Search NASA⌕ Search

SEARCH · Search NASA

Results for “multi-objective”

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 433 records · Page 24
All sourcesNASA NTRSDOE OSTI
NASA NTRSDate not supplied

The Inspectability Metric: A Formalized System Of Measurement Enabling The Design For Inspection Framework

Nondestructive evaluation (NDE) engineers are often confronted with structural design choices that present challenges to meeting inspection requirements. These challenges, at best, increase the resources needed to design an inspection solution and, at worst, require resource intensive redesign of the structure. If the inspectability of the structure can be determined early in the design cycle, these challenging inspection scenarios can be avoided. The emergence of additive manufacturing has further compounded this problem by enabling the creation of highly optimized structures with no regard to inspection constraints. Design for inspection (DFI) offers a framework to integrate nondestructive evaluation (NDE) into the design process to alleviate the mechanisms that produce uninspectable designs. DFI is the concept of including inspectability in a multi-objective optimization framework so that it can be considered in parallel to other metrics such as mass and manufacturability. This allows rapid evaluation of the trade-off between design metrics to find solutions that meet the inspection needs of a particular material system, structural concept, or vehicle program. To enable DFI, there must be a system by which the inspectability of a structure can be measured. This system must be agile to produce results quickly, it must be versatile to work with the type of incomplete information one would encounter early in the design process (such as lack of inspection requirements), and it must be delivered in a form that is easily understood by designers. To meet this need, this presentation introduces the novel inspectability metric as a system to measure inspectability. The inspectability metric is a standardized, automation friendly procedure that uses simulations to determine inspectability. Along with guidelines to properly process designs and integrate with existing workflows, the inspectability metric provides a suite of simulation tests to interrogate the ability to find defects and the sensitivity to variability. The testing rubric is designed to maximize the coverage of the parameter space while minimizing the number of simulations needed. The inspectability metric has been in development in collaboration with industry partners to ensure compatibility with modern simulation tools and aerospace design workflows. In this study, we will demonstrate how the inspectability metric is able to determine the inspectability of multiple types of structures, including aerospace composites and additively manufactured parts. We will then show how the inspectability score can be plugged into existing design optimization tasks, such as structural sizing algorithms or design for manufacturing (DFM) frameworks.

Design for inspection↗
NASA NTRS2024

Optimization of Mission Enabling Broadband Lyman Ultraviolet Series to Infrared Aluminum Lithium Fluoride Reflectors

The Habitable Worlds Observatory (HWO) concept NASA flagship mission aims extend the ultraviolet (UV) observation range and sensitivity capabilities of the Hubble Space Telescope (HST) with respective goals of reaching photons down to 100 nm and containing an ultraviolet multi-object spectrograph with sensitivity improved 30-100 times. Adding the sensitivity capabilities near the Lyman series spectral range of ~100-121.6 nm reduces the reliance of redshift for key diagnostic ion lines such as O VI (103.2 nm), and heavily ionized gases of Ne VIII (77.5 nm) and MgX (62.5 nm). Clean and unoxidized aluminum (Al) reflectors provide optimal efficiency over a broadband down to the extreme ultraviolet (EUV) wavelength of ~83 nm. To prevent oxidation and maintain high spectral efficiency down to ~100 nm, the Al reflector is conventionally overcoated with a thin lithium fluoride (LiF) layer. We investigate the optimization of the deposition of the Al+LiF coating material composition to attain high spectral efficiency broadband reflectors down to 100 nm and enable the scientific goals of HWO. Lithium hexafluoroaluminate (Li3AlF6) is investigated as a different composition form of LiF to overcoat Al. The spectral efficiency and durability of the reflectors is evaluated. We will demonstrate the use of Li3AlF6 as an overcoat to protect aluminum yields ultra-high efficiency at the Hydrogen Lyman-alpha (HLyα, 121.6 nm) line with experimental reflectance values peaking up to 99%.

optical coatings↗
NASA NTRS2024

Optimization of Mission Enabling Broadband Lyman Ultraviolet Series to Infrared Aluminum Lithium Fluoride Reflectors

