputers & Mathematics with Applications, 2025
- Soybean maturity prediction using two‐dimensional contour plots from drone‐based time series imageryThe Plant Phenome Journal, 2025
Summary
We predict soybean days to maturity from drone image time series using two-dimensional contour plots, giving breeders an automated alternative to visual maturity ratings. - SC-NeRF: NeRF-Based Point Cloud Reconstruction Using a Stationary Camera for Agricultural ApplicationsIn , 2025
Summary
We reconstruct 3D plant point clouds from a single fixed camera and a rotating sample, making high-throughput phenotyping practical where moving cameras are not. - Genomic and phenomic prediction for soybean seed yield, protein, and oilThe Plant Genome, 2025
Summary
We combine genomic and image-based phenomic prediction to forecast soybean yield, protein, and oil, improving how breeders select promising varieties. - Constructing generalizable microstructure–property maps across diverse microstructure classesMRS Communications, 2025
Summary
We test whether models linking material microstructure to properties can transfer across very different microstructure types, a step toward general data-driven materials design. - AI-guided high-throughput investigation of conjugated polymer doping reveals importance of local polymer order and dopant-polymer separationMatter, 2025
- Toward a general framework for AI-enabled prediction in crop improvementTheoretical and Applied Genetics, 2025
Summary
We propose a framework for combining symbolic and machine-learning prediction in crop improvement, showing how it can sharpen predictions as biological systems grow more complex. - Real time 3D reconstruction for enhanced cybersecurity of additive manufacturing processesJournal of Manufacturing Processes, 2025
- Enhancing yield prediction from plot-level satellite imagery through genotype and environment feature disentanglementFrontiers in Plant Science, 2025
Summary
We improve in-season crop yield prediction from plot-level satellite imagery by separating genotype and environment signals, helping breeders and growers allocate resources. - 3D Neural Operator-Based Flow Surrogates around 3D geometries: Signed Distance Functions and Derivative ConstraintsarXiv (Cornell University), 2025Preprint
Summary
We build neural operator surrogates that predict 3D fluid flow around complex shapes from a signed distance representation, cutting the cost of high-fidelity simulation for design. - Benchmarking scientific machine-learning approaches for flow prediction around complex geometriesCommunications Engineering, 2025
Summary
We benchmark scientific machine learning methods for predicting flow around complex geometries, moving the field beyond the simple shapes most studies rely on. - Predicting Time-Dependent Flow Over Complex Geometries Using Operator NetworksarXiv (Cornell University), 2025Preprint
Summary
We present a geometry-aware operator network that predicts unsteady flow over complex shapes from a signed distance field and flow history, generalizing across geometries. - Accessing the effect of phyllotaxy and planting density on light interception in field-grown maize using 3D reconstructionsSmart Agricultural Technology, 2025
Summary
We study how maize leaf arrangement and planting density affect light capture in the field, using 3D models to understand how dense planting can still yield well. - Accessing the Effect of Phyllotaxy and Planting Density on Light Use Efficiency in Field-Grown Maize Using 3d ReconstructionsSSRN Electronic Journal, 2025Preprint
-
- Octree-Based Shifted Boundary Method: Evaluating the impact of hanging-node removal on convergence and solver performance for linear PDEsAdvances in Computational Science and Engineering, 2025
Summary
We study how removing hanging nodes in octree-based adaptive meshes affects the accuracy and convergence of the shifted boundary method, sharpening finite element analysis over complex geometries. - WeedNet: A Foundation Model-Based Global-to-Local AI Approach for Real-Time Weed Species Identification and ClassificationResearch Square, 2025Preprint
Summary
We build WeedNet, a foundation-model approach that identifies weed species from images in real time, combining global and local recognition to support weed management. - Plot‐level satellite imagery can substitute for UAVs in assessing maize phenotypes across multistate field trialsPlants People Planet, 2025
Summary
We show that plot-level satellite imagery can stand in for drones when assessing maize traits across many sites, making large multistate breeding trials easier to phenotype. -
- Finding sustainable, resilient, and scalable solutions for future indoor agriculturenpj Science of Plants, 2025
Summary
We examine the technologies that can make indoor farming more sustainable and affordable, from life cycle analysis and digital twins to flexible energy use and engineered plants. We argue that real progress needs teams that cross traditional disciplines. - MolGen-Transformer: A molecule language model for the generation and latent space exploration of pi-conjugated moleculesChemRxiv, 2025Preprint
Summary
We build a molecular language model that generates and explores new pi-conjugated molecules, widening the space of candidate materials for organic electronics. - Octree-based adaptive mesh refinement and the shifted boundary method for efficient fluid dynamics simulationsAdvances in Computational Science and Engineering, 2025
Summary
We combine octree-based adaptive mesh refinement with the shifted boundary method to simulate incompressible and thermal flows efficiently, refining the mesh where the flow is most active. - A shifted boundary method for thermal flowsJournal of Computational Physics, 2025
Summary
We extend the shifted boundary method, which avoids fitting a mesh exactly to complex shapes, to flows that also carry heat. It handles complicated geometries without the cost of remeshing. - Simulating incompressible flows over complex geometries using the shifted boundary method with incomplete adaptive octree meshesJournal of Computational Physics, 2025
- Towards Large Reasoning Models for AgriculturearXiv (Cornell University), 2025Preprint
Summary
We develop large reasoning models tailored to agriculture, where sound decisions depend on local climate, economics, and practice, going beyond what general language models handle. - MaizeEar-SAM: Zero-Shot Maize Ear PhenotypingarXiv (Cornell University), 2025Preprint
Summary
We use a zero-shot segmentation model to measure maize ear yield traits from images without task-specific training, supporting genetics research and breeding.
