Search NASA⌕ Search

SEARCH · Search NASA

Results for “subset sampling”

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 703 records · Page 39

Developing Natural Language Processing and Supervised Learning Techniques to Classify Mars Tasks

As NASA's Human Research Program (HRP) prepares for long-duration Mars missions, understanding astronaut tasks is crucial. This study, conducted at NASA Glenn Research Center (GRC), employed Natural Language Processing (NLP) and machine learning techniques to analyze and classify Mars tasks. A list of 1,058 Mars tasks was provided by HRP experts including binary labeling of 18 Human System Task Categories (HSTCs). We developed an NLP model using Google's BERT language model to capture the semantic and syntactic nuances of these tasks. Supervised training was initially applied to a subset of the NLP-analyzed tasks to assess the model's effectiveness in classifying the remaining tasks. Incorporating HSTC descriptions significantly enhanced the classification accuracy for 9 out of the 18 HSTCs and reduced training time. To address the issue of severe class imbalance in the HSTC data, we introduced innovative weighting and sampling techniques for data augmentation. We then fine-tune BERT to implement a pairwise relatedness scoring method, allowing us to cluster tasks based on their relatedness and similarity, getting a step closer to labeling the tasks without supervision. In this presentation we guide you through data preprocessing, deciphering key syntax components using BERT, and performing supervised classification of the Mars tasks. This work showcases the potential use of advanced NLP techniques to analyze Mars missions to be incorporated into various crew health and performance analyses.

GenAI↗

THE SPECTRAL ENERGY DISTRIBUTIONS AND INFRARED LUMINOSITIES OF z Almost-Equal-To 2 DUST-OBSCURED GALAXIES FROM Herschel AND Spitzer

Dust-obscured galaxies (DOGs) are a subset of high-redshift (z Almost-Equal-To 2) optically-faint ultra-luminous infrared galaxies (ULIRGs, e.g., L{sub IR} > 10{sup 12} L{sub Sun} ). We present new far-infrared photometry, at 250, 350, and 500 {mu}m (observed-frame), from the Herschel Space Telescope for a large sample of 113 DOGs with spectroscopically measured redshifts. Approximately 60% of the sample are detected in the far-IR. The Herschel photometry allows the first robust determinations of the total infrared luminosities of a large sample of DOGs, confirming their high IR luminosities, which range from 10{sup 11.6} L{sub Sun} 10{sup 13} L{sub Sun }. The rest-frame near-IR (1-3 {mu}m) spectral energy distributions (SEDs) of the Herschel-detected DOGs are predictors of their SEDs at longer wavelengths. DOGs with 'power-law' SEDs in the rest-frame near-IR show observed-frame 250/24 {mu}m flux density ratios similar to the QSO-like local ULIRG, Mrk 231. DOGs with a stellar 'bump' in their rest-frame near-IR show observed-frame 250/24 {mu}m flux density ratios similar to local star-bursting ULIRGs like NGC 6240. None show 250/24 {mu}m flux density ratios similar to extreme local ULIRG, Arp 220; though three show 350/24 {mu}m flux density ratios similar to Arp 220. For the Herschel-detected DOGs, accurate estimates (within {approx}25%) of total IR luminosity can be predicted from their rest-frame mid-IR data alone (e.g., from Spitzer observed-frame 24 {mu}m luminosities). Herschel-detected DOGs tend to have a high ratio of infrared luminosity to rest-frame 8 {mu}m luminosity (the IR8 = L{sub IR}(8-1000 {mu}m)/{nu}L{sub {nu}}(8 {mu}m) parameter of Elbaz et al.). Instead of lying on the z = 1-2 'infrared main sequence' of star-forming galaxies (like typical LIRGs and ULIRGs at those epochs) the DOGs, especially large fractions of the bump sources, tend to lie in the starburst sequence. While, Herschel-detected DOGs are similar to scaled up versions of local ULIRGs in terms of 250/24 {mu}m flux density ratio, and IR8, they tend to have cooler far-IR dust temperatures (20-40 K for DOGs versus 40-50 K for local ULIRGs) as measured by the rest-frame 80/115 {mu}m flux density ratios (e.g., observed-frame 250/350 {mu}m ratios at z = 2). DOGs that are not detected by Herschel appear to have lower observed-frame 250/24 {mu}m ratios than the detected sample, either because of warmer dust temperatures, lower IR luminosities, or both.