The Habitable Worlds Observatory (HWO) concept NASA flagship mission aims extend the ultraviolet (UV) observation range and sensitivity capabilities of the Hubble Space Telescope (HST) with respective goals of reaching photons down to 100 nm and containing an ultraviolet multi-object spectrograph with sensitivity improved 30-100 times. Adding the sensitivity capabilities near the Lyman series spectral range of ~100-121.6 nm reduces the reliance of redshift for key diagnostic ion lines such as O VI (103.2 nm), and heavily ionized gases of Ne VIII (77.5 nm) and MgX (62.5 nm). Clean and unoxidized aluminum (Al) reflectors provide optimal efficiency over a broadband down to the extreme ultraviolet (EUV) wavelength of ~83 nm. To prevent oxidation and maintain high spectral efficiency down to ~100 nm, the Al reflector is conventionally overcoated with a thin lithium fluoride (LiF) layer. We investigate the optimization of the deposition of the Al+LiF coating material composition to attain high spectral efficiency broadband reflectors down to 100 nm and enable the scientific goals of HWO. Lithium hexafluoroaluminate (Li3AlF6) is investigated as a different composition form of LiF to overcoat Al. The spectral efficiency and durability of the reflectors is evaluated. We will demonstrate the use of Li3AlF6 as an overcoat to protect aluminum yields ultra-high efficiency at the Hydrogen Lyman-alpha (HLyα, 121.6 nm) line with experimental reflectance values peaking up to 99%.

optical coatings↗
NASA NTRSDate not supplied

Optimal Communication Topology Construction and Sensor Selection for Independent Airspace Surveillance

The paper presents an approach with no estimation feedback to sensors selection and communication network topology computation for independent airspace surveillance with maximum outcome and minimum cost using ground based distributed sensing, computing and communication network infrastructure. The selection criteria includes maximum airspace coverage with minimal resources, minimum communication time and power consumption while guaranteeing the system observability and providing in-time high quality information to both stationary and mobile users. The developed algorithms use multi-objective optimization strategy taking into account trade-offs between conflicting objectives and are implemented using off-the-shelf computational tools. The algorithms are validated in a desktop simulation environment using synthetic sensors data generated for a simulated multi-vehicle flight scenario in the selected regional airspace and parameters of a notional wireless communication network.

Distributed sensing↗
DOE OSTI2024

Predicting Li-ion Battery Performance for Impurity-doped NMC Cathodes Using Deep Learning

With the electric vehicle (EV) market expansion and the energy sector's shift towards electrification, the demand for battery metals, including lithium (Li), cobalt (Co), and nickel (Ni), is set to surpass supply. A critical knowledge gap exists in the purity standards for battery precursors and the impact of impurities on battery performance. Addressing this, our study employs a Deep Machine Learning (DL) based multi-objective optimization approach to interpret the relationship between metal impurities in domestic battery resources and their effects on battery performance. We analyze experimental data from Li-ion batteries with NMC (Nickel-Manganese-Cobalt oxide) cathodes over 1000 cycles, representing approximately ~6-8 months of operation, to establish a baseline of performance without impurities. Leveraging this data, we develop a Physics-Informed Deep Learning (PIDL) framework to extend our findings to cases that include metal impurities (e.g., Fe, Cu, Al) ranging from (0.001 - 0.01) %, respectively. By incorporating physics-based features, our PIDL model can accurately estimate the performance of NMC cathodes doped with various metal impurities to provide rapid design decisions. This research paves the way for informed decisions in Li-ion battery material design and optimization, ensuring the sustainable growth of the EV market and the broader energy sector.

