Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Cluster computing”

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 1,333 records · Page 74
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Big Data and AI at DoE's Legacy Sites - 20546

More than 30 years have passed since DOE started the decommissioning of nuclear weapon complexes and the clean-up of soil and groundwater. All the sites have been collecting and archiving soil and groundwater monitoring datasets; particularly contaminant concentration time-series. These datasets provide unparalleled opportunities to understand the system behavior (including more fundamental hydrological and geochemical processes, the response to various perturbations, the long-term trend and environmental decay rate towards the regulatory limit). This understanding is critical for providing multiple lines of evidences that can support site closure. In this study, we explore the machine learning (ML) and artificial intelligence (AI) applications to the long-term soil and groundwater management at DoE's legacy sites. ML can improve our understanding of the subsurface systems, which is critical for long-term monitoring and management of the sites, while AI can automate or support some of decision-making processes (e.g., anomaly detection, monitoring well placements). The particular focuses are to develop general algorithms to: (1) to identify distinct spatiotemporal patterns and to identify several groups that have similar temporal behaviors, using unsupervised clustering methods, (2) identify the different temporal scales of hydrological responses to climate perturbations by time-series analysis, and (3) reduce the number of monitoring wells by identifying the minimum sufficient number of wells to capture the heterogeneity of the groundwater contaminant plume and concentration distribution, using the Gaussian Process model. We demonstrate our methodology at the Savannah River Site F-Area. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2025

Net Present Value Optimization of a Natural Gas Combined Cycle Plant with CO 2 Capture using a Water-Lean Solvent Considering Transient Electricity Price for Multiple Regions

Global CO 2 emissions are increasing at about a 1.5% rate per year. Fossil fuel-based plants are one of the main contributors to this rise. In the power generation industry, fossil fuel plants are dominant, and many plants are under development. In this study, a natural gas combined cycle (NGCC) power plant with postcombustion capture using a leading water-lean solvent is considered. For optimal design and operating schedule, large-scale dynamic optimization is undertaken for net present value (NPV) optimization. The first principle dynamic model of NGCC is developed, including a model of the highly efficient H-class gas turbines. For computational tractability of the dynamic optimization problem, a reduced-order model is developed by using the Hankel singular value decomposition. A waterlean solvent, N-(2-ethoxyethyl)-3-morpholinopropan-1-amine, is used for carbon capture. A model of the capture system is developed in Aspen Plus, which is used to develop a reduced-order model by using ALAMO, a machine learning software. In addition, a reduced model of the CO 2 compression system with a dehydration unit is also considered. The integrated system is used for NPV optimization by using the Python-based PYOMO platform. The PCC process is analyzed for three configurations-conventional packed bed, rotating packed bed (RPB), and a combination of RPB and direct contact cooler. The NPV optimization is performed for 14 regional markets by considering year-long clustered and continuous locational marginal price data with a 1 h interval. Optimization results show that the PCC can achieve 90% CO 2 capture with a positive NPV for six regions. Sensitivity studies conducted by using the PCC configurations indicate that the process is economically feasible for 9 regions out of 14 regional electricity markets with NPV values in the range of 33−540 $MM.

cabon capture↗
DOE OSTI2025

UnigeneFinder: An Automated Pipeline for Gene Calling From Transcriptome Assemblies Without a Reference Genome

ABSTRACT For most species, transcriptome data are much more readily available than genome data. Without a reference genome, gene calling is cumbersome and inaccurate because of the high degree of redundancy in de novo transcriptome assemblies. To simplify and increase the accuracy of de novo transcriptome assembly in the absence of a reference genome, we developed UnigeneFinder. Combining several clustering methods, UnigeneFinder substantially reduces the redundancy typical of raw transcriptome assemblies. This pipeline offers an effective solution to the problem of inflated transcript numbers, achieving a closer representation of the actual underlying genome. UnigeneFinder performs comparably or better, compared with existing tools, on plant species with varying genome complexities. UnigeneFinder is the only available transcriptome redundancy solution that fully automates the generation of primary transcript, coding region, and protein sequences, analogous to those available for high‐quality reference genomes. These features, coupled with the pipeline’s cross‐platform implementation, focus on automation, and an accessible, user‐friendly interface, make UnigeneFinder a useful tool for many downstream sequence‐based analyses in nonmodel organisms lacking a reference genome, including differential gene expression analysis, accurate ortholog identification, functional enrichments, and evolutionary analyses. UnigeneFinder also runs efficiently both on high‐performance computing (HPC) systems and personal computers, further reducing barriers to use.

