About

Lori Graham-Brady is a professor of civil and systems engineering at Johns Hopkins University and serves as the vice dean for faculty in the Whiting School of Engineering. Her research focuses on stochastic finite element methods, probabilistic mechanics, stochastic simulation of material properties, and micromechanics. She is the current director of the JHU Center on AI for Materials in Extreme Environments (CAIMEE) and has previously directed the Center for Materials in Extreme Dynamic Environments, a collaborative program with the Army Research Labs aimed at multiscale modeling and design of materials for extreme dynamic environments, particularly ceramics and composites for armor applications. Graham-Brady was also the founding director of the Center on High-throughput Materials for Extremes (HT-MAX) and the founding associate director of the Hopkins Extreme Materials Institute (HEMI). Her work provides a fundamental understanding of the connections between material-scale uncertainties and the performance and reliability of structures, leveraging machine learning tools to solve complex mechanics problems. She led the development of an AI-focused facility for materials design, emphasizing high-strain and temperature-rate environments. Her contributions to engineering education include a multi-year NSF-funded program on modeling complex systems and developing the scientific basis for multiscale multiphysics models, with a focus on diversity and inclusion. Graham-Brady has held leadership roles within professional societies, including the ASCE Engineering Mechanics Institute, and has received numerous awards such as the Presidential Early Career Awards for Scientists and Engineers, the Walter L. Huber Civil Engineering Research Prize, and the William H. Huggins Award for Excellence in Teaching. She holds secondary appointments in Mechanical Engineering and Materials Science and Engineering, and is a member of several professional societies.

Selected publications

• Enhanced accuracy through ensembling of randomly initialized auto-regressive models for dynamical systems

  Machine learning for computational science and engineering · 2026-02-07

  article
  PublisherDOI

• Physics-constrained Gaussian Processes for Predicting Shockwave Hugoniot Curves

  arXiv (Cornell University) · 2026-01-10

  preprintOpen access

  A physics-constrained Gaussian Process regression framework is developed for predicting shocked material states along the Hugoniot curve using data from a small number of shockwave simulations. The proposed Gaussian process employs a probabilistic Taylor series expansion in conjunction with the Rankine-Hugoniot jump conditions between the various shocked material states to construct a thermodynamically consistent covariance function. This leads to the formulation of an optimization problem over a small number of interpretable hyperparameters and enables the identification of regime transitions, from a leading elastic wave to trailing plastic and phase transformation waves. This work is motivated by the need to investigate shock-driven material response for materials discovery and for offering mechanistic insights in regimes where experimental characterizations and simulations are costly. The proposed methodology relies on large-scale molecular dynamics which are an accurate but expensive computational alternative to experiments. Under these constraints, the proposed methodology establishes Hugoniot curves from a limited number of molecular dynamics simulations. We consider silicon carbide as a representative material and atomic-level simulations are performed using a reverse ballistic approach together with appropriate interatomic potentials. The framework reproduces the Hugoniot curve with satisfactory accuracy while also quantifying the uncertainty in the predictions using the Gaussian Process posterior.

  PublisherDOI

• Physics-constrained Gaussian Processes for Predicting Shockwave Hugoniot Curves

  ArXiv.org · 2026-01-10

  articleOpen access

  A physics-constrained Gaussian Process regression framework is developed for predicting shocked material states along the Hugoniot curve using data from a small number of shockwave simulations. The proposed Gaussian process employs a probabilistic Taylor series expansion in conjunction with the Rankine-Hugoniot jump conditions between the various shocked material states to construct a thermodynamically consistent covariance function. This leads to the formulation of an optimization problem over a small number of interpretable hyperparameters and enables the identification of regime transitions, from a leading elastic wave to trailing plastic and phase transformation waves. This work is motivated by the need to investigate shock-driven material response for materials discovery and for offering mechanistic insights in regimes where experimental characterizations and simulations are costly. The proposed methodology relies on large-scale molecular dynamics which are an accurate but expensive computational alternative to experiments. Under these constraints, the proposed methodology establishes Hugoniot curves from a limited number of molecular dynamics simulations. We consider silicon carbide as a representative material and atomic-level simulations are performed using a reverse ballistic approach together with appropriate interatomic potentials. The framework reproduces the Hugoniot curve with satisfactory accuracy while also quantifying the uncertainty in the predictions using the Gaussian Process posterior.

  PublisherOA PDF

• Physics-informed latent neural operator for real-time predictions of time-dependent parametric PDEs

  Computer Methods in Applied Mechanics and Engineering · 2025-12-09 · 2 citations

  article
  PublisherDOI

• Materials laboratories of the future for alloys, amorphous, and composite materials

  MRS Bulletin · 2025-01-29 · 4 citations

  articleOpen access

  Abstract In alignment with the Materials Genome Initiative and as the product of a workshop sponsored by the US National Science Foundation, we define a vision for materials laboratories of the future in alloys, amorphous materials, and composite materials; chart a roadmap for realizing this vision; identify technical bottlenecks and barriers to access; and propose pathways to equitable and democratic access to integrated toolsets in a manner that addresses urgent societal needs, accelerates technological innovation, and enhances manufacturing competitiveness. Spanning three important materials classes, this article summarizes the areas of alignment and unifying themes, distinctive needs of different materials research communities, key science drivers that cannot be accomplished within the capabilities of current materials laboratories, and open questions that need further community input. Here, we provide a broader context for the workshop, synopsize the salient findings, outline a shared vision for democratizing access and accelerating materials discovery, highlight some case studies across the three different materials classes, and identify significant issues that need further discussion. Graphical abstract