79 ASTRONOMY AND ASTROPHYSICS↗

Root Pulling Force Across Drought in Maize Reveals Genotype by Environment Interactions and Candidate Genes

High-throughput, field-based characterization of root systems for hundreds of genotypes in thousands of plots is necessary for breeding and identifying loci underlying variation in root traits and their plasticity. We designed a large-scale sampling of root pulling force, the vertical force required to extract the root system from the soil, in a maize diversity panel under differing irrigation levels for two growing seasons. We then characterized the root system architecture of the extracted root crowns. We found consistent patterns of phenotypic plasticity for root pulling force for a subset of genotypes under differential irrigation, suggesting that root plasticity is predictable. Using genome-wide association analysis, we identified 54 SNPs as statistically significant for six independent root pulling force measurements across two irrigation levels and four developmental timepoints. For every significant GWAS SNP for any trait in any treatment and timepoint we conducted post hoc tests for genotype-by-environment interaction, using a mixed model ANOVA. We found that 8 of the 54 SNPs showed significant GxE. Candidate genes underlying variation in root pulling force included those involved in nutrient transport. Although they are often treated separately, variation in the ability of plant roots to sense and respond to variation in environmental resources including water and nutrients may be linked by the genes and pathways underlying this variation. While functional validation of the identified genes is needed, our results expand the current knowledge of root phenotypic plasticity at the whole plant and gene levels, and further elucidate the complex genetic architecture of maize root systems.

Woods, Patrick↗

Enigmatic, ultrasmall, uncultivated Archaea

Metagenomics has provided access to genomes of as yet uncultivated microorganisms in natural environments, yet there are gaps in our knowledge-particularly for Archaea-that occur at relatively low abundance and in extreme environments. Ultrasmall cells (<500 nm in diameter) from lineages without cultivated representatives that branch near the crenarchaeal/euryarchaeal divide have been detected in a variety of acidic ecosystems. We reconstructed composite, near-complete similar to 1-Mb genomes for three lineages, referred to as ARMAN (archaeal Richmond Mine acidophilic nanoorganisms), from environmental samples and a biofilm filtrate. Genes of two lineages are among the smallest yet described, enabling a 10% higher coding density than found genomes of the same size, and there are noncontiguous genes. No biological function could be inferred for up to 45% of genes and no more than 63% of the predicted proteins could be assigned to a revised set of archaeal clusters of orthologous groups. Some core metabolic genes are more common in Crenarchaeota than Euryarchaeota, up to 21% of genes have the highest sequence identity to bacterial genes, and 12 belong to clusters of orthologous groups that were previously exclusive to bacteria. A small subset of 3D cryo-electron tomographic reconstructions clearly show penetration of the ARMAN cell wall and cytoplasmic membranes by protuberances extended from cells of the archaeal order Thermoplasmatales. Interspecies interactions, the presence of a unique internal tubular organelle [Comolli, et al. (2009) ISME J 3: 159-167], and many genes previously only affiliated with Crenarchaea or Bacteria indicate extensive unique physiology in organisms that branched close to the time that Cren- and Euryarchaeotal lineages diverged.

59 BASIC BIOLOGICAL SCIENCES↗

Revisiting the Origin of Macromolecular Carbon (MMC) in Lunar Basalts 15556 & 10044