Xue, Bo [Plant Resilience Institute Michigan State↗
DOE OSTI2021

Elaborated thermochemical treatment of HF, CO, N 2 , and H 2 O: Insight into HEAT and its extensions

Here, empirical, highly accurate non-relativistic electronic total atomization energies (eTAEs) are established by combining experimental or computationally converged treatments of the nuclear motion and relativistic contributions with the total atomization energies of HF, CO, N 2 , and H 2 O obtained from the Active Thermochemical Tables. These eTAEs, which have estimated (2σ) uncertainties of less than 10 cm -1 (0.12 kJ mol -1 ), form the basis for an analysis of high-level ab initio quantum chemical calculations that aim at reproducing these eTAEs for the title molecules. The results are then employed to analyze the performance of the high-accuracy extrapolated ab initio thermochemistry, or High-Accuracy Extrapolated Ab Initio Thermochemistry (HEAT), family of theoretical methods. The method known as HEAT-345(Q), in particular, is found to benefit from fortuitous error cancellation between its treatment of the zero-point energy, extrapolation errors in the Hartree-Fock and coupled cluster contributions, neglect of post-(T) core-correlation, and the basis-set error involved in higher-level correlation corrections. In addition to shedding light on a longstanding curiosity of the HEAT protocol—where the cheapest HEAT-345(Q) performs comparably to the theoretically more complete HEAT-456QP procedure—this study lays the foundation for extended HEAT variants that offer substantial improvements in accuracy relative to the established approaches.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Dial-A-Cluster User Manual

The Dial-A-Cluster (DAC) model allows interactive visualization of multivariate time series data. A multivariate time series dataset consists of an ensemble of data points, where each data point consists of a set of time series curves. The example of a DAC dataset used in this guide is a collection of 100 cities in the United States, where each city collects a year's worth of weather data, including daily temperature, humidity, and wind speed measurements.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Elucidating size effects on the yield strength of single-crystal Cu via the Richtmyer–Meshkov instability

Capturing the dynamic response of a material under high strain-rate deformation often demands challenging and time consuming experimental effort. While shock hydrodynamic simulation methods can aid in this area, a priori characterizations of the material strength under shock loading and spall failure are needed in order to parameterize constitutive models needed for these computational tools. Moreover, parameterizations of strain-rate-dependent strength models are needed to capture the full suite of Richtmyer–Meshkov instability (RMI) behavior of shock compressed metals, creating an unrealistic demand for these training data solely on experiments. Herein, we sweep a large range of geometric, crystallographic, and shock conditions within molecular dynamics (MD) simulations and demonstrate the breadth of RMI in Cu that can be captured from the atomic scale. In this work, yield strength measurements from jetted and arrested material from a sinusoidal surface perturbation were quantified as Y RMI = 0.787 ± 0.374 GPa, higher than strain-rate-independent models used in experimentally matched hydrodynamic simulations. Defect-free, single-crystal Cu samples used in MD will overestimate Y RMI , but the drastic scale difference between experiment and MD is highlighted by high confidence neighborhood clustering predictions of RMI characterizations, yielding incorrect classifications.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2020

Visualization of the HIV-1 Env glycan shield across scales

The dense array of N-linked glycans on the HIV-1 envelope glycoprotein (Env), known as the “glycan shield,” is a key determinant of immunogenicity, yet intrinsic heterogeneity confounds typical structure–function analysis. Here, we present an integrated approach of single-particle electron cryomicroscopy (cryo-EM), computational modeling, and site-specific mass spectrometry (MS) to probe glycan shield structure and behavior at multiple levels. We found that dynamics lead to an extensive network of interglycan interactions that drive the formation of higher-order structure within the glycan shield. This structure defines diffuse boundaries between buried and exposed protein surface and creates a mapping of potentially immunogenic sites on Env. Analysis of Env expressed in different cell lines revealed how cryo-EM can detect subtle changes in glycan occupancy, composition, and dynamics that impact glycan shield structure and epitope accessibility. Importantly, this identified unforeseen changes in the glycan shield of Env obtained from expression in the same cell line used for vaccine production. Finally, by capturing the enzymatic deglycosylation of Env in a time-resolved manner, we found that highly connected glycan clusters are resistant to digestion and help stabilize the prefusion trimer, suggesting the glycan shield may function beyond immune evasion.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2022