  PublisherOA PDFDOI

• Physics-Informed Latent Neural Operator for Real-time Predictions of time-dependent parametric PDEs

  ArXiv.org · 2025-01-14

  preprintOpen access

  Deep operator network (DeepONet) has shown significant promise as surrogate models for systems governed by partial differential equations (PDEs), enabling accurate mappings between infinite-dimensional function spaces. However, when applied to systems with high-dimensional input-output mappings arising from large numbers of spatial and temporal collocation points, these models often require heavily overparameterized networks, leading to long training times. Latent DeepONet addresses some of these challenges by introducing a two-step approach: first learning a reduced latent space using a separate model, followed by operator learning within this latent space. While efficient, this method is inherently data-driven and lacks mechanisms for incorporating physical laws, limiting its robustness and generalizability in data-scarce settings. In this work, we propose PI-Latent-NO, a physics-informed latent neural operator framework that integrates governing physics directly into the learning process. Our architecture features two coupled DeepONets trained end-to-end: a Latent-DeepONet that learns a low-dimensional representation of the solution, and a Reconstruction-DeepONet that maps this latent representation back to the physical space. By embedding PDE constraints into the training via automatic differentiation, our method eliminates the need for labeled training data and ensures physics-consistent predictions. The proposed framework is both memory and compute-efficient, exhibiting near-constant scaling with problem size and demonstrating significant speedups over traditional physics-informed operator models. We validate our approach on a range of parametric PDEs, showcasing its accuracy, scalability, and suitability for real-time prediction in complex physical systems.

  PublisherOA PDFDOI

• Numerical and data-driven modeling of spall failure in polycrystalline ductile materials

  Computer Methods in Applied Mechanics and Engineering · 2025-10-21 · 2 citations

  articleSenior authorCorresponding
  PublisherDOI

• Enhanced accuracy through ensembling of randomly initialized auto-regressive models for time-dependent PDEs

  ArXiv.org · 2025-07-05

  articleOpen access

  Systems governed by partial differential equations (PDEs) require computationally intensive numerical solvers to predict spatiotemporal field evolution. While machine learning (ML) surrogates offer faster solutions, autoregressive inference with ML models suffer from error accumulation over successive predictions, limiting their long-term accuracy. We propose a deep ensemble framework to address this challenge, where multiple ML surrogate models with random weight initializations are trained in parallel and aggregated during inference. This approach leverages the diversity of model predictions to mitigate error propagation while retaining the autoregressive strategies ability to capture the system's time dependent relations. We validate the framework on three PDE-driven dynamical systems - stress evolution in heterogeneous microstructures, Gray-Scott reaction-diffusion, and planetary-scale shallow water system - demonstrating consistent reduction in error accumulation over time compared to individual models. Critically, the method requires only a few time steps as input, enabling full trajectory predictions with inference times significantly faster than numerical solvers. Our results highlight the robustness of ensemble methods in diverse physical systems and their potential as efficient and accurate alternatives to traditional solvers. The codes for this work are available on GitHub (https://github.com/Graham-Brady-Research-Group/AutoregressiveEnsemble_SpatioTemporal_Evolution).

  PublisherOA PDF

• Numerical and data-driven modeling of spall failure in polycrystalline ductile materials

  ArXiv.org · 2025-07-04

  preprintOpen accessSenior author

  Developing materials with tailored mechanical performance requires iteration over a large number of proposed designs. When considering dynamic fracture, experiments at every iteration are usually infeasible. While high-fidelity, physics-based simulations can potentially reduce experimental efforts, they remain computationally expensive. As a faster alternative, key dynamic properties can be predicted directly from microstructural images using deep-learning surrogate models. In this work, the spallation of ductile polycrystals under plate-impact loading at strain rates of O(10^6 s^-1) is considered. A physics-based numerical model that couples crystal plasticity and a cohesive zone model is used to generate data for the surrogate models. Three architectures - 3D U-Net, 3D Fourier Neural Operator (FNO-3D), and U-FNO were trained on the particle-velocity field data from the numerical model. The generalization of the models was evaluated using microstructures with varying grain sizes and aspect ratios. U-FNO and 3D U-Net performed significantly better than FNO-3D across all datasets. Furthermore, U-FNO and 3D U-Net exhibited comparable accuracy for every metric considered in this study. However, training the U-FNO requires almost twice the computational effort compared to the 3D U-Net, making it a desirable option for a surrogate model.

  PublisherOA PDFDOI

• Correction: Materials laboratories of the future for alloys, amorphous, and composite materials

  MRS Bulletin · 2025-02-28

  articleOpen access
  PublisherOA PDFDOI

Awards & honors

• Presidential Early Career Awards for Scientists and Engineer…
• International Association for Structural Safety and Reliabil…
• Walter L. Huber Civil Engineering Research Prize
• William H. Huggins Award for Excellence in Teaching
• Fellow of ASCE EMI