Volatile elements influence the geo-chemical evolution of planetary bodies and they are in magmas at every stage, from melting within planetary interiors to eruption at the surface. Analyses of lunar mare basalts supported the hypothesis that lunar mag-mas were depleted in volatiles (H-C-F-Cl-S), relative to their terrestrial analogs [1]. Nevertheless, several early studies of samples returned during the Apollo program proposed that the mare basalt eruptions, including the “fire fountain” eruptions, were propelled by the oxidation of magmatic graphite to CO (and/or CO2) gas [2, 3]. Seminal studies during the 1970’s measured the bulk concentration and isotopic compositions of C from Apollo 11 samples, and identified several carbonaceous compounds, including: (a) gaseous (CO, CO2, and traces of CH4), (b) metallic carbide, and (c) potentially elemental carbon [4-6]. These studies reported a relatively broad range of C contents (~100-400 μg/g) and isotopic values (δC13 = -30 to +20), and suggested that these heterogeneities can be explained by contribution from multiple factors, including: (a) indigenous carbon, (b) solar wind implantation, (c) bombardment and/or meteorite impact, and (d) terrestrial contamination [5,6]. However, unequivocal observations of magmatic graphite in lunar basalts have never been made. Macromolecular carbon (MMC)—graphitic carbon varying from nearly amorphous to highly crystalline varieties—was identified as inclusions hosted by igneous pyroxenes from Martian meteorites and were attributed to being indigenous to Mars [8]. The authors carefully considered the textural and mineralogical relationship of the MMC phases, and concluded that the subset of MMC located within and/or adjacent to cracks, or at a disrupted surface (e.g., cut) were most consistent with terrestrial contamination. The near absence of con-firmed instances of lunar magmatic MMC within the literature [9], combined with the wide range of isotopic values and bulk carbon contents measured in lunar bas-alts begs the question as to whether previously measured carbon is of an indigenous origin, or the result of terrestrial contamination. Using Raman spectroscopy, we have observed MMC in lunar basalts subjected to different forms of anthropogenic modification related to sample preparation including polished sections, sawn surfaces, and fractured surfaces adjacent to sawn sur-faces. We have observed MMC of unknown origin in all of these settings. Here we report the preliminary textural and spectroscopic characteristics of MMC hosted within the groundmass of Apollo 15 (15556) and Apollo 11 basalts (10044) as part of our ongoing investigation of the origin of these carbonaceous materials.

lunar↗

The SOUX AGN sample: SDSS– XMM-Newton optical, ultraviolet, and X-ray selected active galactic nuclei spanning a wide range of parameter space – sample definition

Abstract We assemble a sample of 696 type 1 active galactic nuclei (AGN) up to a redshift of z = 2.5, all of which have an SDSS spectrum containing at least one broad emission line (H α, H β, or Mg ii) and an XMM-Newton X-ray spectrum containing at least 250 counts in addition to simultaneous optical/ultraviolet photometry from the XMM Optical Monitor. Our sample includes quasars and narrow-line Seyfert 1s: thus our AGN span a wide range in luminosity, black hole mass, and accretion rate. We determine single-epoch black hole mass relations for the three emission lines and find that they provide broadly consistent mass estimates whether the continuum or emission line luminosity is used as the proxy for the broad emission line region radius. We explore variations of the UV/X-ray energy index αox with the UV continuum luminosity and with black hole mass and accretion rate, and make comparisons to the physical quasar spectral energy distribution model qsosed. The majority of the AGN in our sample lie in a region of parameter space with 0.02 < L/LEdd < 2 as defined by this model, with narrow-line type 1 AGN offset to lower masses and higher accretion rates than typical broad-line quasars. We find differences in the dependence of αox on UV luminosity between both narrow/broad-line and radio-loud/quiet subsets of AGN: αox has a slightly weaker dependence on UV luminosity for broad-line AGN and radio-loud AGN have systematically harder αox.

79 ASTRONOMY AND ASTROPHYSICS↗

Examining the Effects of 4He Exposure on the Gut-Brain Axis

Beyond low earth orbit, space radiation poses significant risks to astronaut health. Previous studies have shown that the microbial composition of the gastrointestinal (GI) microbiome changes upon exposure to high linear energy transfer radiation. Interestingly, radiation-induced shifts in GI microbiota composition are linked to various neuropsychological disorders. Herein, we aimed to study changes in GI microbiota and behaviors of rats exposed to whole-body radiation (0, 5, or 25 cGy 4 He, 250 MeV/n) at approximately 6 months of age. Fecal samples were collected 24 hours prior to 4 He radiation and 24 hours and 7 days post-exposure for quantitative PCR analyses to assess fecal levels of spore-forming bacteria (SFB), Bifidobacterium, Lactobacillus, and Akkermansia. Rats were also tested in the social odor recognition memory (SORM) test at 7 days following 4 He exposure. A subset of rats was euthanized 90 minutes following completion of the SORM test, and GI tissue from small intestine to colon were prepared for examining overall histological changes and immunohistochemical staining for serotonin (5-HT). No notable pathological changes were observed in GI tissues. Akkermansia spp. and SFB were significantly decreased the 25 cGy group at 24 hours and 7 days post-exposure compared to pre-exposure, respectively. Bifidobacterium and Lactobacillus spp. showed no significant changes. 5-HT production was significantly higher in the proximal small intestine and the cecum in the 25 cGy group compared to the sham group. The 25 cGy group exhibited deficits in recognition in SORM testing at 7 days post-exposure. Taken together, these results suggest a connection between GI microbiome 3 composition, serotonin production, and neurobehavioral performance, and that this connection may be disrupted upon exposure to 25 cGy of 4 He ions

