Search NASA⌕ Search

SEARCH · Search NASA

Results for “TRAINING PLATFORMS”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 523 records · Page 29

MO-DE-BRA-06: MrRSCAL: A Radiological Simulation Tool for Resident Education

Purpose: The goal of this project was to create a readily accessible, comprehensive-yet-flexible interactive MRI simulation tool for use in training and education of radiology residents in particular. This tool was developed to take the place of an MR scanner in laboratory activities, as magnet time has become scarce while hospitals are optimizing clinical scheduling for improved throughput. Methods: MrRSCAL (Magnetic resonance Resident Simulation Console for Active Learning) was programmed and coded using Matlab on a Mac workstation utilizing OS X platform. MR-based brain images were obtained from one of the co-authors and processed to generate parametric maps. Scanner sounds are also generated via mp3 convolution of a single MR gradient slew with a time-profile of gradient waveforms. Results: MrRSCAL facilitates the simulation of multiple MR sequences with the ability to alter MR parameters via an intuitive GUI control panel. The application allows the user to gain real-time understanding of image transformation when varying these said parameters by examining the resulting images. Lab procedures can be loaded and displayed for more directed study. The panel is also configurable, providing a simple interface for elementary labs or a full array of controls for the expert user. Conclusion: Our introduction of MrRSCAL, which is readily available to users with a current laptop or workstation, allows for individual or group study of MR image acquisition with immediate educational feedback as the MR parameters are manipulated. MrRSCAL can be used at any time and any place once installed, offering a new tool for reviewing relaxometric and artifact principles when studying for boards or investigating properties of a pulse sequence. This tool promises to be extremely useful in conveying traditionally difficult and abstract concepts involved with MR to the radiology resident and other medical professionals at large.

60 APPLIED LIFE SCIENCES↗

Secure Federated Learning Across Heterogeneous Cloud and High-Performance Computing Resources: A Case Study on Federated Fine-Tuning of LLaMA 2

Federated learning enables multiple data owners to collaboratively train robust machine learning models without transferring large or sensitive local datasets by only sharing the parameters of the locally trained models. Here, in this article, we elaborate on the design of our Advanced Privacy-Preserving Federated Learning (APPFL) framework, which streamlines end-to-end secure and reliable federated learning experiments across cloud computing facilities and high-performance computing resources by leveraging Globus Compute, a distributed function as a service platform, and Amazon Web Services. We further demonstrate the use case of APPFL in fine-tuning an LLaMA 2 7B model using several cloud resources and supercomputers.

97 MATHEMATICS AND COMPUTING↗

DRAS: Deep Reinforcement Learning for Cluster Scheduling in High Performance Computing

Cluster schedulers are crucial in high-performance computing (HPC). They determine when and which user jobs should be allocated to available system resources. Existing cluster scheduling heuristics are developed by human experts based on their experience with specific HPC systems and workloads. However, the increasing complexity of computing systems and the highly dynamic nature of application workloads have placed tremendous burden on manually designed and tuned scheduling heuristics. More aggressive optimization and automation are needed for cluster scheduling in HPC. In this work, we present an automated HPC scheduling agent named DRAS (Deep Reinforcement Agent for Scheduling) by leveraging deep reinforcement learning. DRAS is built on a hierarchical neural network incorporating special HPC scheduling features such as resource reservation and backfilling. An efficient training strategy is presented to enable DRAS to rapidly learn the target environment. Once being provided a specific scheduling objective given by the system manager, DRAS automatically learns to improve its policy through interaction with the scheduling environment and dynamically adjusts its policy as workload changes. We implement DRAS into a HPC scheduling platform called CQGym. CQGym provides a common platform allowing users to flexibly evaluate DRAS and other scheduling methods such as heuristic and optimization methods. Furthermore, the experiments using CQGym with different production workloads demonstrate that DRAS outperforms the existing heuristic and optimization approaches by up to 50%.

