Search NASA⌕ Search

SEARCH · Search NASA

Results for “evolutionary computation”

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 181 records · Page 10

NeoN: Neuromorphic Control for Autonomous Robotic Navigation

In this paper we describe the use of a new neuromorphic computing framework to implement the navigation system for a roaming, obstacle avoidance robot. Using a Dynamic Adaptive Neural Network Array (DANNA) structure, our TENNLab (Laboratory of Tennesseans Exploring Neural Networks) hardware/software co-design framework and evolutionary optimization (EO) as the training algorithm, we create, train, implement, and test a spiking neural network autonomous robot control system using an array of neuromorphic computing elements built on an FPGA. The simplicity and flexibility of the DANNA neuromorphic computing elements allow for sufficient scale and connectivity on a Xilinx Kintex-7 FPGA to support sensory input and motor control for a mobile robot to navigate a dynamically changing environment. We further describe how more complex capabilities can be added using the same platform, e.g. object identification and tracking.

Mitchell, J. Parker↗

“Teamwork Makes the Dream Work”: Tribal Competition Evolutionary Search as a Surrogate for Free-Energy-Based Structural Predictions

Crystal structure prediction has been a grand challenge in material science owing to the large configurational space that one must explore. Evolutionary (genetic) algorithms coupled with first principles calculations are commonly used in crystal structure prediction to sample the ground and metastable states of materials based on configurational energies. However, crystal structure predictions at finite temperature (T), pressure (P), and composition (X) require a free-energy-based search that is often computationally expensive and tedious. Here in this paper, we introduce a new machine-learning workflow for structure prediction that is based on a concept inspired by the evolution of human tribes in primitive society. Our tribal genetic algorithm (GA) combines configurational sampling with evolutionary optimization to accurately predict entropically stabilized phases at finite (T, P, X), at a computational cost that is an order of magnitude smaller than that required for a free-energy-based search. In a departure from standard GA techniques, the populations of individuals are divided into multiple tribes based on a bond-order fingerprint, and genetic operations are modified to ensure that cluster configurations are sampled adequately to capture entropic contributions. Team competition introduced into the evolutionary process allows winning teams (representing a better set of individuals) to expand their sizes; this translates into a more expanded search of the phase space allowing us to explore solutions near possible global minimum. Each team explores a specific section of the structural phase space and avoids bias on solutions arising from the use of individual populations in a purely energy-based search. We demonstrate the efficacy of our approach by performing the structural prediction of a representative two-dimensional two-body system as well as Lennard-Jones clusters over a range of temperatures up to its melting point. Our approach outperforms the standard GA approaches and enables structural search under “real nonambient conditions” on both bulk systems and finite-sized clusters.

36 MATERIALS SCIENCE↗

In silico prediction of MOFs with high deliverable capacity or internal surface area

Metal–organic frameworks (MOFs) offer unprecedented atom-scale design and structural tunability, largely due to the vast number of possible organic linkers which can be utilized in their assembly. Exploration of this space of linkers allows identification of ranges of achievable material properties as well as discovery of optimal materials for a given application. Experimental exploration of the linker space has to date been quite limited due to the cost and complexity of synthesis, while high-throughput computational studies have mainly explored MOF materials based on known or readily available linkers. Here an evolutionary algorithm for de novo design of organic linkers for metal–organic frameworks is used to predict MOFs with either high methane deliverable capacity or methane accessible surface area. Known chemical reactions are applied in silico to a population of linkers to discover these MOFs. Through this design strategy, MOF candidates are found in the ten symmetric networks acs, cds, dia, hxg, lvt, nbo, pcu, rhr, sod, and tbo. Furthermore, the correlation between deliverable capacities and surface area is network dependent.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Multiobjective Hyperparameter Optimization for Deep Learning Interatomic Potential Training Using NSGA-II

Deep neural network (DNN) potentials are an emerging tool for simulation of dynamical atomistic systems, with the promise of quantum mechanical accuracy at speedups of 10000$\times$. As with other DNN methods, hyperparameters used during training can make a substantial difference in model accuracy, and optimal settings vary with dataset. To enable rapid tuning of hyperparameters for DNN potential training, we developed a scalable multiobjective optimization evolutionary algorithm for supercomputers and tested it on the Summit system at the Oak Ridge Leadership Computing Facility (OLCF). The multiobjective approach is required due to the coupling of two learned values defining the potential: the energy and force. Using a large-scale implementation of the NSGA-II algorithm adapted for training DNN potentials, we discovered several optimal multiobjective combinations, including best choices of activation functions, learning rate scaling scheme, and pairing of the two radial cutoffs used in the three dimensional descriptor function.