Fast emulation of quantum three-body scattering

Here, we develop a class of emulators for solving quantum three-body scattering problems. They are based on combining the variational method for scattering observables and the recently proposed eigenvector continuation concept. The emulators are first trained by the exact scattering solutions of the governing Hamiltonian at a small number of points in its parameter space, and then employed to make interpolations and extrapolations in that space. Through a schematic nuclear-physics model with finite-range two and three-body interactions, we demonstrate the emulators to be extremely accurate and efficient. The computing time for emulation is on the scale of milliseconds (on a laptop), with relative errors ranging from 10 –13 to 10 –4 depending on the case. The emulators also require little memory. We argue that these emulators can be generalized to even more challenging scattering problems. Furthermore, this general strategy may be applicable for building the same type of emulators in other fields, wherever variational methods can be developed for evaluating physical models.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

GPU acceleration of rank-reduced coupled-cluster singles and doubles

Here, we have developed a graphical processing unit (GPU) accelerated implementation of our recently introduced rank-reduced coupled-cluster singles and doubles (RR-CCSD) method. RR-CCSD introduces a low-rank approximation of the doubles amplitudes. This is combined with a low-rank approximation of the electron repulsion integrals via Cholesky decomposition. The result of these two low-rank approximations is the replacement of the usual fourth-order CCSD tensors with products of second- and third-order tensors. In our implementation, only a single fourth-order tensor must be constructed as an intermediate during the solution of the amplitude equations. Owing in large part to the compression of the doubles amplitudes, the GPU-accelerated implementation shows excellent parallel efficiency (95% on eight GPUs). Our implementation can solve the RR-CCSD equations for up to 400 electrons and 1550 basis functions—roughly 50% larger than the largest canonical CCSD computations that have been performed on any hardware. In addition to increased scalability, the RR-CCSD computations are faster than the corresponding CCSD computations for all but the smallest molecules. We test the accuracy of RR-CCSD for a variety of chemical systems including up to 1000 basis functions and determine that accuracy to better than 0.1% error in the correlation energy can be achieved with roughly 95% compression of the ov space for the largest systems considered. We also demonstrate that conformational energies can be predicted to be within 0.1 kcal mol -1 with efficient compression applied to the wavefunction. Finally, we find that low-rank approximations of the CCSD doubles amplitudes used in the similarity transformation of the Hamiltonian prior to a conventional equation-of-motion CCSD computation will not introduce significant errors (on the order of a few hundredths of an electronvolt) into the resulting excitation energies.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Density weighted angular redshift fluctuations: a new cosmological observable

We propose the use of angular fluctuations in the galaxy redshift field as a new way to extract cosmological information in the Universe. This new probe $\delta z (\hat{\mathbf {n}})$ consists of the statistics of sky maps built by projecting redshifts under a Gaussian window of width σz centred upon a redshift zobs, and weighted by the galaxy density field. We compute the angular power spectrum of the $\delta z (\hat{\mathbf {n}})$ field in both numerical simulations and in linear perturbation theory. From these, we find that the $\delta z (\hat{\mathbf {n}})$ field (i) is sensitive to the underlying density and peculiar velocity fields; (ii) is highly correlated, at the $\gtrsim 60\, {{\ \rm per\ cent}}$ level, to the line-of-sight projected peculiar velocity field; (iii) for narrow windows (σz < 0.03), it is almost completely uncorrelated to the projected galaxy angular density field under the same redshift window; and (iv) it is largely unaffected by multiplicative and additive systematic errors on the observed number of galaxies that are redshift-independent over ~σz. We conclude that $\delta z (\hat{\mathbf {n}})$ is a simple and robust tomographic measure of the cosmic density and velocity fields, complementary to angular clustering, that will contribute to more complete exploitations of current and upcoming galaxy redshift surveys.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

The Three Hundred Project: Modeling baryon and hot-gas fraction evolution in simulated clusters