97 MATHEMATICS AND COMPUTING↗

Deployment of Dynamic Neural Network Optimization to Minimize Heat Rate During Ramping for Coal Power Plants (Final Technical Report)

Much success was achieved throughout the course of this project. A successful implementation of Dynamic Neural Network Optimization (D-NNO) was coupled with Adaptive Predictive Controls (APC) and a novel hardware installation comprised of an advanced sensor network (ASN) measuring mass-weighted averages of flue gas constituents above the horizontal superheater of a coal-fired utility boiler. From 2019 through 2023 (including an extension due to COVID delays), the team was able to prototype, evaluate, deploy, iterate, and ultimately finalize an advanced closed-loop control D-NNO system which demonstrated the ability to: •improve unit efficiency ~2.0% relative to unoptimized operation (represented as total fuel fired per MWh generated) •improve unit NOx emission rates 10%+ beyond static optimization baselines •improve unit temperature stability as much as 58% and on average 12% •improve operating load stability as much as 35% The culmination of this project has generated an advanced methodology of deploying specially designed recurrent neural networks (long short-term memory, gated recurrent unit, encoder-decoder networks, transformers, etc.), customized trajectory planning and closed-loop optimization modules capable of adapting to live electric grid responses and demands, self-tuning and adaptive expert controls constantly adjusting prediction parameters to real-time unit behavior, and a hardware/software package able to reliably calculate net unit heat rate (NUHR) in real-time using flue gas constituents, machine learning, and known combustion relationships. Through this real-time NUHR value, immediate feedback on system adjustments relative to operating efficiency was available, allowing for rapid improvements to system performance. In addition to development and deployment of the advanced D-NNO system, the approach methodology has been readily commercialized through the project platform Griffin Open Systems, LLC, the D-NNO software platform host. Similar methodologies to those developed by this project have already been deployed at 5 other units across the United States, with another 6 implementations scheduled, and more expected. Over the course of the project, multiple academic papers were submitted and accepted for publication within esteemed academic journals, and PhD students were trained and graduated, as well as undergraduate students becoming involved and participating to project objectives.

01 COAL, LIGNITE, AND PEAT↗

MLSPICE: Machine Learning based SPICE Modeling Platform for Power Magnetics

Electrical power converters are critical to a wide range of applications ranging from renewable integration to transportation electrification, and can be a key factor determining the size, weight, and efficiency of energy conversion systems. Magnetic components are typically the largest and least efficient components in power electronics. While there have been major strides in the modeling and analysis of power semiconductor devices and circuit simulations, the necessary advances in the design of power magnetics have lagged. In this project, we have transformed the modeling and design of power magnetics with machine learning enabled methods and catalyze simultaneous disruptive improvements for ML-based power electronics design tools. A fully automated open-source machine learning based magnetics modeling platform – the MagNet project - with innovations in full stack have been developed to greatly accelerate the design process and provide new insights to magnetic material and geometry design. The ARPA-E funded MagNet platform contains three major building blocks: 1) a ML-Integrated Data Acquisition System (MIDAS): a highly automated data acquisition testbed which is capable of measuring a large number of magnetic cores with a wide range of electrical circuit excitations; 2) a ML-integrated Core Loss Model (MICLM): a machine-learning trained modeling method for modeling the core loss and saturation effects of magnetic materials for arbitrary excitation waveforms; 3) ML-guided Magnetics SPICE Simulation Tool (PMSPICE): a fully integrated CAD tool which can simulate the magnetics in SPICE. It can help the designers to quickly model the linear and non-linear characteristics of magnetic components and evaluate their behavior in SPICE simulations. The developed MagNet system has fully demonstrated the proposed performance target and has been open sourced to the entire power electronics community to advance the modeling and design of power magnetics from many different angles.

36 MATERIALS SCIENCE↗

Training self-learning circuits for power-efficient solutions

As the size and ubiquity of artificial intelligence and computational machine learning models grow, the energy required to train and use them is rapidly becoming economically and environmentally unsustainable. Recent laboratory prototypes of self-learning electronic circuits, such as “physical learning machines,” open the door to analog hardware that directly employs physics to learn desired functions from examples at a low energy cost. In this work, we show that this hardware platform allows for an even further reduction in energy consumption by using good initial conditions and a new learning algorithm. Using analytical calculations, simulations, and experiments, we show that a trade-off emerges when learning dynamics attempt to minimize both the error and the power consumption of the solution—greater power reductions can be achieved at the cost of decreasing solution accuracy. Finally, we demonstrate a practical procedure to weigh the relative importance of error and power minimization, improving the power efficiency given a specific tolerance to error.