2024
- Evaluating Neural Radiance Fields for 3D Plant Geometry Reconstruction in Field ConditionsPlant Phenomics, 2024
Summary
We evaluate neural radiance field methods for reconstructing plants in 3D from indoor to field conditions, showing where they capture the fine geometric detail phenotyping needs. -
- Identifying representative sub-domains in 3D microstructures for accelerated structure–property mapping in organic photovoltaicComputational Materials Science, 2024
- Rapid Estimation of the Intermolecular Electronic Couplings and Charge-Carrier Mobilities of Crystalline Molecular Organic Semiconductors through a Machine Learning PipelineThe Journal of Physical Chemistry Letters, 2024
Summary
We rapidly estimate the electronic couplings and charge mobilities of crystalline organic semiconductors, speeding the screening of candidate materials for electronics. - Leveraging soil mapping and machine learning to improve spatial adjustments in plant breeding trialsCrop Science, 2024
Summary
We use soil maps and machine learning to correct for field variability in plant breeding trials, giving more accurate yield estimates and better variety selection. - InsectNet: Real-time identification of insects using an end-to-end machine learning pipelinePNAS Nexus, 2024
Summary
We build a machine learning system that identifies insects from ordinary photos in real time, covering both pests and beneficial species. It holds up across life stages and messy field conditions, giving growers fast, reliable identification. - In the Mix: A Workshop Merging Computational Chemistry and Electrochemistry Alongside Data ScienceJournal of Chemical Education, 2024
Summary
We describe In the Mix, a graduate-student-led workshop that brings together computational chemistry, electrochemistry, and data science to train researchers in cross-disciplinary work. - Zero‐shot insect detection via weak language supervisionThe Plant Phenome Journal, 2024
Summary
We detect insects in images with little labeled data by using weak language supervision, tapping large citizen-science image collections to build agricultural AI. - Persistent monitoring of insect-pests on sticky traps through hierarchical transfer learning and slicing-aided hyper inferenceFrontiers in Plant Science, 2024
Summary
We continuously monitor insect pests on sticky traps using transfer learning and image slicing, giving reliable automated pest counts for crop protection. - AIIRA: AI Institute for Resilient AgricultureAI Magazine, 2024
Summary
We lay out the vision of AIIRA, the AI Institute for Resilient Agriculture, which fuses diverse data with domain knowledge to model plants across scales for more resilient farming. -
- Evaluating Large Language Models for G-Code Debugging, Manipulation, and ComprehensionIn , 2024
Summary
We test how well large language models can read, debug, and manipulate G-code, the language that controls 3D printers, toward more reliable additive manufacturing. - STITCH: Surface reconstrucTion using Implicit neural representations with Topology Constraints and persistent HomologyarXiv (Cornell University), 2024Preprint
Summary
We reconstruct surfaces from sparse, irregular point clouds while guaranteeing a single connected shape, using persistent homology to enforce the right topology. - Multi‐sensor and multi‐temporal high‐throughput phenotyping for monitoring and early detection of water‐limiting stress in soybeanThe Plant Phenome Journal, 2024
Summary
We combine multiple sensors over time for high-throughput phenotyping that detects water stress in soybean early, helping breeders and growers respond to drought. - NeuFENet: neural finite element solutions with theoretical bounds for parametric PDEsEngineering With Computers, 2024
-
- AgGym: An agricultural biotic stress simulation environment for ultra-precision management planningarXiv (Cornell University), 2024Preprint
Summary
We build AgGym, a simulation environment for crop biotic stress that lets managers plan ultra-precise, targeted use of fungicides, insecticides, and herbicides. - Direct numerical simulation of electrokinetic transport phenomena in fluids: Variational multi-scale stabilization and octree-based mesh refinementJournal of Computational Physics, 2024
- Active learning for regression of structure–property mapping: the importance of sampling and representationDigital Discovery, 2024
Summary
We develop an active-learning workflow that chooses which microstructures to evaluate, calibrating structure-property models with far fewer expensive property calculations. - Disentangling genotype and environment specific latent features for improved trait prediction using a compositional autoencoderFrontiers in Plant Science, 2024
Summary
We separate genotype-specific and environment-specific features when predicting plant traits, improving trait prediction over standard dimensionality-reduction methods. - Modeling and simulations of high-density two-phase flows using projection-based Cahn-Hilliard Navier-Stokes equationsarXiv (Cornell University), 2024Preprint
Summary
We simulate high-density-ratio two-phase flows, such as molten metal oscillating in microgravity, using a projection-based Cahn-Hilliard Navier-Stokes method. - Out-of-Distribution Detection Algorithms for Robust Insect ClassificationPlant Phenomics, 2024
Summary
We add out-of-distribution detection to insect classifiers so they flag unfamiliar species instead of guessing, making image-based pest identification more trustworthy. - Class‐specific data augmentation for plant stress classificationThe Plant Phenome Journal, 2024
Summary
We develop a class-specific data augmentation strategy that picks the most effective image transformations for each plant stress category, improving deep learning classifiers for stress identification. - FlowBench: A Large Scale Benchmark for Flow Simulation over Complex GeometriesarXiv (Cornell University), 2024Preprint
Summary
We release FlowBench, a large benchmark of fluid flow simulations over complex geometries, so scientific machine learning models can be trained and tested on realistic flow physics. - Data driven discovery and quantification of hyperspectral leaf reflectance phenotypes across a maize diversity panelThe Plant Phenome Journal, 2024
Summary
We use hyperspectral leaf reflectance to derive plant traits that are otherwise slow or costly to measure by hand, discovering and quantifying reflectance-based phenotypes. - Soybean Canopy Stress Classification Using 3D Point Cloud DataAgronomy, 2024
Summary
We classify canopy stress in soybean from three-dimensional point cloud data, capturing plant structure that flat top-down drone images miss.
2023
- Machine Learning Identifies Strong Electronic Contacts in Semiconducting Polymer MeltsMacromolecules, 2023
Summary
We use machine learning to find the local molecular arrangements that create strong electronic coupling in semiconducting polymers, guiding the design of better organic electronic materials. - Electrokinetic Enrichment and Label-Free Electrochemical Detection of Nucleic Acids by Conduction of Ions along the Surface of Bioconjugated BeadsACS Sensors, 2023
Summary
We combine electrokinetic pre-enrichment of nucleic acids on probe-modified microbeads with label-free electrochemical detection, improving sensitivity for nucleic acid sensing. - A Comprehensive Study on Soybean Yield Prediction Using Soil and Hyperspectral Reflectance DataPreprints.org, 2023Preprint
Summary
We predict soybean yield by combining soil data with hyperspectral reflectance from plants and soil, comparing models to see which data sources help breeders most. - Deep learning powered real-time identification of insects using citizen science dataarXiv (Cornell University), 2023Preprint
Summary
We build a deep learning system that identifies insects in real time from citizen-science images, recognizing both beneficial insects and harmful pests for integrated pest management. - Dissecting the genetic architecture of leaf morphology traits in mungbean (Vigna radiata (L.) Wizcek) using genome‐wide association studyThe Plant Phenome Journal, 2023
Summary
We map the genetics of leaf shape in mungbean, an increasingly important protein crop, linking measurable leaf traits to the genomic regions that control them. - Self-supervised maize kernel classification and segmentation for embryo identificationFrontiers in Plant Science, 2023
Summary
We use self-supervised learning to classify and segment maize kernels and identify the embryo without heavy manual labeling, supporting seed quality and food security work. - 3D reconstruction of plants using probabilistic voxel carvingComputers and Electronics in Agriculture, 2023
- Generating Finite Element Codes combining Adaptive Octrees with Complex GeometriesarXiv (Cornell University), 2023Preprint
Summary
We build a domain-specific language that generates finite element code on adaptive octree meshes for complex geometries, letting scientists write high-level PDE solvers that run efficiently in parallel. - Towards Foundational AI Models for Additive Manufacturing: Language Models for G-Code Debugging, Manipulation, and ComprehensionarXiv (Cornell University), 2023Preprint
Summary
We explore large language models as a foundation for additive manufacturing, testing how well they read and reason about the G-code that drives 3D printers. - Self‐supervised learning improves classification of agriculturally important insect pests in plantsThe Plant Phenome Journal, 2023
Summary
We show that self-supervised learning improves classification of agriculturally important insects when labeled images are scarce, helping automate pest monitoring. - Direct numerical simulation of electrokinetic transport phenomena: variational multi-scale stabilization and octree-based mesh refinementIowa State University Digital Repository (Iowa State University), 2023Preprint
Summary
We develop a variational finite element method for direct numerical simulation of electrokinetic transport governed by the Poisson-Nernst-Planck equations, aiding the design of electrochemical and electrokinetic devices. - In-Droplet Electromechanical Cell Lysis and Enhanced Enzymatic Assay Driven by Ion Concentration PolarizationAnalytical Chemistry, 2023
Summary
We drive cell lysis and enzymatic assays inside individual droplets using electromechanical forces, enabling molecular analysis of small numbers of cells. - Deep learning-based 3D multigrid topology optimization of manufacturable designsEngineering Applications of Artificial Intelligence, 2023
- Cyber-agricultural systems for crop breeding and sustainable productionTrends in Plant Science, 2023
Summary
We describe the cyber-agricultural system, a framework that combines widespread sensing, artificial intelligence, and smart machines to speed up crop breeding and make production more efficient and sustainable. - CyRSoXS: a GPU-accelerated virtual instrument for polarized resonant soft X-ray scatteringJournal of Applied Crystallography, 2023
Summary
We build CyRSoXS, a GPU-accelerated virtual instrument that simulates polarized resonant soft X-ray scattering, letting researchers link molecular structure to scattering signatures far faster than before. - Scalable adaptive algorithms for next-generation multiphase flow simulationsIn , 2023
Summary
We develop scalable adaptive algorithms for high-fidelity multiphase flow simulation, refining the mesh only where the interface needs it so large runs stay efficient on parallel machines. - A computational framework for transmission risk assessment of aerosolized particles in classroomsEngineering With Computers, 2023
- Optimal surrogate boundary selection and scalability studies for the shifted boundary method on octree meshesComputer Methods in Applied Mechanics and Engineering, 2023
Summary
We made the shifted boundary method, which avoids fitting a mesh exactly to complex shapes, scalable on adaptive octree grids. It has grown into a family of solvers for flows and heat transfer over complicated geometries. - “Canopy fingerprints” for characterizing three-dimensional point cloud data of soybean canopiesFrontiers in Plant Science, 2023
Summary
We introduce canopy fingerprints, compact descriptors of three-dimensional plant canopy point clouds that capture fine structural detail for high-throughput phenotyping.