Carli B Jones↗

RED NUGGETS AT HIGH REDSHIFT: STRUCTURAL EVOLUTION OF QUIESCENT GALAXIES OVER 10 Gyr OF COSMIC HISTORY

We present an analysis of the size growth seen in early-type galaxies over 10 Gyr of cosmic time. Our analysis is based on a homogeneous synthesis of published data from 16 spectroscopic surveys observed at similar spatial resolution, augmented by new measurements for galaxies in the Gemini Deep Deep Survey. In total, our sample contains structural data for 465 galaxies (mainly early-type) in the redshift range 0.2 < z < 2.7. The size evolution of passively evolving galaxies over this redshift range is gradual and continuous, with no evidence for an end or change to the process around z {approx} 1, as has been hinted at by some surveys which analyze subsets of the data in isolation. The size growth appears to be independent of stellar mass, with the mass-normalized half-light radius scaling with redshift as R{sub e} {proportional_to}(1 + z){sup -1.62{+-}0.34}. Surprisingly, this power law seems to be in good agreement with the recently reported continuous size evolution of UV-bright galaxies in the redshift range z {approx} 0.5-3.5. It is also in accordance with the predictions from recent theoretical models.

79 ASTRONOMY AND ASTROPHYSICS↗

Wavefront Control and Image Restoration with Less Computing

PseudoDiversity is a method of recovering the wavefront in a sparse- or segmented- aperture optical system typified by an interferometer or a telescope equipped with an adaptive primary mirror consisting of controllably slightly moveable segments. (PseudoDiversity should not be confused with a radio-antenna-arraying method called pseudodiversity.) As in the cases of other wavefront- recovery methods, the streams of wavefront data generated by means of PseudoDiversity are used as feedback signals for controlling electromechanical actuators of the various segments so as to correct wavefront errors and thereby, for example, obtain a clearer, steadier image of a distant object in the presence of atmospheric turbulence. There are numerous potential applications in astronomy, remote sensing from aircraft and spacecraft, targeting missiles, sighting military targets, and medical imaging (including microscopy) through such intervening media as cells or water. In comparison with prior wavefront-recovery methods used in adaptive optics, PseudoDiversity involves considerably simpler equipment and procedures and less computation. For PseudoDiversity, there is no need to install separate metrological equipment or to use any optomechanical components beyond those that are already parts of the optical system to which the method is applied. In Pseudo- Diversity, the actuators of a subset of the segments or subapertures are driven to make the segments dither in the piston, tilt, and tip degrees of freedom. Each aperture is dithered at a unique frequency at an amplitude of a half wavelength of light. During the dithering, images on the focal plane are detected and digitized at a rate of at least four samples per dither period. In the processing of the image samples, the use of different dither frequencies makes it possible to determine the separate effects of the various dithered segments or apertures. The digitized image-detector outputs are processed in the spatial-frequency (Fourier-transform) domain to obtain measures of the piston, tip, and tilt errors over each segment or subaperture. Once these measures are known, they are fed back to the actuators to correct the errors. In addition, measures of errors that remain after correction by use of the actuators are further utilized in an algorithm in which the image is phase-corrected in the spatial-frequency domain and then transformed back to the spatial domain at each time step and summed with the images from all previous time steps to obtain a final image having a greater signal-to-noise ratio (and, hence, a visual quality) higher than would otherwise be attainable.