25 ENERGY STORAGE↗
DOE OSTI2024

CEBAF Injector for K Long Beam Conditions

The Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Lab concurrently operates four experimental Halls with distinct bunch charge specifications and repetition rates. Numerous critical beam parameters within CEBAF are configured in the injector, some remaining unchanged throughout the accelerator. Consequently, the injector plays a crucial role in determining final beam characteristics, including bunch structure, beam sizes, bunch lengths, energy spread, and beam transmission. The Jefferson Lab KL experiment is scheduled to take place at CEBAF in Hall D, featuring a much lower bunch repetition rate of 7.80 MHz or 15.59 MHz, below the nominal values of 249.5 MHz or 499 MHz. Although the proposed average current of 5 ?A or 10 ?A is low compared to the maximum CEBAF cur- rent of approximately 180 ?A, the corresponding bunch charge is unusually high for CEBAF injector operation. This study focuses on the behavior of low-repetition-rate, high-bunch- charge (0.32 to 0.64 pC) beams in the CEBAF injector. We investigated the evolution and transmission of low-charge beams to space-charge dominated high-charge beams in the front end of the CEBAF injector for two configurations: the pre-existing CEBAF Phase 1 injector upgrade, operated at 130 kV, and the existing CEBAF Phase 2 injector upgrade, operated at 140 kV, 180 kV, and to be operated at 200 kV. The electron beam through the CEBAF injector is characterized using beam dynamics simulations and comparisons with the available measurements performed at 130 kV. Multi-objective genetic optimizations of the CEBAF injector were performed to determine the operating magnetic elements and RF settings for the evolution and transmission of low, moderate, and high charge beams in the CEBAF injector at 180 kV and 200 kV DC gun voltages. Subsequently, simulations at the same voltages were conducted to obtain the beam characteristics at the front end of the CEBAF injector. The laser spot size and laser pulse length at the cathode were varied to observe their effects on beam transmission in the injector at different voltages (130 kV, 180 kV, and 200 kV). Experimental studies at 130 kV, 140 kV, and 180 kV validate the simulations. Beam study measurements are carried out using EPICS tools, while optimizations and simulations are facilitated by General Particle Tracer. Based on the findings, optimal parameters for the upcoming Jefferson Lab KL experiment are proposed, utilizing a lower repetition rate and higher bunch charge

Pokharel, Sunil↗
DOE OSTI2024

Optimization of R290 variable geometry heat exchangers

Air-to-refrigerant heat exchangers (HX) are vital components in Heating, Ventilating and Air Conditioning and Refrigeration (HVAC&R) equipment. Recent literature has proposed utilizing shape-optimized, non-round tubes to reduce component size and refrigerant charge, thus enabling the adoption of low-GWP natural refrigerants like R290. However, most designs are restricted to fixed tube configurations, limiting the overall performance potential. In this work, multi-objective optimization is used in a staged approach to develop variable geometry multi-pass air-to-R290 HXs with minimal volume and airside pressure drop. Non-round tubes are used throughout the domain and each pass may have different tube arrangements to maximize HX performance. Compared to conventional fin-tube HXs, the optimized designs demonstrate more than 60% reductions in envelope volume, 30% reductions in face area, and up to 24% reduction in airside pressure drop. Additionally, the refrigerant charge reduction was up to 64%, thus enabling safe use of flammable and mildly-flammable refrigerants.

42 ENGINEERING↗
DOE OSTI2024

Optimization and Experimental Validation of Annular Finned PCM-HX for a Domestic Hot Water Heater Application

The load profile for domestic water heating is time-dependent and can result in high energy demand during peak operating times. Shifting this peak load can have significant environmental and economic impacts. Phase change material (PCM)-based thermal energy storage (TES) is a potentially useful technology for peak load shifting in domestic hot water (DHW) applications thanks to its high latent heat and energy density. In this study, an annular finned-tube PCM-HX design concept was optimized for a load-shifting TES unit to meet the Department of Energy standard for a medium-usage DHW heater using a resistance-capacitance model (RCM) integrated with a Multi-Objective Genetic Algorithm. The optimized design comprised 70 identical annular finned-tube PCM-HX units connected in parallel and utilizing RT62HC as the PCM. A single PCM-HX unit was prototyped and tested in a vertically oriented setup with upward heat transfer fluid (HTF) flow. The hot water supply time was defined based on a cutoff temperature of 51.7°C. The as-designed mass flow rate (1.5 g/s) was tested to assess the performance of the prototyped PCM-HX unit for RCM validation. For the experimental investigation, RTD sensor bundles measured HTF temperature at the PCM-HX inlet and outlet, and a Coriolis flow meter accurately measured the HTF mass flow rate. The simulated discharging power underpredicted the experimental result by about 12%, and the simulated hot water supply time underpredicted the experimental result by approximately 13% for the as-designed mass flow rate (1.5 g/s). The average deviation of the hot water supply temperature between the experimental and RCM results during the complete PCM solidification process was 1.3 K for the as-designed mass flow rate. The overall good agreement between the experimental and RCM results provides confidence that computationally efficient models such as RCM can be utilized for design optimization of PCM-HXs.