Stern, Menachem (ORCID:0000000302158082)↗

Application of Manufacturing Quality Management Principles to PV System Installations

To help SETO/DOE achieve its goals, the IBTS team proposed a project addressing system reliability by improving installation standards and quality management. The proposed approach was designed to help achieve measurable reductions in installation defect density and improvements in the performance of PV systems by optimizing design and installation of residential and commercial PV systems. This approach addressed the soft costs associated with installations and quality management. The project demonstrated improved system reliability and reduced PV system installation costs. The software developed improved operations, decreased risk, and increased the overall value of PV systems across their lifecycle. The project used several data collection methods, including extensive industry surveys, face-to-face high-level interviews at industry conferences, stakeholder teleconferences, and in-depth interviews conducted by IBTS staff. Results from the research found the industry needs a uniform assessment method for national providers to be more efficient; the software should support both code officials and installers; most industry stakeholders would find value in a centralized software system that allows them to collect, report, and review information on in-process and completed solar installations; and mobile solutions that bridge existing knowledge gaps with inspectors and integrate with existing methodologies (such as permitting software) are of great value. The software solution developed is web-based, allowing for national access, and is built on a Google Firebase platform that can handle significant users and data. It can be used onsite or remotely, allowing for code compliance to continue despite ongoing pandemic related delays or shutdowns for local economies. The information provided by the software tool allows users to uniformly assess a system for compliance and use that aggregated data to identify training topics or create internal process designed to improving issues and reducing occurrence. This solution has multiple benefits in managing quality at time of use and promoting an increase in future safety and quality through education. Perhaps most importantly, this software increases public safety by ensuring compliance of installed systems and allows for local AHJs to remotely engage specialized and qualified solar specific expertise for oversite of the installation in their jurisdictions. Data analysis provides the quality feedback loop identifying the root cause of failure and drives installation practices to improve through training and education, resulting in systems with higher performance, greater reliability, and reduced operations and maintenance costs. With the successful completion of this project, the industry can expect reduced soft costs and increased performance and safety and will ultimately benefit from longer performing systems that cost less to operate.

14 SOLAR ENERGY↗

An adaptable product for material processing and life science missions

The Experiment Control System II (ECS-II) is designed to make available to the microgravity research community the same tools and mode of automated experimentation that their ground-based counterparts have enjoyed for the last two decades. The design goal was accomplished by combining commercial automation tools familiar to the experimenter community with system control components that interface with the on-orbit platform in a distributed architecture. The architecture insulates the tools necessary for managing a payload. By using commercial software and hardware components whenever possible, development costs were greatly reduced when compared to traditional space development projects. Using commercial-off-the-shelf (COTS) components also improved the usability documentation, and reducing the need for training of the system by providing familiar user interfaces, providing a wealth of readily available documentation, and reducing the need for training on system-specific details. The modularity of the distributed architecture makes it very amenable for modification to different on-orbit experiments requiring robotics-based automation.

Wassick, Gregory↗

Digital Twin + AI: Control Room of the Future [Slides]

The control room functions as the central brain of the grid, essential for balancing supply and demand and ensuring moment-to-moment grid reliability. Like the human brain, which processes sensory data to make decisions, control room operators analyze operational data from power generation, transmission, and distribution to make informed decisions. Currently, decision-making primarily rests with operators due to hardware and software limitations. However, with technological advancements, Digital Twins and AI are becoming high interest points in the control room's decision-making pilot programs. NREL is developing a comprehensive decision-making platform that integrates Digital Twins, AI, and advanced visualization techniques. As this integration progresses, the role of Digital Twins will evolve from conducting automated simulations to serving as a Trustworthy AI enabler, offering verification and validation of AI-generated response for power systems or providing physics-aware synthetic data of AI pre-training.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Web-Based Tools for Data-Informed Remedy Optimization: Software Theory and User Guide