The baryon fraction of galaxy clusters, expressed as the ratio between the mass in baryons (including both stars and cold or hot gas) and the total mass, is a powerful tool to provide information on the cosmological parameters, while the hot-gas fraction provides indications on the physics of the intracluster plasma and its interplay with the processes that drive galaxy formation. Using cosmological hydrodynamical simulations of about 300 simulated massive galaxy clusters with a median mass M 500 ≈ 7 × 10 14 M ⊙ at z = 0, we model the relations between total mass and either baryon fraction or the hot gas fractions at overdensities Δ = 2500, 500, and 200 with respect to the cosmic critical density, and their evolution from z ∼ 0 to z ∼ 1.3. We utilized the simulated galaxy clusters from the Three Hundred project, which include star formation and feedback from both supernovae and active galactic nuclei. We fit the simulation results for such scaling relations against three analytic forms (linear, quadratic, and logarithmic in a logarithmic plane) and three forms for the redshift dependence, and we considered as a variable both the inverse of the cosmic scale factor, (1 + z), and the Hubble expansion rate, E(z). We show that power-law dependencies on cluster mass poorly describe the investigated relations. A power law fails to simultaneously capture the flattening of the total baryon and gas fractions at high masses, their drop at low masses, and the transition between these two regimes. The other two functional forms provide a more accurate description of the curvature in mass scaling. The fractions measured within smaller radii exhibit a stronger evolution than those measured within larger radii. From the analysis of these simulations, we evince that as long as we include systems in the mass range herein investigated, the baryon or gas fraction can be accurately related to the total mass through either a parabola or a logarithm in the logarithmic plane. The trends are common to all modern hydro simulations, although the amplitude of the drop at low masses might differ. Being able to observationally determine the gas fraction in groups will thus provide constraints on the baryonic physics.

galaxy clusters↗
DOE OSTI2023

Clustering Acoustic Background Noise in the Stratosphere Using Machine Learning

Infrasound, characterized by low-frequency sound inaudible to humans (<20 Hz), emanates from natural and anthropogenic sources. Its efficacy for monitoring phenomena necessitates robust sensing networks. Traditional ground-based infrasound sensors have limitations due to atmospheric dynamics and noise interference. Balloon-bore sensors have emerged as an alternative, offering reduced noise and improved capabilities. This study bridges clustering algorithms with balloon borne infrasound data, a domain yet to be explored. Employing K-Means, DBSCAN, and GMM algorithms on normalized and reshaped data and only normalized data from a New Zealand-based NASA balloon flight, insights into background noise at stratospheric altitudes were revealed. Despite challenges arising from distinguishing signals amid unique background noise, this research provides vital reference material for noise analysis and calibration. Beyond infrasound event capture, the dataset enriches comprehension of background noise characteristics in the southern hemisphere.

47 OTHER INSTRUMENTATION↗
DOE OSTI2024

Are Marine Low Cloud Droplet Concentrations Buffered by Entrained Aitken‐Mode Aerosol (Final Technical Report)

During the summertime, the high-latitude oceans come to life with green phytoplankton, which gain their energy from sunlight and are food for sea creatures small and large. When the phytoplankton are eaten or die, sulfur-rich gases are released under the ocean surface and mix into the air. Observations suggest that, over the Southern Ocean, frequent storms lift this air high into the atmosphere while raining out particulates like salt. As a result, the sulfur-rich air then spawns high concentrations of small ‘Aitken-mode’ aerosol particles. We hypothesize that these particles work their way down into the marine boundary layer, where they can replenish the supply of cloud-condensation nuclei scavenged by frequent precipitation. Further, this process maintains high concentrations of liquid cloud droplets in austral summer, promoting more sunlight to be reflected to space. We call this ‘Aitken buffering’. The primary objective of this project has been to document and test what role Aitken-mode aerosols play in clouds over the Southern Ocean and elsewhere. This effort has included three main components: 1) developing a computer model that realistically simulates the aerosol processes and the small-scale turbulent air motions that move aerosols around and create the clouds, 2) using that model to interpret and extend these observations for process understanding, by allowing different factors that contribute to the aerosol budget, such as surface wind speed, precipitation, surface gas exchange, etc. to be separated, and 3) studying Aitken-mode aerosol and its variability with a focus over the Southern Ocean and Antarctica, using data from Atmospheric Radiation Measurement (ARM) sites and other available observations. Initial computer studies in more idealized conditions found that elevated concentrations of Aitken-mode aerosols above the clouds could help prevent the breakup of those clouds by acting as cloud-condensation nuclei after they were entrained into the cloudy boundary layer. Simulations of a day during the ACE-ENA field campaign showed that Aitken-mode aerosols could also prevent cloud breakup under more realistic conditions. A new method that extracts information about Aitken-mode aerosols from measurements of aerosols onto which cloud droplets can form finds that Aitken-mode aerosols do vary seasonally over the Southern Ocean, with a peak in summertime, as described above. Other work during this project has focused on understanding how patterns of water vapor, clouds and precipitation are coupled within low-lying clouds over the oceans, and also on how cloud droplets cluster within clouds and how the distribution of cloud droplet sizes change as dry air is mixed into clouds, with the latter studies also using observations from ACE-ENA.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