2022
- Protocols for In Vivo Doubled Haploid (DH) Technology in Maize Breeding: From Haploid Inducer Development to Haploid Genome DoublingIn Methods in molecular biology, 2022
- Physics-aware machine learning surrogates for real-time manufacturing digital twinManufacturing Letters, 2022
- Out-of-plane faradaic ion concentration polarization: stable focusing of charged analytes at a three-dimensional porous electrodeLab on a Chip, 2022
Summary
We demonstrate out-of-plane faradaic ion concentration polarization, a microfluidic technique that stably focuses charged biomolecules for preconcentration and bioanalysis. - Electronic, redox, and optical property prediction of organic π-conjugated molecules through a hierarchy of machine learning approachesChemical Science, 2022
Summary
We train a hierarchy of machine learning models to predict the electronic, redox, and optical properties of organic molecules. This lets researchers screen many more candidate molecules for energy and electronics uses in far less time. - Dissecting the Root Phenotypic and Genotypic Variability of the Iowa Mung Bean Diversity PanelFrontiers in Plant Science, 2022
Summary
We characterize root traits across the Iowa mung bean diversity panel and map their genetics, supporting the breeding of this drought-tolerant, protein-rich crop. - Feature Engineering for Microstructure–Property Mapping in Organic PhotovoltaicsIntegrating materials and manufacturing innovation, 2022
- Breakup dynamics in primary jet atomization using mesh- and interface- refined Cahn-Hilliard Navier-StokesarXiv (Cornell University), 2022Preprint
Summary
We perform interface-resolved simulations of jet breakup in two-phase flows using an adaptively refined Cahn-Hilliard Navier-Stokes method that sharpens the interface where it matters. - A fully-coupled framework for solving Cahn-Hilliard Navier-Stokes equations: Second-order, energy-stable numerical methods on adaptive octree based meshesComputer Physics Communications, 2022
- A projection-based, semi-implicit time-stepping approach for the Cahn-Hilliard Navier-Stokes equations on adaptive octree meshesJournal of Computational Physics, 2022
Summary
We develop a fast, stable time-stepping method for simulating two fluids that mix and separate, running on adaptive grids that add detail only where it is needed. This makes large two-phase flow simulations affordable. - Computational framework for resolving boundary layers in electrochemical systems using weak imposition of Dirichlet boundary conditionsFinite Elements in Analysis and Design, 2022
- Stochastic Conservative Contextual Linear BanditsIn 2022 IEEE 61st Conference on Decision and Control (CDC), 2022
Summary
We formulate conservative contextual linear bandits for real-time decisions under adversarial contexts, guaranteeing safety constraints are met while the model keeps learning. - How important is microstructural feature selection for data-driven structure-property mapping?MRS Communications, 2022
- Optimization framework for
patient‐specific modeling under uncertaintyInternational Journal for Numerical Methods in Biomedical Engineering, 2022Summary
We build an optimization-based uncertainty quantification framework for calibrating patient-specific computational models from unreliable clinical data. - Plant phenotyping with limited annotation: Doing more with lessThe Plant Phenome Journal, 2022
Summary
We show how deep learning can extract plant traits from images with far fewer labeled examples, making phenotyping practical when annotation is scarce. - A graph based approach to model charge transport in semiconducting polymersnpj Computational Materials, 2022
Summary
We introduce a graph based method to model how electric charge moves through semiconducting polymers, capturing how molecular packing shapes performance. It reproduces known results while staying fast enough for large studies. - Algorithm 1025: PARyOpt: A Software for P arallel A synchronous R emote Ba y esian Opt imizationACM Transactions on Mathematical Software, 2022
Summary
We release PARyOpt, a Python Bayesian optimization library built for remote, asynchronous, and expensive function evaluations such as large simulations. - Computational characterization of charge transport resiliency in molecular solidsMolecular Systems Design & Engineering, 2022
Summary
We characterize how resilient charge transport is in molecular materials by treating each molecule as a graph and measuring the centrality of its charge pathways, guiding organic electronics design. - Deep learning-based phenotyping for genome wide association studies of sudden death syndrome in soybeanFrontiers in Plant Science, 2022
Summary
We use deep learning to phenotype soybean disease severity from images and feed those measurements into genome-wide association studies, helping breed for disease resistance. - Simulation-guided analysis of resonant soft X-ray scattering for determining the microstructure of triblock copolymersMolecular Systems Design & Engineering, 2022
Summary
We use simulation to interpret resonant soft X-ray scattering and determine the nanoscale morphology of multiblock copolymers, sharpening a chemically sensitive characterization tool. - A vorticity-based criterion to characterise leading edge dynamic stall onsetJournal of Fluid Mechanics, 2022
Summary
We propose a vorticity-based criterion, the boundary enstrophy flux, that gives a physically grounded and conservative way to detect the onset of leading-edge dynamic stall. - A scalable adaptive-matrix SPMV for heterogeneous architecturesIn 2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS), 2022
Summary
We design a scalable adaptive-matrix sparse matrix-vector product that runs efficiently across CPU and GPU architectures, speeding the core kernel of many parallel simulation codes. - Construction and high throughput exploration of phase diagrams of multi-component organic blendsComputational Materials Science, 2022
- Multi-fidelity machine learning models for structure–property mapping of organic electronicsComputational Materials Science, 2022
2021
- Distributed multigrid neural solvers on megavoxel domainsIn , 2021
Summary
We train large neural PDE solvers across many GPUs using a distributed multigrid scheme, producing full-field solutions on very large three-dimensional domains. - Differentiable Spline ApproximationsNeural Information Processing Systems, 2021
Summary
We extend gradient-based optimization to splines and other piecewise-smooth models by deriving accurate differentiable approximations, broadening what differentiable programming can fit. - Designing asymmetrically modified nanochannel sensors using virtual EISElectrochimica Acta, 2021
- Impedance-Based Nanoporous Anodized Alumina/ITO Platforms for Label-Free BiosensorsACS Applied Materials & Interfaces, 2021
Summary
We fabricate and model an impedance-based, label-free biosensor built on nanoporous anodized alumina and ITO, achieving sensitive detection without redox labels. - UAS-Based Plant Phenotyping for Research and Breeding ApplicationsPlant Phenomics, 2021
Summary
We review how unmanned aircraft systems enable flexible, high-throughput plant phenotyping for research and breeding, surveying platforms, sensors, and analysis workflows. - Using Machine Learning to Develop a Fully Automated Soybean Nodule Acquisition Pipeline (SNAP)Plant Phenomics, 2021
Summary
We build a fully automated machine learning pipeline that acquires and counts soybean root nodules from images, quantifying a key nitrogen-fixation trait at scale. - Self-supervised agricultural insect pest classification2021Preprint
Summary
We use self-supervised learning to classify agricultural insect pests when labeled data is limited, easing automated pest monitoring for farmers. - DiffNet: Neural Field Solutions of Parametric Partial Differential EquationsarXiv (Cornell University), 2021Preprint
Summary