This report documents the development and application of two web-based decision-support tools for pump-and-treat (P&T) groundwater remediation systems: PTOLEMY (Pump-and-Treat Optimized Location Evaluation to Maximize Yields) and OPTIMA (Optimization for Pump-and-Treat Implementation, Management, & Assessment). These tools enhance remedy design and management by leveraging advanced computational methods – specifically deep learning and multi-objective optimization – within a user-friendly platform. By integrating data-driven models with established hydrogeological knowledge, PTOLEMY and OPTIMA enable more efficient evaluation of well placement and operational strategies, helping site managers balance multiple remediation objectives under complex conditions. Both tools are implemented as modules within the SOCRATES (Suite Of Comprehensive Rapid Analysis Tools for Environmental Sites) web platform, which provides data access, visualization, and analytics to support remedy optimization across sites in the U.S. Department of Energy Office of Environmental Management complex. PTOLEMY is a rapid screening module designed to identify promising locations for new extraction wells. It employs a multi-channel three-dimensional convolutional neural network (MC3D-CNN) trained on high-fidelity simulation data to predict the relative performance (in terms of contaminant mass recovery) of potential well sites. Through an interactive web interface, PTOLEMY visualizes the probability of high performance across a site, highlighting areas where an extraction well is likely to yield above-threshold contaminant removal over a multi-year period. PTOLEMY’s map-based displays and exportable results support transparent communication of screening analyses. By focusing attention on the most favorable candidate locations, the tool augments traditional engineering judgment and physics-based modeling, providing a data informed basis for subsequent detailed evaluations. OPTIMA is a multi objective optimization module designed to find wellfield layouts and operating schedules that meet various cleanup goals. It quickly evaluates thousands of candidate setups – combinations of well locations, timing, and rates – and returns a small set of best trade-off options for comparison. At its core, OPTIMA uses a U-Net-based surrogate model – a deep-learning emulator of a groundwater flow and transport simulator – to dramatically accelerate scenario evaluations. Coupling this fast surrogate with the NSGA-II (Non-dominated Sorting Genetic Algorithm II) evolutionary algorithm, OPTIMA explores a wide decision space of well locations and schedules to identify Pareto-optimal solutions that trade off key objectives (e.g., minimizing cleanup time, maximizing contaminant mass removal, and minimizing plume extent). The tool outputs a family of optimal configurations and visualizes their trade-offs (Pareto frontiers of cleanup metrics and maps of optimized well placements). Site managers can use these results to understand the range of viable strategies and to select candidate designs for more detailed verification. OPTIMA is currently under active development and not yet fully released; this guide provides early documentation to support planning and gather user feedback.

54 ENVIRONMENTAL SCIENCES↗

Digital Twin Framework for PIP-II Linac: AI-Driven Multi-Scale Modeling from Ion Source to 800 MeV

The PIP-II linac will enable >1.2 MW beam power for DUNE, requiring unprecedented operational reliability across its warm front-end (RFQ, MEBT) and five distinct SRF sections operating at 162.5/325/650 MHz. We present a comprehensive digital twin framework uniquely combining a fully differentiable fast beam transport code with neural network surrogates trained on high-fidelity PIC simulations, capturing space charge and nonlinear dynamics beyond traditional envelope codes while achieving 10⁴ speedup at <1% accuracy. End-to-end differentiability enables gradient-based optimization across 500+ parameters simultaneously previously impossible with conventional tools while the model incorporates static/dynamic errors and serves as a virtual commissioning platform for diverse hardware integration. The framework facilitates reinforcement learning for pulsed/CW mode transitions, predictive maintenance through anomaly detection, and autonomous tuning algorithm development with real-time execution capability. Validation against physics simulations shows excellent agreement for the front-end, with initial results demonstrating potential for 30% commissioning time reduction and proactive fault mitigation, providing a scalable blueprint for operating next-generation high-intensity accelerators.

Pathak, Abhishek [Fermilab] (ORCID:000000021704208↗

Using hypermedia to develop an intelligent tutorial/diagnostic system for the Space Shuttle Main Engine Controller Lab

This is a tutorial/diagnostic system for training personnel in the use of the Space Shuttle Main Engine Controller (SSMEC) Simulation Lab. It also provides a diagnostic capable of isolating lab failures at least to the major lab component. The system was implemented using Hypercard, which is an program of hypermedia running on Apple Macintosh computers. Hypercard proved to be a viable platform for the development and use of sophisticated tutorial systems and moderately capable diagnostic systems. This tutorial/diagnostic system uses the basic Hypercard tools to provide the tutorial. The diagnostic part of the system uses a simple interpreter written in the Hypercard language (Hypertalk) to implement the backward chaining rule based logic commonly found in diagnostic systems using Prolog. Some of the advantages of Hypercard in developing this type of system include sophisticated graphics, animation, sound and voice capabilities, its ability as a hypermedia tool, and its ability to include digitized pictures. The major disadvantage is the slow execution time for evaluation of rules (due to the interpretive processing of the language). Other disadvantages include the limitation on the size of the cards, that color is not supported, that it does not support grey scale graphics, and its lack of selectable fonts for text fields.