Lyon, Richard G.↗

THE CHARACTERISTIC STAR FORMATION HISTORIES OF GALAXIES AT REDSHIFTS z {approx} 2-7

A large sample of spectroscopically confirmed star-forming galaxies at redshifts 1.4 {<=} z{sub spec} {<=} 3.7, with complementary imaging in the near- and mid-IR from the ground and from the Hubble Space Telescope and Spitzer Space Telescope, is used to infer the average star formation histories (SFHs) of typical galaxies from z {approx} 2 to 7. For a subset of 302 galaxies at 1.5 {<=} z{sub spec} < 2.6, we perform a detailed comparison of star formation rates (SFRs) determined from spectral energy distribution (SED) modeling (SFRs[SED]) and those calculated from deep Keck UV and Spitzer/MIPS 24 {mu}m imaging (SFRs[IR+UV]). Exponentially declining SFHs yield SFRs[SED] that are 5-10 times lower on average than SFRs[IR+UV], indicating that declining SFHs may not be accurate for typical galaxies at z {approx}> 2. The SFRs of z {approx} 2-3 galaxies are directly proportional to their stellar masses (M{sub *}), with unity slope-a result that is confirmed with Spitzer/IRAC stacks of 1179 UV-faint (R>25.5) galaxies-for M{sub *} {approx}> 5 Multiplication-Sign 10{sup 8} M{sub Sun} and SFRs {approx}> 2 M{sub Sun} yr{sup -1}. We interpret this result in the context of several systematic biases that can affect determinations of the SFR-M{sub *} relation. The average specific SFRs at z {approx} 2-3 are remarkably similar within a factor of two to those measured at z {approx}> 4, implying that the average SFH is one where SFRs increase with time. A consequence of these rising SFHs is that (1) a substantial fraction of UV-bright z {approx} 2-3 galaxies had faint sub-L* progenitors at z {approx}> 4; and (2) gas masses must increase with time from z = 2 to 7, over which time the net cold gas accretion rate-as inferred from the specific SFR and the Kennicutt-Schmidt relation-is {approx}2-3 times larger than the SFR. However, if we evolve to higher redshift the SFHs and masses of the halos that are expected to host L* galaxies at z {approx} 2, then we find that {approx}< 10% of the baryons accreted onto typical halos at z {approx}> 4 actually contribute to star formation at those epochs. These results highlight the relative inefficiency of star formation even at early cosmic times when galaxies were first assembling.

79 ASTRONOMY AND ASTROPHYSICS↗

Search for GeV gamma-ray bursts with the ARGO-YBJ detector: summary of eight years of observations

The search for gamma-ray burst (GRB) emission in the energy range of 1-100 GeV in coincidence with the satellite detection has been carried out using the Astrophysical Radiation with Ground-based Observatory at YangBaJing (ARGO-YBJ) experiment. The high-altitude location (4300 m a.s.l.), the large active surface (∼6700 m{sup 2} of Resistive Plate Chambers), the wide field of view (∼2 sr, limited only by the atmospheric absorption), and the high duty cycle (>86%) make the ARGO-YBJ experiment particularly suitable to detect short and unexpected events like GRBs. With the scaler mode technique, i.e., counting all the particles hitting the detector with no measurement of the primary energy and arrival direction, the minimum threshold of ∼1 GeV can be reached, overlapping the direct measurements carried out by satellites. During the experiment lifetime from 2004 December 17 to 2013 February 7, a total of 206 GRBs occurring within the ARGO-YBJ field of view (zenith angle θ ≤ 45°) have been analyzed. This is the largest sample of GRBs investigated with a ground-based detector. Two light curve models have been assumed and since in both cases no significant excess has been found, the corresponding fluence upper limits in the 1-100 GeV energy region have been derived, with values as low as 10{sup –5} erg cm{sup –2}. The analysis of a subset of 24 GRBs with known redshift has been used to constrain the fluence extrapolation to the GeV region together with possible cutoffs under different assumptions on the spectrum.

79 ASTRONOMY AND ASTROPHYSICS↗

On Finding Black Holes in Photometric Microlensing Surveys

There are expected to be millions of isolated black holes in the galaxy resulting from the deaths of massive stars. Measuring the abundance and properties of this remnant population would shed light on the end stages of stellar evolution and the evolution paths of black hole systems. Detecting isolated black holes is currently only possible via gravitational microlensing, which has so far yielded one definitive detection. The difficulty in finding microlensing black holes lies in having to choose a small subset of events, based on characteristics of their light curves, to allocate expensive and scarce follow-up resources to confirm the identity of the lens. Current methods either rely on simple cuts in parameter space without using the full distribution information or are only effective on small subsets of events. In this paper, we present a new lens classification method. The classifier takes in posterior constraints on light-curve parameters and combines them with a Galactic simulation to estimate the lens class probability. This method is flexible and can be used with any set of microlensing light-curve parameters, making it applicable to large samples of events. We make this classification framework available via the popclass Python package. We apply the classifier to ~10,000 microlensing events from the Optical Gravitational Lensing Experiment survey and find 23 high-probability black hole candidates. Our classifier also suggests that the only known isolated black hole is an observational outlier, according to current Galactic models, and the allocation of astrometric follow-up on this event was a high-risk strategy.