42 ENGINEERING↗
DOE OSTI2024

Accelerated CO2 Storage Optimization Using Multi-Resolution Fourier Neural Operator at the Illinois Basin Decatur Project (IBDP)

This paper presents a deep learning-based approach for optimizing CO2 injection in carbon capture and storage (CCS) operations. We developed a multi-resolution machine learning model to significantly reduce data generation costs. Utilizing this proxy model, we implemented a multi-objective genetic algorithm to optimize well control during the CO2 injection process. The proposed approach was applied to the Illinois Basin Decatur Project (IBDP), successfully optimizing the CO2 injection schedule based on three key objectives: maximizing the amount of CO2 stored, maximizing sweep efficiency, and minimizing pressure increase. The use of the proxy model accelerated the optimization workflow by two orders of magnitude, while the cost of data generation for the proxy model was reduced by 90% by utilizing a coarse-scale model.

accelerated CO2 storage optimization↗
DOE OSTI2025

A Joint Search for the Electromagnetic Counterpart to the Gravitational-Wave Binary Black-Hole Merger Candidate S250328ae with the Dark Energy Camera and the Prime Focus Spectrograph

The first detection of an optical counterpart to a gravitational wave signal revealed that collaborative efforts between instruments with different specializations provide a unique opportunity to acquire impactful multi-messenger data. We present results of such a joint search with the Dark Energy Camera (DECam) and Prime Focus Spectrograph (PFS) for the optical counterpart of the LIGO-Virgo-KAGRA event S250328ae, a binary black hole merger candidate of high significance detected at a distance of 511$\pm$82 Mpc and localized within an area of 3 (15) square degrees at 50% (90%) confidence. We observed the 90% confidence area with DECam and identified 36 high-confidence transient candidates after image processing, candidate selection, and candidate vetting. We observed with PFS to obtain optical spectra of DECam candidates, Swift-XRT candidates, and potential host galaxies of S250328ae. In total, 3897 targets were observed by seven pointings covering ~50% of the 90% confidence area. After template fitting and visual inspection, we identified 12 SNe, 159 QSOs, 2975 galaxies, and 131 stars. With the joint observations of DECam and PFS, we found variability in 12 SNe, 139 QSOs, 37 galaxies, and 2 stars. We do not identify any confident optical counterparts, though the association is not ruled out for three variable candidates that are not observed by PFS and 6 QSO candidates without clear variability if the optical counterpart of S250328ae is faint. Despite the lack of confident optical counterparts, this paper serves as a framework for future collaborations between wide-field imagers and multi-object spectrographs to maximize multi-messenger analyses.

Zhang, Haibin [Natl. Astron. Observ. of Japan] (OR↗
DOE OSTI2025

Surrogate Neural Architecture Codesign Package (SNAC-Pack)

Neural Architecture Search is a powerful approach for automating model design, but existing methods struggle to accurately optimize for real hardware performance, often relying on proxy metrics such as bit operations. We present Surrogate Neural Architecture Codesign Package (SNAC-Pack), an integrated framework that automates the discovery and optimization of neural networks focusing on FPGA deployment. SNAC-Pack combines Neural Architecture Codesign's multi-stage search capabilities with the Resource Utilization and Latency Estimator, enabling multi-objective optimization across accuracy, FPGA resource utilization, and latency without requiring time-intensive synthesis for each candidate model. We demonstrate SNAC-Pack on a high energy physics jet classification task, achieving 63.84% accuracy with resource estimation. When synthesized on a Xilinx Virtex UltraScale+ VU13P FPGA, the SNAC-Pack model matches baseline accuracy while maintaining comparable resource utilization to models optimized using traditional BOPs metrics. This work demonstrates the potential of hardware-aware neural architecture search for resource-constrained deployments and provides an open-source framework for automating the design of efficient FPGA-accelerated models.

Weitz, Jason [UC, San Diego] (ORCID:00090004631535↗
DOE OSTI2025

Machine Learning-Driven Optimization of Building Enclosures for Moisture Durability and Thermal Performance

The design of moisture-durable building enclosures with low embodied carbon often involves an iterative process of selecting the materials for the specific exposure conditions to meet the performance requirements. While hygrothermal simulations are commonly used to evaluate moisture durability, they often require advanced expertise for proper implementation. Machine learning (ML) provides a promising alternative by streamlining the design process and minimizing the reliance on complex simulations. This study presents a machine learning-based approach for predicting moisture durability in residential wall assemblies. The ML model was trained to estimate the mold index and maximum moisture content of various layers under typical exposure conditions. The model achieved a high predictive accuracy, with a coefficient of determination (R²) exceeding 0.90 when compared to traditional hygrothermal simulations on materials that were not part of training the ML model. Building on these results, the ML model was developed into a practical tool for optimizing wall assembly designs. This tool allows users to automatically optimize material selections based on energy, moisture, and carbon performance criteria. By incorporating multi-objective optimization, the tool identifies configurations that minimize embodied carbon while maintaining moisture safety and code-compliant thermal performance. Additionally, it provides insights into how material choices influence assembly durability, energy efficiency, and carbon reduction. The tool will be implemented in the Building Science Advisor (BSA) to enhance its performance and provide more granularity on the results. This research highlights the potential for ML-driven tools to simplify the design of high-performance building enclosures, offering architects and engineers a faster, more efficient way to balance critical performance factors.