Coletti, Mark↗

ARENA: Adversary-Resistant Evolving Neural Architectures

Neural networks are becoming the cornerstone for national security prediction tasks. However, designing them requires significant research and trial/error, as they have many hyperparameters, including their computation graph (“architecture”). Neural architecture search (NAS) employs secondary optimizers to search for architectures maximizing objectives like accuracy. Evolutionary algorithms (EAs) are the most used class of optimizer for NAS. However, existing Python libraries for writing EAs limit the complexity of experiments a user can design. In this project, we built ARENA, a Python framework that encodes complex, hyper-realistic EAs. ARENA collects detailed information as it runs and is flexible enough to encode non-EA search algorithms. We tested ARENA on 4 toy optimization problems by encoding 3 search algorithms for each—random search, an EA, and simulated annealing. We also designed an EA that performs NAS on the MNIST dataset. Our experiments suggest the potential for immediate mission impact through solving lab-wide optimization problems.

97 MATHEMATICS AND COMPUTING↗

Can An Evolutionary Process Create English Text?

Critics of the conventional theory of biological evolution have asserted that while natural processes might result in some limited diversity, nothing fundamentally new can arise from 'random' evolution. In response, biologists such as Richard Dawkins have demonstrated that a computer program can generate a specific short phrase via evolution-like iterations starting with random gibberish. While such demonstrations are intriguing, they are flawed in that they have a fixed, pre-specified future target, whereas in real biological evolution there is no fixed future target, but only a complicated 'fitness landscape'. In this study, a significantly more sophisticated evolutionary scheme is employed to produce text segments reminiscent of a Charles Dickens novel. The aggregate size of these segments is larger than the computer program and the input Dickens text, even when comparing compressed data (as a measure of information content).

97 MATHEMATICS AND COMPUTING↗

From calibration to parameter learning: Harnessing the scaling effects of big data in geoscientific modeling

Abstract The behaviors and skills of models in many geosciences (e.g., hydrology and ecosystem sciences) strongly depend on spatially-varying parameters that need calibration. A well-calibrated model can reasonably propagate information from observations to unobserved variables via model physics, but traditional calibration is highly inefficient and results in non-unique solutions. Here we propose a novel differentiable parameter learning (dPL) framework that efficiently learns a global mapping between inputs (and optionally responses) and parameters. Crucially, dPL exhibits beneficial scaling curves not previously demonstrated to geoscientists: as training data increases, dPL achieves better performance, more physical coherence, and better generalizability (across space and uncalibrated variables), all with orders-of-magnitude lower computational cost. We demonstrate examples that learned from soil moisture and streamflow, where dPL drastically outperformed existing evolutionary and regionalization methods, or required only ~12.5% of the training data to achieve similar performance. The generic scheme promotes the integration of deep learning and process-based models, without mandating reimplementation.

54 ENVIRONMENTAL SCIENCES↗

Deep Reinforcement Learning-Based Control of Energy Storage for Interarea Oscillation Damping

With the increasing electricity consumption and lack of transmission investment, today's power systems are operated much closer to their limits, raising concerns of inter-area oscillations that deteriorate the system stability. Here, this article presents a novel energy storage placement and control approach for enhanced damping of interarea oscillations. Combining the residual analysis and dominant mode analysis, we are able to identify the advantageous locations for placing energy storage that achieve improved damping performance. To overcome the challenges, such as fixed control parameters and insufficient damping, we propose to use a deep reinforcement learning-based approach for energy storage control. A state-of-the-art guided surrogate-gradient-based evolutionary strategy is used to train a learning agent in a robust, efficient, and reproducible manner. Parallel computing is also adopted to speed up the training process. The proposed strategy has been tested on both medium and large-scale systems. The proposed methods have demonstrated their effectiveness in mitigating various interarea oscillations within a timeframe of 20 s, thereby averting system collapse and enhancing power grid stability effectively.