We introduce DiffNet, a mesh-based neural network that predicts full-field solutions to parametric partial differential equations, offering a differentiable alternative to conventional solvers. - Interdisciplinary strategies to enable data-driven plant breeding in a changing climateOne Earth, 2021
- Fast inverse design of microstructures via generative invariance networksNature Computational Science, 2021
Summary
We built generative neural networks that respect physical symmetries to design material microstructures with targeted properties, orders of magnitude faster than conventional optimization. - Detection of the Progression of Anthesis in Field-Grown Maize Tassels: A Case StudyPlant Phenomics, 2021
Summary
We detect the progression of anthesis in field maize tassels from images, capturing the timing of pollen shed that matters for breeding and reproduction studies. - Polarized X-ray scattering measures molecular orientation in polymer-grafted nanoparticlesNature Communications, 2021
Summary
We built a fast computational twin of a powerful X-ray characterization tool, so that microstructural details that were previously impossible to recover can be reconstructed in near real time. - How useful is active learning for image‐based plant phenotyping?The Plant Phenome Journal, 2021
Summary
We evaluate whether active learning reduces the labeling burden for image-based plant phenotyping, identifying where it helps and where it does not. - Deep Multiview Image Fusion for Soybean Yield Estimation in Breeding ApplicationsPlant Phenomics, 2021
Summary
We fuse multiple drone-image views with deep learning to estimate soybean seed yield in breeding trials, giving a faster alternative to manual harvest measurement. - Following the crystal growth of anthradithiophenes through atomistic molecular dynamics simulations and graph characterizationMolecular Systems Design & Engineering, 2021
Summary
We combine atomistic molecular dynamics with graph analysis to follow how organic semiconductor crystals grow from the melt, linking processing to structure. - Case study of SARS-CoV-2 transmission risk assessment in indoor environments using cloud computing resourcesIn , 2021
Summary
We assess indoor SARS-CoV-2 transmission risk using accessible high-fidelity flow simulation, showing how airflow shapes exposure in occupied spaces. - Industrial scale Large Eddy Simulations with adaptive octree meshes using immersogeometric analysisComputers & Mathematics with Applications, 2021
- Scalable adaptive PDE solvers in arbitrary domainsIn Zenodo (CERN European Organization for Nuclear Research), 2021
Summary
We build scalable, highly adaptive finite element solvers for PDEs in and around arbitrary geometries, enabling accurate simulation on complex domains at large scale. -
- Crop yield prediction integrating genotype and weather variables using deep learningPLoS ONE, 2021
Summary
We predict crop yield by integrating genotype and weather data with recurrent neural networks, capturing how varieties respond across climates for breeding and monitoring. -
- Iowa Urban FEWS: Integrating Social and Biophysical Models for Exploration of Urban Food, Energy, and Water SystemsFrontiers in Big Data, 2021
Summary
We integrate social and biophysical models into an Iowa urban food-energy-water framework, letting stakeholders explore how cities can become more sustainable. - Computational study of natural ventilation in a sustainable building with complex geometrySustainable Energy Technologies and Assessments, 2021
- Identification and utilization of genetic determinants of trait measurement errors in image-based, high-throughput phenotypingThe Plant Cell, 2021
Summary
We show that errors in automated trait measurement are themselves heritable and map their genetic basis, improving the reliability of high-throughput phenotyping for genetic studies.
2020
- Deep Generative Models that Solve PDEs: Distributed Computing for\nTraining Large Data-Free ModelsarXiv (Cornell University), 2020Preprint
Summary
We train deep generative neural networks that solve partial differential equations, using distributed computing to scale training to large scientific problems. - Nanoscale control of internal inhomogeneity enhances water transport in desalination membranesScience, 2020
Summary
Using computational modeling, we showed that nanoscale unevenness in reverse osmosis membranes, long thought to be a defect, actually improves water transport. The finding changed how the field links membrane structure to desalination performance. - Computer vision and machine learning enabled soybean root phenotyping pipelinePlant Methods, 2020
Summary
We build a computer vision and machine learning pipeline that measures soybean root architecture traits from images, replacing slow and variable manual scoring. - Soybean Root System Architecture Trait Study through Genotypic, Phenotypic, and Shape-Based ClustersPlant Phenomics, 2020
Summary
We study root system architecture across a large soybean accession set, linking root trait diversity to genotype and phenotype for breeding. - Flow sculpting enabled anaerobic digester for energy recovery from low-solid content wasteRenewable Energy, 2020
- Predicting county-scale maize yields with publicly available dataScientific Reports, 2020
Summary
We predict maize yields at county scale from publicly available weather and soil data using machine learning, supporting large-area monitoring and decisions. - Thinner biological tissues induce leaflet flutter in aortic heart valve replacementsProceedings of the National Academy of Sciences, 2020
Summary
We simulate how thinner tissue in aortic heart valves induces leaflet flutter, connecting valve mechanics to durability and the design of replacements. - InvNet: Encoding Geometric and Statistical Invariances in Deep Generative ModelsIn Proceedings of the AAAI Conference on Artificial Intelligence, 2020
Summary
We build InvNet, a generative model that encodes geometric and statistical invariances so deep generative methods can respect the constraints of scientific design problems. - Simulating two-phase flows with thermodynamically consistent energy stable Cahn-Hilliard Navier-Stokes equations on parallel adaptive octree based meshesJournal of Computational Physics, 2020
- Concentration Enrichment, Separation, and Cation Exchange in Nanoliter-Scale Water-in-Oil DropletsJournal of the American Chemical Society, 2020
Summary
We achieve concentration enrichment, separation, and cation exchange inside nanoliter droplets, expanding what droplet chemistry can do with tiny fluid volumes. - Inertial focusing in triangular microchannels with various apex anglesBiomicrofluidics, 2020
Summary
We study inertial focusing of particles in triangular microchannels, showing how the apex angle controls the number and location of focusing positions. - Modeling electrochemical systems with weakly imposed Dirichlet boundary conditionsarXiv (Cornell University), 2020Preprint
Summary
We model electrochemical systems with weakly imposed Dirichlet boundary conditions in finite elements, improving the analysis and design of electrochemical devices. - Determination of the Free Energies of Mixing of Organic Solutions through a Combined Molecular Dynamics and Bayesian Statistics ApproachJournal of Chemical Information and Modeling, 2020
Summary
We determine the free energies of mixing of organic solutions from molecular simulation, providing data needed to design printing formulations for solution-processed semiconductors. - Automated trichome counting in soybean using advanced image‐processing techniquesApplications in Plant Sciences, 2020
Summary
We automate trichome counting in soybean with image processing, quantifying these protective leaf hairs that influence resistance to herbivores. - Usefulness of interpretability methods to explain deep learning based plant stress phenotypingarXiv (Cornell University), 2020Preprint
Summary
We test whether interpretability methods truly explain deep learning models for plant stress, assessing how much to trust their explanations. - Quantifying the effects of noise on early states of spinodal decomposition:In Elsevier eBooks, 2020