Salonvaara, Mikael [ORNL] (ORCID:0000000318991554)↗
DOE OSTI2026

Use of Digital Real-Time Simulation and Optimization to Identify Maximum Real Power Injection on the Banshee Distribution Network: Preprint

This study investigates the hosting capacity of the Banshee Distribution Network by optimizing the real power injection at carefully selected Distributed Energy Resource (DER) locations. The analysis is conducted within the framework of power system operational constraints, including bus voltage ranges, thermal line ratings, and transformer loading limits. A Python-based Genetic Algorithm (GA), implemented using the PyGAD library, is employed to iteratively identify the optimal power injection configuration that maximizes network utilization while preserving system reliability. The methodology integrates a real-time simulation environment using the Real-Time Digital Simulator (RTDS), allowing high-fidelity evaluation of power flow and voltage behavior under each proposed injection scenario. By coupling the optimization algorithm with real-time simulation feedback, this approach ensures that both static and dynamic constraints are enforced during the evaluation process. The GA leverages evolutionary operators such as selection, crossover, and mutation to navigate the nonlinear search space efficiently. The results of the study delineate the feasible hosting capacity at three targeted buses, reflecting maximum real power levels that can be injected without causing voltage violations, transformer overloading, or line congestion. These findings provide a decision- support tool for distribution planners and utilities aiming to integrate higher penetrations of DERs in existing infrastructure. Additionally, the work lays the foundation for extending such optimization techniques to multi-objective formulations, including economic dispatch and reactive power coordination, in future studies.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2025

DEVELOPMENT AND APPLICATION OF RISK ANALYSIS TOOLKIT FOR PLANT RESOURCE OPTIMIZATION

This paper presents the development of methods and tools that are being designed to optimize plant operations (e.g., maintenance/replacement schedules and optimal maintenance postures for plant components) in a manner that is more cost effective than current approaches and makes better use of available component health and cost data. These methods include both data- and model-based optimization methods. Model-based optimization methods directly include reliability and cost models to determine an optimal plant operational strategy. We consider gradient-based and evolutionary (based on genetic algorithms) optimization methods. The second class of methods target more specific use cases (e.g., project schedule optimization) and are not based on reliability models directly, but they require specific component reliability and cost data. This class of methods is based on variants of the knapsack problem with an aim to determine an optimal project schedule that maximizes the overall NPV. This paper also presents multi-objective methods designed to identify an optimal maintenance posture based on a Pareto frontier analysis. Rather than dictating the “right” tradeoff (i.e., identify the absolute best posture), we show how it is possible to perform a trade space exploration approach (i.e., identify value and costs of several postures and let the analysis account for desired value and cost metrics). This is performed by identifying maintenance postures that maximize value (e.g., system availability) and minimize operational costs, i.e., the Pareto frontier in a value-cost trade space. For all these methods we present detailed applicative examples that show their validity from a decision-making perspective.

97 - MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Implementation of fuel management multi-cycle optimization capabilities in RAVEN optimization framework

Optimization in nuclear fuel-management assists the core reload engineer with finding optimal out-of-core and in-core strategies. RAVEN is INL’s open source software that is equipped with fuel-management optimization capabilities including single-cycle, single- and multi-objective optimization of pressurized water reactors (PWRs) loading patterns (LP) of a fresh core using genetic algorithm (GA) and non-dominated sorting genetic algorithm (NSGA-II). In practice, however, medium and long term planning of fuel-management needs a multi-cycle approach, where the history and availability of fuel assemblies is considered in the optimization process. In this paper, we present a description of an initial expansion of RAVEN fuel-management optimization capabilities for a multi-cycle optimization framework. N-th cycle optimization capabilities that account for the unique history of recycled fuel assembly in the core were added. The multi-cycle optimization approach taken is formulated as a cycle-wise optimization problem where out-of-core decisions are used to onset each cycle optimization. Out-of-core decisions are managed externally to the in-core optimization by a fuel inventory management module. A proof-of-concept optimization problem is also presented.