25 ENERGY STORAGE↗

The Eastern Renewable Generation Integration Study: Insights on System Stress: Preprint

The Eastern Renewable Generation Integration Study (ERGIS) explores the operational impacts of the wide spread adoption of wind and solar photovoltaics (PV) resources in North America's Eastern and Quebec Interconnections. We explore the impact of large scale adoption of wind and solar generation on the unit commitment and economic dispatch of the largest coordinated power system in the world by simulating hourly and five-minute operations. Using NREL's high-performance computing capabilities and new methodologies to model operations, we found that the modeled system, as simulated with evolutionary change in 2026, could balance the variability and uncertainty of wind and solar PV at a five-minute level under a variety of conditions. Our simulations achieve instantaneous penetrations that exceed 50% of load while meeting an annual penetration of 30% on an energy basis. The system meets balanced load and supply in all intervals, with modest curtailment, using technologies and practices that are widely available today. However, a variety of the conditions present in these simulations deviate substantially from historical practice. In this work, we analyze potentially stressful system conditions that occur in the simulations and identify opportunities for innovation, regulatory reform, and changes in operating practices that require further analysis to enable the transition to a system with more wind and solar PV.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Evolutionary inverse design of defects at graphene 2D lateral interfaces

Grain boundaries (GBs) in two-dimensional (2D) materials often have a profound impact on various material properties from mechanical to optical to electronic, yet predicting all possible GB formations is a challenge. Here, we introduce a workflow based on an evolutionary algorithm for exploring possible GBs formed at a lateral 2D interface. In a departure from conventional genetic algorithm based structure optimization methods, we perform genetic operations in the near interface region that allow us to be computationally efficient. We benchmark our method using graphene, which is a well-studied 2D material with a wide range of point defects. An empirical potential was used as the surrogate of the evolutionary search. More than 11.5 × 106 structures in total were evaluated for 128 GB orientations, and for each orientation, the ten best structures are recorded. A subset of low energy GBs predicted by empirical potential based search was relaxed by first-principles calculations and used to validate the energetic rank order. With the validated formation energy, we rank-ordered the best 128 GB structures and performed a detailed statistical analysis of primitive rings to find the correlation between the ring distribution and the formation energy. We found that for low energy GBs (below 0.5eV/Å), Stone–Wales defects will dominate, while structures with a higher energy (0.5–1.1eV/Å) show an increasing population of heptagons and nine-membered rings to form seven-nine defect pairs. For structures with energy higher than 1.1eV/Å, the percentage of octagons and nine-membered rings increases, which indicates that these two types of rings are not energetically favorable. Our proposed methodology is broadly applicable to explore defective low dimensional materials and represents a powerful tool that enables a systematic search of GBs of lateral interfaces for 2D materials.

Zhang, Jianan↗

Biocontainment of genetically modified organisms by synthetic protein design

Genetically modified organisms (GMOs) are increasingly deployed at large scales and in open environments. Genetic biocontainment strategies are needed to prevent unintended proliferation of GMOs in natural ecosystems. Existing biocontainment methods are insufficient because they impose evolutionary pressure on the organism to eject the safeguard by spontaneous mutagenesis or horizontal gene transfer, or because they can be circumvented by environmentally available compounds. In this paper, we computationally redesign essential enzymes in the first organism possessing an altered genetic code (Escherichia coli strain C321.ΔA) to confer metabolic dependence on non-standard amino acids for survival. The resulting GMOs cannot metabolically bypass their biocontainment mechanisms using known environmental compounds, and they exhibit unprecedented resistance to evolutionary escape through mutagenesis and horizontal gene transfer. Finally, this work provides a foundation for safer GMOs that are isolated from natural ecosystems by a reliance on synthetic metabolites.