- Leaf Angle eXtractor: A high‐throughput image processing framework for leaf angle measurements in maize and sorghumApplications in Plant Sciences, 2020
Summary
We build Leaf Angle eXtractor, a high-throughput image pipeline that measures maize leaf angle, a trait central to planting density and light capture. - Challenges and Opportunities in Machine-Augmented Plant Stress PhenotypingTrends in Plant Science, 2020
- Women in Mechanical Engineering: A Departmental Effort to Improve Recruitment, Retention, and Engagement of Women StudentsIn , 2020
- Equilibrium microstructures of diblock copolymers under 3D confinementComputational Materials Science, 2020
- Computational investigation of the impact of glass flexure on thermal efficiency of double-pane windowsJournal of Thermal Analysis and Calorimetry, 2020
- An octree-based immersogeometric approach for modeling inertial migration of particles in channelsComputers & Fluids, 2020
2019
- Defining Cell Cluster Size by Dielectrophoretic Capture at an Array of Wireless Electrodes of Several Distinct LengthsMicromachines, 2019
Summary
We use dielectrophoretic capture at a microelectrode array to define the size of biological cell clusters, a step toward studying cluster behavior in insulin release and cancer spread. -
- Enabling Resilient Food-Energy-Water Systems Using Scientific Computing and Data AnalysisIn , 2019
- A Weakly Supervised Deep Learning Framework for Sorghum Head Detection and CountingPlant Phenomics, 2019
Summary
We build a weakly supervised deep learning framework that detects and counts sorghum crop-heads from images, easing a yield-related measurement that is tedious by hand. - Time Lapse Photography for High-Throughput Phenotyping of CornIowa State University Research and Demonstration Farms Progress Reports, 2019
Summary
We use time-lapse photography for high-throughput phenotyping of corn, capturing plant performance across the season without repeated manual field measurements. - Solving PDEs in space-timeIn , 2019
Summary
We introduced a four-dimensional, tree-based method that solves time-dependent equations across space and time at once, without storing large matrices, advancing high-performance computing for engineering simulation. - In silico design of crop ideotypes under a wide range of water availabilityFood and Energy Security, 2019
Summary
We design crop ideotypes in silico across a wide range of water availability, searching for plant traits that stay optimal and sustainable under a changing climate. - High throughput, automated prediction of focusing patterns for inertial microfluidicsarXiv (Cornell University), 2019Preprint
Summary
We automatically predict particle focusing patterns in inertial microfluidic flows at high throughput, replacing error-prone visual inspection of stable focusing positions. - Shape-design for stabilizing micro-particles in inertial microfluidic flowsIowa State University Digital Repository (Iowa State University), 2019
Summary
We numerically design microparticle shapes that stabilize at the channel centerline in inertial flows, guiding the engineering of particles for microfluidic sorting. - A Case Study of Deep Reinforcement Learning for Engineering Design: Application to Microfluidic Devices for Flow SculptingJournal of Mechanical Design, 2019
Summary
We apply deep reinforcement learning to engineering design, showing how a learning agent can efficiently explore large design spaces. - Hydrogel-based transparent soils for root phenotyping in vivoProceedings of the National Academy of Sciences, 2019
Summary
We develop transparent hydrogel-based soils that let roots be imaged in vivo, making root phenotypes easier to measure than in opaque soil or sand. - Optimization Framework for Patient-Specific Cardiac ModelingCardiovascular Engineering and Technology, 2019
- Semi-Automated Feature Extraction from RGB Images for Sorghum Panicle ArchitectureIowa State University Research and Demonstration Farms Progress Reports, 2019
Summary
We extract sorghum panicle features semi-automatically from RGB images, quantifying inflorescence architecture that influences cereal yield. - Plant disease identification using explainable 3D deep learning on hyperspectral imagesPlant Methods, 2019
Summary
We identify plant diseases with an explainable 3D deep learning model on hyperspectral images, improving accuracy while revealing which spectral bands drive the prediction. -
- Development of Optimized Phenomic Predictors for Efficient Plant Breeding Decisions Using Phenomic-Assisted Selection in SoybeanPlant Phenomics, 2019
Summary
We develop optimized phenomic predictors that make phenomic-assisted plant breeding more efficient, narrowing the gap with genomic selection. - Machine Learning Approach for Prescriptive Plant BreedingScientific Reports, 2019
Summary
We fuse high-dimensional phenotypic data with machine learning to enable prescriptive plant breeding, supporting in-season yield prediction and selection. -
- GRATE: A framework and software for GRaph based Analysis of Transmission Electron Microscopy images of polymer filmsComputational Materials Science, 2019
- Interpretable deep learning for guided microstructure-property explorations in photovoltaicsnpj Computational Materials, 2019
Summary
We use interpretable deep learning to link the microstructure of organic solar cell films to their photovoltaic performance, guiding the search for better morphologies. - Encoding Invariances in Deep Generative ModelsarXiv (Cornell University), 2019Preprint
Summary
We encode geometric and statistical invariances into generative adversarial networks so they can be trained reliably from limited scientific data. -
-
-
- PIRM2018 Challenge on Spectral Image Super-Resolution: Methods and ResultsIn Lecture notes in computer science, 2019
- FlowSculpt: software for efficient design of inertial flow sculpting devicesLab on a Chip, 2019
Summary
We release FlowSculpt, software that efficiently designs inertial flow sculpting devices by predicting how sequences of pillars shape a microfluidic stream. - Soybean Root PhenomicsIowa State University Research and Demonstration Farms Progress Reports, 2019
Summary
We study soybean root phenomics, characterizing below-ground root traits and root-soil interactions that traditional above-ground selection overlooks. - Utilization of Reduced Haploid Vigor for Phenomic Discrimination of Haploid and Diploid Maize SeedlingsThe Plant Phenome Journal, 2019
Summary
We use the reduced vigor of haploid embryos to discriminate them phenotypically, offering a non-transgenic way to select haploids in doubled-haploid breeding. -
- A residual-based variational multiscale method with weak imposition of boundary conditions for buoyancy-driven flowsComputer Methods in Applied Mechanics and Engineering, 2019
-
2018
- A deep learning framework to discern and count microscopic nematode eggsScientific Reports, 2018
Summary
We build a deep learning framework that identifies and counts microscopic nematode eggs in images, automating detection of a destructive crop pest. - NTIRE 2018 Challenge on Spectral Reconstruction from RGB ImagesIn , 2018
Summary
We review the first NTIRE challenge on reconstructing full hyperspectral images from ordinary RGB photos, benchmarking methods that recover spectral detail from three channels. - Microstructure design using graphsnpj Computational Materials, 2018
Summary
We introduced a way to represent complex 3D material microstructures as graphs, turning tangled morphologies into computable objects. This made it practical to characterize phase separation and transport pathways in organic solar cells and polymer blends. - Parallel-In-Space-Time, Adaptive Finite Element Framework for Nonlinear Parabolic EquationsSIAM Journal on Scientific Computing, 2018
Summary