42 - ENGINEERING↗
DOE OSTI2024

Proof-of-Concept for Sensor Modeling in MOOSE for the Design of Autonomous Nuclear Reactor Control

Autonomous operation is essential for the deployment of microreactors and fission batteries, both in terrestrial and space applications. For this reason, recent studies have investigated autonomous control by using adaptive model predictive control and multi-objective optimization for heat pipe–cooled microreactors under normal and heat pipe failure conditions. However, prototypes of microreactors and fission batteries do not exist yet, and even the design space has not been narrowed down conclusively, making the instrumentation and control system design difficult. For this reason, there is a need for flexible computational capabilities to create a numerical stand-in of potential microreactor and fission battery designs. The latter can be used to design and test control strategies to support autonomous operations. In this poster, we describe the initial implementation of a pluggable sensor system for the easy implementation of realistic sensor models in the multiphysics object-oriented simulation environment (MOOSE) framework. This new capability will enable MOOSE users to create a numerical stand-in of microreactors and fission batteries, ultimately allowing them to easily test new control algorithms, and instrumentation strategies for advanced systems in the design phase.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2025

LC-Opt: Benchmarking Reinforcement Learning and Agentic AI for End-to-End Liquid Cooling Optimization in Data Centers

Liquid cooling is critical for thermal management in high-density data centers with the rising AI workloads. However, machine learning-based controllers are essential to unlock greater energy efficiency and reliability, promoting sustainability. We present LC-Opt, a Sustainable Liquid Cooling (LC) benchmark environment, for reinforcement learning (RL) control strategies in energy-efficient liquid cooling of high-performance computing (HPC) systems. Built on the baseline of a high-fidelity digital twin of Oak Ridge National Lab's Frontier Supercomputer cooling system, LC-Opt provides detailed Modelica-based end-to-end models spanning site-level cooling towers to data center cabinets and server blade groups. RL agents optimize critical thermal controls like liquid supply temperature, flow rate, and granular valve actuation at the IT cabinet level, as well as cooling tower (CT) setpoints through a Gymnasium interface, with dynamic changes in workloads. This environment creates a multi-objective real-time optimization challenge balancing local thermal regulation and global energy efficiency, and also supports additional components like a heat recovery unit (HRU). We benchmark centralized and decentralized multi-agent RL approaches, demonstrate policy distillation into decision and regression trees for interpretable control, and explore LLM-based methods that explain control actions in natural language through an agentic mesh architecture designed to foster user trust and simplify system management. LC-Opt democratizes access to detailed, customizable liquid cooling models, enabling the ML community, operators, and vendors to develop sustainable data center liquid cooling control solutions.

Naug, Avisek [Hewlett Packard Enterprise]↗
DOE OSTI2026

Surrogate Neural Architecture Codesign Package (SNAC-Pack)

Neural architecture search (NAS) is a powerful approach for automating model design, but existing methods often optimize for accuracy alone or rely on proxy metrics such as bit operations (BOPs) that correlate poorly with hardware cost. This gap is particularly large for FPGA deployment, where cost is dominated by a multi-dimensional budget of lookup tables, DSPs, flip-flops, BRAM, and latency. We present the Surrogate Neural Architecture Codesign Package (SNAC-Pack), an open-source AutoML framework for hardware-aware neural architecture codesign and end-to-end FPGA deployment. SNAC-Pack runs a multi-objective global search with Optuna and NSGA-II, loading trials to a shared SQLite store that enables parallel workers across compute nodes. A hardware surrogate model outputs per-trial resource and latency estimates, avoiding the synthesis cost that would otherwise dominate the search loop. A local search stage then applies quantization-aware training (QAT) together with iterative magnitude pruning in a combined compression loop, after which the final model is synthesized to FPGA firmware via the hls4ml Python library. A YAML configuration and an optional agentic frontend let users run the pipeline on new datasets without modifying the framework. We demonstrate SNAC-Pack on jet classification at the Large Hadron Collider and superconducting qubit readout, discovering compact architectures that match or exceed strong baselines on the task metric while reducing FPGA resource utilization and, in the qubit readout case, reducing the design space exploration process from months of manual fine-tuning to hours of automated search.

Weitz, Jason [UC, San Diego]↗