59 BASIC BIOLOGICAL SCIENCES↗

Discovery and annotation of small proteins using genomics, proteomics and computational approaches

Small proteins (10 200 amino acids aa in length) encoded by short open reading frames (sORF) play important regulatory roles in various biological processes, including tumor progression, stress response, flowering, and hormone signaling. However, ab initio discovery of small proteins has been relatively overlooked. Recent advances in deep transcriptome sequencing make it possible to efficiently identify sORFs at the genome level. In this study, we obtained 2.6 million expressed sequence tag (EST) reads from Populus deltoides leaf transcriptome and reconstructed full-length transcripts from the EST sequences. We identified an initial set of 12,852 sORFs encoding proteins of 10 200 aa in length. Three computational approaches were then used to enrich for bona fide protein-coding sORFs from the initial sORF set: (1) codingpotential prediction, (2) evolutionary conservation between P. deltoides and other plant species, and (3) gene family clustering within P. deltoides. As a result, a high-confidence sORF candidate set containing 1469 genes was obtained. Analysis of the protein domains, non-protein-coding RNA motifs, sequence length distribution, and protein mass spectrometry data supported this high-confidence sORF set. In the high-confidence sORF candidate set, known protein domains were identified in 1282 genes (higher-confidence sORF candidate set), out of which 611 genes, designated as highest-confidence candidate sORF set, were supported by proteomics data. Of the 611 highest-confidence candidate sORF genes, 56 were new to the current Populus genome annotation. This study not only demonstrates that there are potential sORF candidates to be annotated in sequenced genomes, but also presents an efficient strategy for discovery of sORFs in species with no genome annotation yet available.

99 GENERAL AND MISCELLANEOUS↗

AI-enhanced Codesign for Next-Generation Neuromorphic Circuits and Systems

This report details work that was completed to address the Fiscal Year 2022 Advanced Science and Technology (AS&T) Laboratory Directed Research and Development (LDRD) call for “AI-enhanced Co-Design of Next Generation Microelectronics.” This project required concurrent contributions from the fields of 1) materials science, 2) devices and circuits, 3) physics of computing, and 4) algorithms and system architectures. During this project, we developed AI-enhanced circuit design methods that relied on reinforcement learning and evolutionary algorithms. The AI-enhanced design methods were tested on neuromorphic circuit design problems that have real-world applications related to Sandia’s mission needs. The developed methods enable the design of circuits, including circuits that are built from emerging devices, and they were also extended to enable novel device discovery. We expect that these AI-enhanced design methods will accelerate progress towards developing next-generation, high-performance neuromorphic computing systems.

42 ENGINEERING↗

An Evolutionary Algorithm to Personalize Stool-Based Colorectal Cancer Screening

Fecal immunochemical testing (FIT) is an established method for colorectal cancer (CRC) screening. Measured FIT-concentrations are associated with both present and future risk of CRC, and may be used for personalized screening. However, evaluation of personalized screening is computationally challenging. In this study, a broadly applicable algorithm is presented to efficiently optimize personalized screening policies that prescribe screening intervals and FIT-cutoffs, based on age and FIT-history. We present a mathematical framework for personalized screening policies and a bi-objective evolutionary algorithm that identifies policies with minimal costs and maximal health benefits. The algorithm is combined with an established microsimulation model (MISCAN-Colon), to accurately estimate the costs and benefits of generated policies, without restrictive Markov assumptions. The performance of the algorithm is demonstrated in three experiments. In Experiment 1, a relatively small benchmark problem, the optimal policies were known. The algorithm approached the maximum feasible benefits with a relative difference of 0.007%. Experiment 2 optimized both intervals and cutoffs, Experiment 3 optimized cutoffs only. Optimal policies in both experiments are unknown. Compared to policies recently evaluated for the USPSTF, personalized screening increased health benefits up to 14 and 4.3%, for Experiments 2 and 3, respectively, without adding costs. Generated policies have several features concordant with current screening recommendations. The method presented in this paper is flexible and capable of optimizing personalized screening policies evaluated with computationally-intensive but established simulation models. It can be used to inform screening policies for CRC or other diseases. For CRC, more debate is needed on what features a policy needs to exhibit to make it suitable for implementation in practice.