Oreilly, Daniel↗

RAPID: Collaborative Commanding and Monitoring of Lunar Assets

RAPID (Robot Application Programming Interface Delegate) software utilizes highly robust technology to facilitate commanding and monitoring of lunar assets. RAPID provides the ability for intercenter communication, since these assets are developed in multiple NASA centers. RAPID is targeted at the task of lunar operations; specifically, operations that deal with robotic assets, cranes, and astronaut spacesuits, often developed at different NASA centers. RAPID allows for a uniform way to command and monitor these assets. Commands can be issued to take images, and monitoring is done via telemetry data from the asset. There are two unique features to RAPID: First, it allows any operator from any NASA center to control any NASA lunar asset, regardless of location. Second, by abstracting the native language for specific assets to a common set of messages, an operator may control and monitor any NASA lunar asset by being trained only on the use of RAPID, rather than the specific asset. RAPID is easier to use and more powerful than its predecessor, the Astronaut Interface Device (AID). Utilizing the new robust middleware, DDS (Data Distribution System), developing in RAPID has increased significantly over the old middleware. The API is built upon the Java Eclipse Platform, which combined with DDS, provides platform-independent software architecture, simplifying development of RAPID components. As RAPID continues to evolve and new messages are being designed and implemented, operators for future lunar missions will have a rich environment for commanding and monitoring assets.

Torres, Recaredo J.↗

NASA GeneLab: Open Science for Life in Space

The NASA GeneLab project capitalizes on multi-omic technologies to maximize the return on spaceflight experiments. To do this, GeneLab maintains a publicly accessible database (GLDS) that houses spaceflight and spaceflight relevant multi-omics data and collaborates with NASA principal investigators and projects to generate additional omics data. GeneLab houses more than 350 transcriptomic, proteomic, metabolomic and epigenomic datasets from plant, animal and microbial experiments, with a growing number of these having been produced by the GeneLab Sequencing Lab. The GLDS contains rich metadata about each experiment and has integrated radiation dosimetry data from experiments flown on the Space Shuttle, International Space Station, and Free Flying spacecrafts. With the increasing amount and complexity of omics data being generated, GeneLab utilizes community-defined, common models for metadata and terminology so that omics data and results are discoverable and reliably reproducible. GeneLab uses the ISA-Tab specification and semantic model for organizing and representing omics metadata. In addition to metadata standards, data files must be open-source file or common exchange formats to ensure accessibility and usability by all users. To ease data ingestion and transfer, the web-based submission tool allows PIs a user-friendly user interface to curate, organize, and publish their space relevant omics data. In the more recent years, data curation and submission portal has incorporated the FAIR principles making data findable, accessible, interoperable, and reusable. To increase reusability of data, GeneLab has implemented an effort to present processed data in the GLDS in addition to the raw omics data. The processed data will enable interpretation of the data by a larger group of students, scientists and the general public. Standard pipelines for the transformation of raw data into visualizations were developed by four GeneLab Analysis Working Groups (animals, plants, microbes, multi-omics) comprised of over 200 scientists from NASA, industry, and academia. To explore the data, the GLDS provides users various tools for data analysis, collaborative workspace for file storage and sharing, and a visualization portal. The analysis platform built using the Galaxy toolshed provides access to a broad variety of users including those with limited bioinformatics experience and students to learn how to analyze spaceflight omics data. The visualization portal takes GeneLab one step closer to data democratization by removing all bioinformatics requisites to interpret transcriptomics data hosted in the repository. To train the next generation of scientists, NASA offers training programs such as GeneLab 4 High School (GL4HS) and GeneLab 4 Universities. NLM Curation at a Scale Workshop 2022 | NASA GeneLab (GL4U) to teach students bioinformatics and computational biology methods to analyze omics data. Discoveries made using GeneLab have begun and will continue to deepen our understanding of biology, advance the field of genomics, and help to discover cures for diseases, create better diagnostic tools, and ultimately allow astronauts to better withstand the rigors of long-duration spaceflight.