We build a parallel-in-space-and-time adaptive finite element framework for nonlinear time-dependent PDEs, tailored for massively parallel computation. - A Novel Multirobot System for Plant PhenotypingRobotics, 2018
Summary
We design and demonstrate a multi-robot system for distributed plant phenotyping, cutting the labor and cost of collecting large field datasets. -
- A linearised model for calculating inertial forces on a particle in the presence of a permeate flowJournal of Fluid Mechanics, 2018
Summary
We derive a linearised model for the inertial forces on a particle in a porous channel at moderate Reynolds number, informing particle transport in filtration and biology. - An explainable deep machine vision framework for plant stress phenotypingProceedings of the National Academy of Sciences, 2018
Summary
We showed that deep learning for classifying plant stress can be made interpretable, so scientists can trust and check its decisions. This helped launch the field of AI-augmented plant science. - Predicting County Level Corn Yields Using Deep Long Short Term Memory ModelsarXiv (Cornell University), 2018Preprint
Summary
We predict county-level corn yields with deep long short-term memory networks, giving useful pre-harvest forecasts from public data. - Measurements of maize root plasticity under water stress in hydroponic chambersbioRxiv (Cold Spring Harbor Laboratory), 2018Preprint
Summary
We measure how maize roots adjust their depth, diameter, and density under water stress in hydroponics, quantifying root plasticity relevant to drought. - Simulating two-phase flows using a thermodynamically consistent coupled Cahn-Hilliard Navier-Stokes frameworkBulletin of the American Physical Society, 2018
- Flow Shape Design for Microfluidic Devices Using Deep Reinforcement LearningarXiv (Cornell University), 2018Preprint
Summary
We design microfluidic flow shapes using deep reinforcement learning, letting an agent discover pillar sequences that sculpt flow for diagnostics and other uses. - Hyperspectral band selection using genetic algorithm and support vector machines for early identification of charcoal rot disease in soybean stemsPlant Methods, 2018
Summary
We select the most informative hyperspectral bands with a genetic algorithm and support vector machine to detect charcoal rot in soybean, simplifying disease sensing. - Process optimization for microstructure-dependent properties in thin film organic electronicsMaterials Discovery, 2018
- Big Data and Parkinson’s Disease: Exploration, Analyses, and Data Challenges.In Proceedings of the ... Annual Hawaii International Conference on System Sciences/Proceedings of the Annual Hawaii International Conference on System Sciences, 2018
Summary
We apply big data analytics to clinical and laboratory data for Parkinson’s disease, exploring how large-scale analysis can support earlier detection. -
-
- An automated tassel detection and trait extraction pipeline to support high-throughput field imaging of maizeIn , 2018
Summary
We build an automated pipeline that detects maize tassels and extracts their traits from field images, supporting plant breeding and precision agriculture. - HOMEs for plants and microbes – a phenotyping approach with quantitative control of signaling between organisms and their individual environmentsLab on a Chip, 2018
-
- Deep Learning for Plant Stress Phenotyping: Trends and Future PerspectivesTrends in Plant Science, 2018
- Physics-aware Deep Generative Models for Creating Synthetic MicrostructuresarXiv (Cornell University), 2018Preprint
- uFlow: software for rational engineering of secondary flows in inertial microfluidic devicesMicrofluidics and Nanofluidics, 2018
- Shaped 3D microcarriers for adherent cell culture and analysisMicrosystems & Nanoengineering, 2018
- Crowdsourcing image analysis for plant phenomics to generate ground truth data for machine learningPLoS Computational Biology, 2018
- Semiautomated Feature Extraction from RGB Images for Sorghum Panicle Architecture GWASPLANT PHYSIOLOGY, 2018
2017
- Optimizing isotope substitution in graphene for thermal conductivity minimization by genetic algorithm driven molecular simulationsApplied Physics Letters, 2017
Summary
We combine a genetic algorithm with molecular dynamics to design isotope-engineered graphene, systematically tuning its structure to control thermal conductivity. - A data-driven identification of morphological features influencing the fill factor and efficiency of organic photovoltaic devicesComputational Materials Science, 2017
- Interpretable Deep Learning applied to Plant Stress PhenotypingarXiv (Cornell University), 2017Preprint
Summary
We apply interpretable deep learning to plant stress phenotyping, building a model that identifies specific stresses while showing which image features drive its decisions. - Incorporating a stochastic data‐driven inflow model for uncertainty quantification of wind turbine performanceWind Energy, 2017
Summary
We add a stochastic, data-driven inflow model to wind turbine analysis, capturing the natural variability of wind for more realistic design under uncertainty. - Morphological consequences of ligand exchange in quantum dot - Polymer solar cellsOrganic Electronics, 2017
- Deploying Fourier Coefficients to Unravel Soybean Canopy DiversityFrontiers in Plant Science, 2017
Summary
We use Fourier coefficients to quantify soybean canopy shape, capturing diversity in light interception and canopy closure that affects crop growth and yield. - Integrating optimization with thermodynamics and plant physiology for crop ideotype designarXiv (Cornell University), 2017Preprint
Summary
We integrate optimization, parallel computing, and plant physiology into a framework for designing crop ideotypes, exploring which trait combinations perform best. - Thermal performance analysis of residential attics containing high performance aerogel-based radiant barriersEnergy and Buildings, 2017
- A farm-level precision land management framework based on integer programmingPLoS ONE, 2017
Summary
We build a farm-level precision land management framework that integrates seed selection and irrigation decisions to guide farmland planning. -
- Deep Learning for Engineering Big Data AnalyticsIn Big Data Analytics, 2017
Summary
We recast engineering problems as inverse and design problems and use deep learning to solve them, learning the conditions that produce desired outcomes. - A real-time phenotyping framework using machine learning for plant stress severity rating in soybeanPlant Methods, 2017
Summary
We build a real-time machine learning phenotyping framework that collects plant traits accurately and quickly, accelerating the pace of plant research. - Thermally induced texture flip in semiconducting polymer stabilized by epitaxial relationship.Bulletin of the American Physical Society, 2017
- An optimization approach to identify processing pathways for achieving tailored thin film morphologiesComputational Materials Science, 2017
- Nanoscale Morphology of Doctor Bladed versus Spin‐Coated Organic Photovoltaic FilmsAdvanced Energy Materials, 2017
Summary
We compare the nanoscale morphology of doctor-bladed and spin-coated organic photovoltaic films, linking processing method to the structure that governs efficiency. - Autonomous Mobile Sensing Platform for Spatio-Temporal Plant PhenotypingIn , 2017
Summary
We design an autonomous ground-based mobile platform that collects multi-modal data for spatio-temporal plant phenotyping in agricultural research fields. - Deep Learning for Flow Sculpting: Insights into Efficient Learning using Scientific Simulation DataScientific Reports, 2017
Summary
We use deep learning for flow sculpting, learning efficient representations that speed the design of pillar sequences that shape microfluidic flow. - Computer vision and machine learning for robust phenotyping in genome-wide studiesScientific Reports, 2017
Summary
We use computer vision and machine learning for robust plant stress phenotyping, replacing slow manual evaluation and enabling genetic studies of stress response.