60 APPLIED LIFE SCIENCES↗

MOOSE ProbML: Parallelized probabilistic machine learning and uncertainty quantification for computational energy applications

Here, this paper presents the development and demonstration of massively parallel probabilistic machine learning (ML) and uncertainty quantification (UQ) capabilities within the Multiphysics Object-Oriented Simulation Environment (MOOSE), an open-source computational platform for parallel finite element and finite volume analyses. In addressing the computational expense and uncertainties inherent in complex multiphysics simulations, this paper integrates Gaussian process (GP) variants, active learning, Bayesian inverse UQ, adaptive forward UQ, Bayesian optimization, evolutionary optimization, and Markov chain Monte Carlo (MCMC) within MOOSE. It also elaborates on the interaction among key MOOSE systems — Sampler, MultiApp, Reporter, and Surrogate — in enabling these capabilities. The modularity offered by these systems enables development of a multitude of probabilistic ML and UQ algorithms in MOOSE. Example code demonstrations include parallel active learning and parallel Bayesian inference via active learning. The impact of these developments is illustrated through five applications relevant to computational energy applications: UQ of nuclear fuel fission product release, using parallel active learning Bayesian inference; very rare events analysis in nuclear microreactors using active learning; advanced manufacturing process modeling using multi-output GPs (MOGPs) and dimensionality reduction; fluid flow using deep GPs (DGPs); and tritium transport model parameter optimization for fusion energy, using batch Bayesian optimization. These capabilities are part of the MOOSE framework.

97 - MATHEMATICS AND COMPUTING↗

Poplar Genomics: State of the Science

Populus has become a model for genomics research in woody perennial plants due to the release of its genome sequence and the important roles it has been playing in research areas related to alternative energy and climate change. Over the past five years, enormous advances have been made in both experimental and computational genomics in Populus. In the area of Populus experimental genomics, traditional genetic approaches have been advanced to the genome scale with resolution to the gene and/or single nucleotide level on the one hand, and on the other hand, the modern omics approaches have been successfully applied to analyses of gene functions, including transcriptome profiling using microarrays as well as the next-generation DNA sequencing technology, proteome analysis using matrix assisted laser desorption ionization timeof-flight mass spectrometry, and metabolome characterization using gas chromatography/timeof-flight mass spectrometry. In the area of Populus computational genomics, significant progress has been made in sequence-based discovery of predicted gene function, comparative analysis of gene families, development of online genomic databases, and studies of the evolutionary dynamics at both the gene and genome level. While exciting achievements have been obtained in Populus genomics, several challenges need to be addressed in the future: 1) better annotation of the Populus genome; 2) robust technology for large-scale gene expression analysis; 3) an efficient system for genome-wide mutagenesis; and 4) high performance computational pipelines to keep up with the pace of the rapid accumulation of data and to integrate omics data into functional systems biology platforms.

54 ENVIRONMENTAL SCIENCES↗

The Effects of the pep Nuclear Reaction and Other Improvements in the Nuclear Reaction Rate Library on Simulations of the Classical Nova Outburst

Nova explosions occur on the white dwarf (WD) component of a cataclysmic variable binary stellar system which is accreting matter lost by its companion. When sufficient material has been accreted by the WD, a thermonuclear runaway (TNR) occurs and ejects material in what is observed as a classical nova (CN) explosion. We have continued our studies of TNRs on 1.25 M sun and 1.35 M sun WDs (ONeMg composition) under conditions which produce mass ejection and a rapid increase in the emitted light, by examining the effects of changes in the nuclear reaction rates on both the observable features and the nucleosynthesis during the outburst. In order to improve our calculations over previous work, we have incorporated a modern nuclear reaction network into our one-dimensional, fully implicit, hydrodynamic computer code. We find that the updates in the nuclear reaction rate libraries change the amount of ejected mass, peak luminosity, and the resulting nucleosynthesis. Because the evolutionary sequences on the 1.35 M sun WD reach higher temperatures, the effects of library changes are more important for this mass. In addition, as a result of our improvements, we discovered that the pep reaction (p + e - + p {yields} d + ?) was not included in our previous studies of CN explosions (or to the best of our knowledge those of other investigators). Although the energy production from this reaction is not important in the Sun, the densities in WD envelopes can exceed 104 g cm-3 and the presence of this reaction increases the energy generation during the time that the p-p chain is operating. Since it is only the p-p chain that is operating during most of the accretion phase prior to the final rise to the TNR, the effect of the increased energy generation is to reduce the evolution time to the peak of the TNR and, thereby, the accreted mass as compared to the evolutionary sequences done without this reaction included. As expected from our previous work, the reduction in accreted mass has important consequences on the characteristics of the resulting TNR and is discussed in this paper.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