79 ASTRONOMY AND ASTROPHYSICS↗

A machine learning approach to emulation and biophysical parameter estimation with the Community Land Model, version 5

Abstract. Land models are essential tools for understanding and predicting terrestrial processes and climate–carbon feedbacks in the Earth system, but uncertainties in their future projections are poorly understood. Improvements in physical process realism and the representation of human influence arguably make models more comparable to reality but also increase the degrees of freedom in model configuration, leading to increased parametric uncertainty in projections. In this work we design and implement a machine learning approach to globally calibrate a subset of the parameters of the Community Land Model, version 5 (CLM5) to observations of carbon and water fluxes. We focus on parameters controlling biophysical features such as surface energy balance, hydrology, and carbon uptake. We first use parameter sensitivity simulations and a combination of objective metrics including ranked global mean sensitivity to multiple output variables and non-overlapping spatial pattern responses between parameters to narrow the parameter space and determine a subset of important CLM5 biophysical parameters for further analysis. Using a perturbed parameter ensemble, we then train a series of artificial feed-forward neural networks to emulate CLM5 output given parameter values as input. We use annual mean globally aggregated spatial variability in carbon and water fluxes as our emulation and calibration targets. Validation and out-of-sample tests are used to assess the predictive skill of the networks, and we utilize permutation feature importance and partial dependence methods to better interpret the results. The trained networks are then used to estimate global optimal parameter values with greater computational efficiency than achieved by hand tuning efforts and increased spatial scale relative to previous studies optimizing at a single site. By developing this methodology, our framework can help quantify the contribution of parameter uncertainty to overall uncertainty in land model projections.

54 ENVIRONMENTAL SCIENCES↗

An imaging spectrometer experiment for the Shuttle

An imaging spectrometer experiment concept for earth remote sensing, developed as part of NASA's Multispectral Linear Array program, will map a series of test sites at high spatial and spectral resolution from the vantage point of the Shuttle payload bay. The instrument covers the spectral range from 0.4 to 2.5 micrometers with a sampling interval of 10 nanometers in the visible and near infrared (to 1.0 micrometer) and 20 nanometers in the short wavelength infrared (1.0 to 2.5 micrometers). Resolution corresponding to a ground instantaneous field of view (pixel size) of 30 meters is provided over a swath width of 12 kilometers. On-board data editing is utilized to select a subset of the data stream for transmission to the ground. The instrument utilizes silicon and mercury cadmium telluride area array detectors. Pointing mirrors are included to permit specific test sites to be imaged from the Shuttle orbit.

Wellman, J. B.↗

The EXOSAT database and archive

The EXOSAT database provides on-line access to the results and data products (spectra, images, and lightcurves) from the EXOSAT mission as well as access to data and logs from a number of other missions (such as EINSTEIN, COS-B, ROSAT, and IRAS). In addition, a number of familiar optical, infrared, and x ray catalogs, including the Hubble Space Telescope (HST) guide star catalog are available. The complete database is located at the EXOSAT observatory at ESTEC in the Netherlands and is accessible remotely via a captive account. The database management system was specifically developed to efficiently access the database and to allow the user to perform statistical studies on large samples of astronomical objects as well as to retrieve scientific and bibliographic information on single sources. The system was designed to be mission independent and includes timing, image processing, and spectral analysis packages as well as software to allow the easy transfer of analysis results and products to the user's own institute. The archive at ESTEC comprises a subset of the EXOSAT observations, stored on magnetic tape. Observations of particular interest were copied in compressed format to an optical jukebox, allowing users to retrieve and analyze selected raw data entirely from their terminals. Such analysis may be necessary if the user's needs are not accommodated by the products contained in the database (in terms of time resolution, spectral range, and the finesse of the background subtraction, for instance). Long-term archiving of the full final observation data is taking place at ESRIN in Italy as part of the ESIS program, again using optical media, and ESRIN have now assumed responsibility for distributing the data to the community. Tests showed that raw observational data (typically several tens of megabytes for a single target) can be transferred via the existing networks in reasonable time.