2016
- A finite element approach to self-consistent field theory calculations of multiblock polymersJournal of Computational Physics, 2016
- Morphology of diblock copolymers under confinementBulletin of the American Physical Society, 2016
- A Bayesian Network approach to County-Level Corn Yield Prediction using historical data and expert knowledgearXiv (Cornell University), 2016Preprint
Summary
We predict county-level corn yields before harvest with a Bayesian network, giving probabilistic forecasts useful for production and market decisions. - Microstructure taxonomy based on spatial correlations: Application to microstructure coarseningActa Materialia, 2016
- Contaminant transport at large Courant numbers using Markov matricesBuilding and Environment, 2016
- Development and verification of the Fraunhofer attic thermal modelJournal of Building Performance Simulation, 2016
Summary
We develop and verify a thermal model of attic spaces, one of the most dynamic parts of the building envelope, to help evaluate attic designs and materials. - Exploring future climate trends on the thermal performance of attics: Part 1 – Standard roofsEnergy and Buildings, 2016
- A methodology for optimal placement of sensors in enclosed environments: A dynamical systems approachBuilding and Environment, 2016
- Quantifying mechanical ventilation performance: The connection between transport equations and Markov matricesBuilding and Environment, 2016
- Tuning domain size and crystallinity in isoindigo/PCBM organic solar cells via solution shearingOrganic Electronics, 2016
- Constructing low-dimensional stochastic wind models through hierarchical spatial temporal decompositionarXiv (Cornell University), 2016Preprint
Summary
We construct low-dimensional stochastic wind models from high-fidelity data, reproducing wind variability that standard turbine simulations miss. - A framework for parametric design optimization using isogeometric analysisComputer Methods in Applied Mechanics and Engineering, 2016
- Utilizing morphological correlators for device performance to optimize ternary blend organic solar cells based on block copolymer additivesSolar Energy Materials and Solar Cells, 2016
- Research Needs and Challenges in the FEW System: Coupling Economic Models with Agronomic, Hydrologic, and Bioenergy Models for Sustainable Food, Energy, and Water SystemsIowa State University Digital Repository (Iowa State University), 2016
Summary
We report research needs for coupled food-energy-water systems, identifying where economic and biophysical models must be linked to address sustainability. - A Quasi-Dynamic Approach to modelling Hydrodynamic FocusingBulletin of the American Physical Society, 2016
- Deep Action Sequence Learning for Causal Shape TransformationarXiv (Cornell University), 2016Preprint
Summary
We use deep recurrent networks to learn action sequences that cause a target shape transformation, applying sequence learning to design problems like flow sculpting. - An Integrated Experimental-Computational Investigation of Connected Spaces as Natural Ventilation TypologiesIn Iowa State University Digital Repository (Iowa State University), 2016
Summary
We combine experiments and computation to study how building shape affects passive cooling by natural ventilation, comparing a conical-roofed house design. - Morphology control in polymer blend fibers—a high throughput computing approachModelling and Simulation in Materials Science and Engineering, 2016
Summary
We model and control the morphology of polymer blend fibers with a high-throughput computational approach, guiding the manufacture of functional fibers. - Automated Design for Microfluid Flow Sculpting: Multiresolution Approaches, Efficient Encoding, and CUDA ImplementationJournal of Fluids Engineering, 2016
Summary
We automate the design of microfluidic flow sculpting devices with a multiresolution search that finds pillar sequences producing target flow shapes. - Publisher’s Note: “Optimization of micropillar sequences for fluid flow sculpting” [Phys. Fluids 28, 012003 (2016)]Physics of Fluids, 2016
- A Diffuse Interface Model for Incompressible Two-Phase Flow with Large Density RatiosIn Modeling and simulation in science, engineering & technology, 2016
- Incompressible two-phase flow: Diffuse interface approach for large density ratios, grid resolution study, and 3D patterned substrate wetting problemComputers & Fluids, 2016
- Vertical Phase Separation in Small Molecule:Polymer Blend Organic Thin Film Transistors Can Be Dynamically ControlledAdvanced Functional Materials, 2016
Summary
We study vertical phase separation in small-molecule and polymer blends, showing how stratification governs the performance of organic thin-film transistors.
2015
- A Novel Framework for Visual Detection and Exploration of Performance Bottlenecks in Organic Photovoltaic Solar Cell MaterialsComputer Graphics Forum, 2015
Summary
We build a visualization framework that detects and explores performance-related features in organic photovoltaic bulk heterojunction morphologies. - Parallel framework for wormlike chains using self consistent field theoryBulletin of the American Physical Society, 2015
- High Fidelity CFD Modeling of Natural Ventilation in a Solar HouseIn , 2015
Summary
We use high-fidelity CFD to model natural ventilation in a solar house, quantifying how airflow supports thermal comfort and air quality in sustainable buildings. - Constructing Markov matrices for real-time transient contaminant transport analysis for indoor environmentsBuilding and Environment, 2015
- Utilizing Wide Band Gap, High Dielectric Constant Nanoparticles as Additives in Organic Solar CellsThe Journal of Physical Chemistry C, 2015
Summary
We show, in experiment and theory, that high-dielectric BaTiO3 nanoparticle additives change charge behavior in polythiophene and fullerene solar cells. - Achieving Bicontinuous Microemulsion Like Morphologies in Organic PhotovoltaicsACS Macro Letters, 2015
Summary
We create bicontinuous, microemulsion-like morphologies in organic solar cells, forming the interpenetrating donor and acceptor domains thought to be optimal. - Polymer/solvent bicontinuous microemulsions for use as organic solar cell active layersBulletin of the American Physical Society, 2015
- Hierarchical Feature Extraction for Efficient Design of Microfluidic Flow PatternsIn Iowa State University Digital Repository (Iowa State University), 2015
Summary
We use deep neural networks to learn hierarchical features for the efficient design of microfluidic flow sculpting devices. - Genome-wide association analysis of seedling root development in maize (Zea mays L.)BMC Genomics, 2015
Summary
We map the genetics of maize seedling root development through a genome-wide association study, linking root traits to nutrient uptake and productivity. - Quantifying the Effects of Noise on Diffuse Interface Models: Cahn-Hilliard-Cook equationsBulletin of the American Physical Society, 2015
- Tailoring the morphology of polymer blend particles: 3D simulations and linear stability analysisBulletin of the American Physical Society, 2015
- Parallel Framework for Dimensionality Reduction of Large-Scale DatasetsScientific Programming, 2015
Summary
We build a parallel framework for dimensionality reduction of large-scale data, preserving key properties while cutting computational cost. - Machine Learning for High-Throughput Stress Phenotyping in PlantsTrends in Plant Science, 2015
Summary
This widely cited framework set out how machine learning could be woven into high-throughput plant phenotyping, shaping the research agenda for AI in the plant and agricultural sciences. - Micropillar sequence design for inertial fluid flow sculptingBulletin of the American Physical Society, 2015
- Optimization of micropillar sequences for fluid flow sculptingarXiv (Cornell University), 2015Preprint
Summary
We optimize sequences of micropillars for flow sculpting, searching the rich design space to produce desired inertial flow deformations. - Automated, high throughput exploration of process–structure–property relationships using the MapReduce paradigmMaterials Discovery, 2015
- Using graphs to interrogate the atomic structure of polymer blendsBulletin of the American Physical Society, 2015
- Morphology optimization for enhanced performance in organic photovoltaicsBulletin of the American Physical Society, 2015
2014
- A massively parallel space-time formulation for SCFTBulletin of the American Physical Society, 2014
- Federated Computing for the Masses–Aggregating Resources to Tackle Large-Scale Engineering ProblemsComputing in Science & Engineering, 2014
Summary
We aggregate distributed computing resources through federation, giving ordinary users access to capacity beyond a single center for demanding problems. - High-Resolution Performance Analysis of a Large Building with Linear Dispersion Ductwork SystemIn Iowa State University Digital Repository (Iowa State University), 2014
Summary
We perform high-resolution performance analysis of a large building’s thermal loads, showing how better thermal management can cut its substantial energy use. - Revolutionizing science through simulation: A junior researcher’s perspective on research challenges in uncertain timesMerrill Series on The Research Mission of Public Universities, 2014
- Electrode Materials, Thermal Annealing Sequences, and Lateral/Vertical Phase Separation of Polymer Solar Cells from Multiscale Molecular SimulationsACS Applied Materials & Interfaces, 2014
Summary
We show how electrode materials and thermal annealing sequences control the sensitive bulk-heterojunction morphology of polymer solar cells. - Analysis of Maize (Zea mays L.) Seedling Roots with the High-Throughput Image Analysis Tool ARIA (Automatic Root Image Analysis)PLoS ONE, 2014
Summary
We analyze maize seedling roots with a high-throughput platform, quantifying the genetic and phenotypic variation in root architecture that affects establishment. - A software framework for data dimensionality reduction: application to chemical crystallographyIntegrating materials and manufacturing innovation, 2014
Summary
We release a software framework for data dimensionality reduction applied to materials science, helping establish process-structure-property relationships. - Micropillar sequence designs for fundamental inertial flow transformationsLab on a Chip, 2014
Summary
We design micropillar sequences that produce fundamental inertial flow transformations, enabling passive flow control for reactions, separation, and materials. - Graph-based surrogate models for quantifying performance in organic solar cellsAPS, 2014
- How do evaporating thin films evolve? Unravelling phase-separation mechanisms during solvent-based fabrication of polymer blendsApplied Physics Letters, 2014
Summary
We reveal how evaporating thin films phase-separate as they dry, connecting solvent-based processing to the morphology of the resulting polymer film.