CYPminer: an automated cytochrome P450 identification, classification, and data analysis tool for genome data sets across kingdoms

Background: Cytochrome P450 monooxygenases (termed CYPs or P450s) are hemoproteins ubiquitously found across all kingdoms, playing a central role in intracellular metabolism, especially in metabolism of drugs and xenobiotics. The explosive growth of genome sequencing brings a new set of challenges and issues for researchers, such as a systematic investigation of CYPs across all kingdoms in terms of identification, classification, and pan-CYPome analyses. Such investigation requires an automated tool that can handle an enormous amount of sequencing data in a timely manner. Results: CYPminer was developed in the Python language to facilitate rapid, comprehensive analysis of CYPs from genomes of all kingdoms. CYPminer consists of two procedures i) to generate the Genome-CYP Matrix (GCM) that lists all occurrences of CYPs across the genomes, and ii) to perform analyses and visualization of the GCM, including pan-CYPomes (pan- and core-CYPome), CYP co-occurrence networks, CYP clouds, and genome clustering data. The performance of CYPminer was evaluated with three datasets from fungal and bacterial genome sequences. Conclusions: CYPminer completes CYP analyses for large-scale genomes from all kingdoms, which allows systematic genome annotation and comparative insights for CYPs. CYPminer also can be extended and adapted easily for broader usage.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2023

Computing the Properties of Matter with Leadership Computing Resources (Closeout Report for DE-SC0018121)

In order to add more capabilities to Halide, we have designed a new framework called Tiramisu and integrated this framework into Halide. Since Tiramisu enables Halide to target heterogeneous architectures, our development efforts have been refocused on Tiramisu. Most high-performance computer systems today are complex and increasingly heterogeneous; they may have CPUs, GPUs and FPGAs. Achieving best performance requires taking full advantage of all these different architectures. To address this issue, we have designed Tiramisu, an optimization framework that enables Halide (and other DSLs) to target heterogeneous architectures. Tiramisu is an optimization framework that takes as input a high level, architecture-independent representation of code and a set of scheduling and data mapping commands that guide code transformation. The input can either be generated by a domain-specific language (DSL) compiler such as Halide or directly written by a programmer. Tiramisu then applies the user-specified code and data-layout transformations and generates an architecture-specific, low-level intermediate representation (IR) that takes advantage of modern architectural features such as multicore parallelism, non-uniform memory (NUMA) hierarchies, clusters, and accelerators like GPUs and FPGAs. We integrated Tiramisu within Halide and implemented a representative set of benchmarks to evaluate this integration. Tiramisu is now open source and is available for public use (http://tiramisu-compiler.org/). A paper about Tiramisu was published, it shows that Tiramisu extends Halide with many new capabilities and that Tiramisu can generate efficient code for multicores, GPUs, FPGAs and distributed heterogeneous systems. The performance of code generated by the Tiramisu backends matches or exceeds hand optimized reference implementations. For example, the multicore backend matches the highly optimized Intel MKL library on many kernels and shows speedups reaching 4x over the original Halide. In addition to making Tiramisu more robust, we have used Tiramisu to implement a set of representative tensor operation for constructing baryon building blocks required for multi baryon contractions in LQCD. In order to implement this code, we needed to generalize Tiramisu in two ways: first we needed to support indirect array accesses, and second, we needed to add support for complex numbers to Tiramisu. The code generated by Tiramisu is 6x faster than the reference code. Our efforts towards an MPI based multi-node version of tiramisu have matured and the resulting code scales well on multiple nodes (tests up to 512 KNL nodes have been undertaken).