GeneLab↗

Digital Twin Framework for PIP-II Linac: AI-Driven Multi-Scale Modeling from Ion Source to 800 MeV

The PIP-II linac will enable >1.2 MW beam power for DUNE, requiring unprecedented operational reliability across its warm front-end (RFQ, MEBT) and five distinct SRF sections operating at 162.5/325/650 MHz. We present a comprehensive digital twin framework uniquely combining a fully differentiable fast beam transport code with neural network surrogates trained on high-fidelity PIC simulations, capturing space charge and nonlinear dynamics beyond traditional envelope codes while achieving 10⁴× speedup at <1% accuracy. End-to-end differentiability enables gradient-based optimization across 500+ parameters simultaneously—previously impossible with conventional tools—while the model incorporates static/dynamic errors and serves as a virtual commissioning platform for diverse hardware integration. The framework facilitates reinforcement learning for pulsed/CW mode transitions, predictive maintenance through anomaly detection, and autonomous tuning algorithm development with real-time execution capability. Validation against physics simulations shows excellent agreement for the front-end, with initial results demonstrating potential for 30% commissioning time reduction and proactive fault mitigation, providing a scalable blueprint for operating next-generation high-intensity accelerators.

Pathak, Abhishek [Fermilab] (ORCID:000000021704208↗

Establishing a Platform for Interest and Education in Interventional Radiology Amongst Radiology Trainees

Purpose: Over recent times, procedural Radiologists have begun to establish themselves as the distinct subspecialty of Interventional Radiology (IR). The Interventional Radiology Society of Australasia (IRSA) was established in 1982 to share collaborative ideas, encourage research, and promote education. IRSA developed a weekend registrar workshop attended by Radiology Registrars from Australia and New Zealand. In the 2018 event, we surveyed the Registrars to identify their interest in IR training before and after the workshop. Materials and Methods: The event was held over a weekend and consisted of both lectures and hands-on workshops. A survey was handed to all 67 registrants of the workshop and there was a 55% response rate including 78% of females in attendance. Results: Before the workshop, trainees rated their interest in IR training at a mean of 3.7 out of 5. After the workshop, trainees rated their interest in IR training as an average of 4.4 out of 5 (p < 0.001). The difference in interest between males and females before the workshop (4.0 vs. 3.1) was significant (p = 0.003), however after the workshop (4.5 vs. 4.1) was not significant (p = 0.07). The change in interest from attending the workshop was significant between genders, p = 0.03 (male interest increased mean 0.5, female increased mean 1.0). Conclusion: We show that a program of lectures and workshops designed to generate interest in IR leads to a significant increase in training interest, particularly amongst females. Other subspecialty groups should consider this type of intervention and promote ongoing education and inspiration. Level of Evidence: Cross-sectional study, Level IV.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Toward an Autonomous Workflow for Single Crystal Neutron Diffraction

The operation of the neutron facility relies heavily on beamline scientists. Some experiments can take one or two days with experts making decisions along the way. Leveraging the computing power of HPC platforms and AI advances in image analyses, here we demonstrate an autonomous workflow for the single-crystal neutron diffraction experiments. The workflow consists of three components: an inference service that provides real-time AI segmentation on the image stream from the experiments conducted at the neutron facility, a continuous integration service that launches distributed training jobs on Summit to update the AI model on newly collected images, and a frontend web service to display the AI tagged images to the expert. Ultimately, the feedback can be directly fed to the equipment at the edge in deciding the next-step experiment without requiring an expert in the loop. With the analyses of the requirements and benchmarks of the performance for each component, this effort serves as the first step toward an autonomous workflow for real-time experiment steering at ORNL neutron facilities.

Yin, Junqi↗

Autonomous phase mapping of gold nanoparticles synthesis with differentiable models of spectral shape

Autonomous experimentation–or self-driving labs–offers a systematic approach to accelerate materials discovery by integrating automated synthesis, characterization, and data-driven decision-making. We present a closed-loop workflow for the on-demand synthesis and structural characterization of colloidal gold nanoparticles, enabling direct mapping from composition to nanoscale structure. Our framework leverages differentiable models of spectral shape to address two central tasks in self-driving labs: (a) phase mapping, or identifying compositional regions with distinct structural behavior; and (b) material retrosynthesis, or optimizing compositions for target structure. Using functional data analysis, we develop a data-driven model with generative pre-training, active learning, and high-throughput experiments to predict spectral responses across composition space. We demonstrate the approach on seed-mediated growth of gold nanoparticles, showcasing its ability to extract design rules, reveal secondary interactions, and efficiently navigate morphology space. Gradient-based optimization of the models enables inverse design, making this a unified platform.

36 MATERIALS SCIENCE↗