Reynolds, A. P.↗

Copper transformation, speciation, and detoxification in anoxic and suboxic freshwater sediments

The complex chemistry of copper (Cu) in freshwater sediments at low concentrations is not well understood. We evaluated the transformation processes of Cu added to freshwater sediments under suboxic and anoxic conditions. Freshwater sediments from three sources in Michigan with different characteristics (Spring Creek, River Raisin, and Maple Lake) were spiked with 30 or 60 mg kg -1 Cu and incubated under a nitrogen atmosphere. After 28-d, each treatment subset was amended with organic matter (OM) to promote anoxic conditions and evaluate its effects on Cu speciation. OM addition triggered a shift from suboxic to anoxic conditions, and sequential extractions showed that Cu accordingly shifted from acid-soluble to oxidizable fractions. Extended X-ray absorption fine-structure (EXAFS) spectroscopy revealed that Cu sulfides dominated all anoxic samples except for Spring Creek 30 mg kg -1 , where Cu(I) was predominantly complexed to thiol groups of OM. Covellite and chalcopyrite (CuFeS 2 ) were the predominant Cu species in nearly all anoxic samples, as determined by Raman spectroscopy, scanning electron microscopy, and X-ray absorption near-edge structure (XANES) spectroscopy. Copper reduction also occurred under suboxic conditions: for two of three sediments, around 80% had been reduced to Cu(I), while the remaining 20% persisted as Cu(II) complexed to OM. However, in the third coarsest (i.e., Spring Creek), around 50% of the Cu had been reduced, forming Cu(I)-OM complexes, while the remainder was Cu(II)-OM complexes. Toxicity tests showed that survival of H. azteca and D. magna were significantly lower in suboxic treatments. Anoxic sediments triggered a near-complete transformation of Cu to sulfide minerals, reducing its toxicity.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A Look Inside the Black Box: Using graph-theoretical descriptors to interpret a Continuous-Filter Convolutional Neural Network (CF-CNN) trained on the global and local minimum energy structures of neutral water clusters

A Continuous Filter Convolutional Neural Network (CF-CNN) was trained to predict the potential energy of water cluster networks \ce{(H2O)_{\textit{N}}}, \textit{N}=10--30, corresponding to local minima lying within 5 kcal/mol from the putative minima taken from a newly published database containing over 5 million unique networks. The chemical sampling space of the database was characterized using chemical descriptors derived from graph theory, which led to the identification of important trends in the topology, connectivity, polygon structures associated with the various networks as a function of cluster size. The resulting graphs are available alongside the original database at \url{https://sites.uw.edu/wdbase/}. The CF-CNN trained on a subset of 500,000 networks for (\textit{N}=10, 30) yielded a mean absolute error of 0.002$\pm$0.002 kcal/mol per water molecule, giving the trained CF-CNN the highest accuracy of any neural network-based surrogate model to date. In addition, clusters of sizes not included in the training set exhibited errors of the same magnitude, indicating that the CF-CNN ptotocol is general enough to accurately predict energies of networks for both smaller and larger sizes than those used during training. The graph-theoretical descriptors were developed in order to analyze the properties of the full database and interpret the predictive power of the CF-CNN. Using topology measures, such as the Wiener index and the average shortest path length along with two similarity measures, we showed that all networks from the test set were within the range of the ones from the training set, suggesting that the training set covered the chemical space of interest quite well. Our graph analysis suggests that the mean degree and number of polygons for networks with larger errors tend to lie further from the mean than those with lower errors. The generality of the used CF-CNN was thus demonstrated, while the use of the graph-theoretical descriptors assisted in interpreting the predicted results.

Bilbrey, Jenna A.↗

Applying Machine Learning to Understand Write Performance of Large-scale Parallel Filesystems

In high-performance computing (HPC), I/O performance prediction offers the potential to improve the efficiency of scientific computing. In particular, accurate prediction can make runtime estimates more precise, guide users toward optimal checkpoint strategies, and better inform facility provisioning and scheduling policies. HPC I/O performance is notoriously difficult to predict and model, however, in large part because of inherent variability and a lack of transparency in the behaviors of constituent storage system components. In this work we seek to advance the state of the art in HPC I/O performance prediction by (1) modeling the mean performance to address high variability, (2) deriving model features from write patterns, system architecture and system configurations, and (3) employing Lasso regression model to improve model accuracy. We demonstrate the efficacy of our approach by applying it to a crucial subset of common HPC I/O motifs, namely, file-per-process checkpoint write workloads. We conduct experiments on two distinct production HPC platforms—Titan at the Oak Ridge Leadership Computing Facility and Cetus at the Argonne Leadership Computing Facility—to train and evaluate our models. We find that we can attain ≤ 30% relative error for 92.79% and 99.64% of the samples in our test set on these platforms, respectively.

Xie, Bing↗