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2020

Studying the effects of overlapping objects in dark energy

Observing the clustering of galaxies allows us to calculate cosmological parameters necessary for understanding dark energy. However, as the density of observed objects increases, the probability of these objects blending likewise increases, causing multiple galaxies to be observed as one. This affects the inferred values of parameters such as the galaxy bias (b) and the matter energy density (ΩM). To see whether the bias from incorrectly inferring the galaxy count is significant, we compare the correlation functions in simulated data for “true” and “observed” data sets with one-to-one and multiple-to-one correspondences, respectively. For each data set, we create two correlation functions: one “measured” function directly relying on the galaxies’ positions using the TreeCorr python library, and one “model” derived mathematically from the galaxies’ power spectrum using the Cosmological Core Library (CCL). By minimizing the residual between these two functions, we compute the ideal values for b and ΩM across the various possible redshifts that position the galaxies in three dimensional space. This minimization is done with an Markov chain Monte Carlo (MCMC) estimate that finds one value of ΩM and ten values for b corresponding to the ten redshift bins ranging from z = 0.2 to z = 1.2. We find that neither b nor ΩM is particularly affected by inclusion of blended galaxies. Though there is room for improvement, the data suggests that the fluctuations we found are a result of noise or limitations on the modeling rather than blending explicitly.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

ROOT RNTuple and EOS: The Next Generation of Event Data I/O

For several years, the ROOT team is developing the new RNTuple I/O subsystem in preparation of the next generation of collider experiments. Both HL-LHC and DUNE are expected to start data taking by the end of this decade. They pose unprecedented challenges to event data I/O in terms of data rates, event sizes, and event complexity. At the same time, the I/O landscape is becoming more diverse. HPC cluster file systems and object stores, NVMe disk cache layers in analysis facilities, and S3 storage on cloud resources are mixing with traditional XRootD-managed spinning disk pools.The ROOT team will finalize a first production version of the RNTuple binary format by the end of 2024. After this point, ROOT will provide backward compatibility for RNTuple data. This contribution provides an overview of the RNTuple feature set, the related R&D activities and the long-term vision for RNTuple. We report on performance, interface design, tooling, robustness, integration with experiment frameworks, and validation results, as well as recent R&D on parallel reading and writing and exploitation of modern hardware and storage systems. We will give an outlook on possible future features after a first production release.Collaboratively, the IT and EP departments at CERN have launched a formal project within the Research and Computing sector to evaluate the novel data format for physics analysis data utilized in LHC experiments and other fields. This part of the project focuses on validating the scalability of the EOS storage backend during the transition from the over 25 years old TTree production format to the newly developed RNTuple format, using both replicated and erasure-coded storage profiles.

Blomer, Jakob [CERN]↗
DOE OSTI2020

Rapid Identification of X-ray Diffraction Patterns Based on Very Limited Data by Interpretable Convolutional Neural Networks

Large volumes of data from material characterizations call for rapid and automatic data analysis to accelerate materials discovery. Herein, we report a convolutional neural network (CNN) that was trained based on theoretical data and very limited experimental data for fast identification of experimental X-ray diffraction (XRD) patterns of metal–organic frameworks (MOFs). To augment the data for training the model, noise was extracted from experimental data and shuffled; then it was merged with the main peaks that were extracted from theoretical spectra to synthesize new spectra. For the first time, one-to-one material identification was achieved. Theoretical MOFs patterns (1012) were augmented to a whole data set of 72 864 samples. It was then randomly shuffled and split into training (58 292 samples) and validation (14 572 samples) data sets at a ratio of 4:1. For the task of discriminating, the optimized model showed the highest identification accuracy of 96.7% for the top 5 ranking on a test data set of 30 hold-out samples. Neighborhood component analysis (NCA) on the experimental XRD samples shows that the samples from the same material are clustered in groups in the NCA map. Analysis on the class activation maps of the last CNN layer further discloses the mechanism by which the CNN model successfully identifies individual MOFs from the XRD patterns. Furthermore, this CNN model trained by the data augmentation technique would not only open numerous potential applications for identifying XRD patterns for different materials, but also pave avenues to autonomously analyze data by other characterization tools such as FTIR, Raman, and NMR spectroscopies.

36 MATERIALS SCIENCE↗