2013
- Engineering fluid flow using sequenced microstructuresNature Communications, 2013
Summary
We showed that carefully sequenced microstructures can sculpt fluid flows in programmable ways, a new route to engineering flows in microfluidic devices for sorting, mixing, and diagnostics. - Application of Computational Homology and Graph-Theoretic Approaches for Quantitative Chemical Imaging in Atom Probe TomographyMicroscopy and Microanalysis, 2013
- Exploring the Use of Elastic Resource Federations for Enabling Large-Scale Scientific Workflows2013
Summary
We use elastic federations of computing resources to run uncertainty quantification, design optimization, and parametric studies that map onto many machines. - A stochastic approach to modeling the dynamics of natural ventilation systemsEnergy and Buildings, 2013
- Sensitivity analysis of current generation in organic solar cells—comparing bilayer, sawtooth, and bulk heterojunction morphologiesSolar Energy Materials and Solar Cells, 2013
-
- A graph-theoretic approach for characterization of precipitates from atom probe tomography dataComputational Materials Science, 2013
- Quantifying organic solar cell morphology: a computational study of three-dimensional mapsEnergy & Environmental Science, 2013
Summary
We built the computational pipeline that links three-dimensional nanoscale morphology to the performance of organic solar cells, an early foundation for the group’s materials work.
2012
- The use of sequences of pillars to engineer fluid cross-sectional shape \textitvia inertial flow deformationsBulletin of the American Physical Society, 2012
- Nanoscale surface roughness affects low Reynolds number flow: Experiments and modelingApplied Physics Letters, 2012
Summary
We show that nanoscale surface roughness alters low Reynolds number flow, and that this stochastic roughness can be harnessed to direct microfluidic transport. - A computational framework to investigate charge transport in heterogeneous organic photovoltaic devicesComputer Methods in Applied Mechanics and Engineering, 2012
- Computer simulation of heterogeneous polymer photovoltaic devicesModelling and Simulation in Materials Science and Engineering, 2012
Summary
We simulate heterogeneous polymer photovoltaic devices computationally, linking the blended active-layer structure to device efficiency. - Enhanced charge separation in organic photovoltaic films doped with ferroelectric dipolesEnergy & Environmental Science, 2012
Summary
We show that doping organic photovoltaic films enhances the separation of photogenerated electron-hole pairs, a key step for efficient solar cells. - WiME: a departmental effort to improve recruitment, retention and engagement of women students in Mechanical EngineeringIn Iowa State University Digital Repository (Iowa State University), 2012
- Computational characterization of bulk heterojunction nanomorphologyJournal of Applied Physics, 2012
Summary
We computationally characterize the bulk-heterojunction nanomorphology of organic solar cells, quantifying the structure that controls device efficiency. - A graph-based formulation for computational characterization of bulk heterojunction morphologyOrganic Electronics, 2012
- Modeling morphology evolution during solvent-based fabrication of organic solar cellsComputational Materials Science, 2012
- Cantilever deflection associated with hybridization of monomolecular DNA filmJournal of Applied Physics, 2012
Summary
We model the deflection of a microcantilever caused by molecular binding such as DNA hybridization, informing the design of label-free biosensors.
2011
- A near real-time framework for extracting tip-sample forces in dynamic atomic force microscopy (dAFM)arXiv (Cornell University), 2011Preprint
Summary
We build a near real-time framework that extracts tip-sample forces in atomic force microscopy, supporting high-speed nanoscale imaging. - Thermal comparison between ceiling diffusers and fabric ductwork diffusers for green buildingsEnergy and Buildings, 2011
- Exploring the Effect of Stick-Slip Friction Transition Across Tape-Roller Interface on the Transmission of Lateral VibrationIEEE Transactions on Magnetics, 2011
- Dependence of recombination mechanisms and strength on processing conditions in polymer solar cellsApplied Physics Letters, 2011
Summary
We study how charge recombination mechanisms and strength in polymer solar cells depend on processing conditions, informing more efficient device fabrication. - Experimental analysis of the surface roughness evolution of etched glass for micro/nanofluidic devicesJournal of Micromechanics and Microengineering, 2011
Summary
We experimentally analyze how the surface roughness of etched microchannels evolves, since roughness shapes fluid behavior in micro and nanoscale devices. - Computationally efficient solution to the Cahn–Hilliard equation: Adaptive implicit time schemes, mesh sensitivity analysis and the 3D isoperimetric problemJournal of Computational Physics, 2011
- A novel graph-based formulation for characterizing morphology with application to organic solar cellsarXiv (Cornell University), 2011Preprint
Summary
We introduce a graph-based formulation to characterize the morphology of organic solar cells, turning complex nanostructure into quantitative descriptors. - Modeling morphology evolution during solvent-based fabrication of\norganic solar cellsarXiv (Cornell University), 2011Preprint
Summary
We model how morphology evolves during solvent-based fabrication of organic films, tracking phase separation as the solvent evaporates.
2010
- Stability in the almost everywhere sense: A linear transfer operator approachJournal of Mathematical Analysis and Applications, 2010
- Experimental Data Analysis of the Vortex Structures in the Wakes of Flapping WingsIn 28th AIAA Applied Aerodynamics Conference, 2010
2009
- Using data to account for lack of data: Linking material informatics with stochastic analysisJOM, 2009
- Using stabilized finite elements for understanding the performance of organic solar cellsBulletin of the American Physical Society, 2009
- Data Mining and Informatics for Quantitative Atom Probe TomographyMicroscopy and Microanalysis, 2009
- Transfer operator method for control in fluid flowsIn , 2009
Summary
We present a transfer operator method, drawn from ergodic theory, for the optimal control of complex fluid flows.
2008
- A non-linear dimension reduction methodology for generating data-driven stochastic input modelsJournal of Computational Physics, 2008
- A seamless approach towards stochastic modeling: Sparse grid collocation and data driven input modelsFinite Elements in Analysis and Design, 2008
- A stochastic multiscale framework for modeling flow through random heterogeneous porous mediaJournal of Computational Physics, 2008
- A scalable framework for the solution of stochastic inverse problems using a sparse grid collocation approachJournal of Computational Physics, 2008
2007
- Modeling diffusion in random heterogeneous media: Data-driven models, stochastic collocation and the variational multiscale methodJournal of Computational Physics, 2007
- Sparse grid collocation schemes for stochastic natural convection problemsJournal of Computational Physics, 2007
2006
- Modelling dendritic solidification with melt convection using the extended finite element methodJournal of Computational Physics, 2006
- Sparse grid collocation schemes for stochastic convection problemsBulletin of the American Physical Society, 2006
2005
- On the control of solidification using magnetic fields and magnetic field gradientsInternational Journal of Heat and Mass Transfer, 2005
2004
- Control of solidification of non-conducting materials using tailored magnetic fieldsJournal of Crystal Growth, 2004
- Using magnetic field gradients to control the directional solidification of alloys and the growth of single crystalsJournal of Crystal Growth, 2004
- Melt flow control using magnetic fields and magnetic field gradients2004