Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model interpretation”

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 631 records · Page 35

Less is More: Bigger Data from Compressive Measurements

Compressive sensing approaches are beginning to take hold in (scanning) transmission electron microscopy (S/TEM) [1,2,3]. Compressive sensing is a mathematical theory about acquiring signals in a compressed form (measurements) and the probability of recovering the original signal by solving an inverse problem [4]. The inverse problem is underdetermined (more unknowns than measurements), so it is not obvious that recovery is possible. Compression is achieved by taking inner products of the signal with measurement weight vectors. Both Gaussian random weights and Bernoulli (0,1) random weights form a large class of measurement vectors for which recovery is possible. The measurements can also be designed through an optimization process. The key insight for electron microscopists is that compressive sensing can be used to increase acquisition speed and reduce dose. Building on work initially developed for optical cameras, this new paradigm will allow electron microscopists to solve more problems in the engineering and life sciences. We will be collecting orders of magnitude more data than previously possible. The reason that we will have more data is because we will have increased temporal/spatial/spectral sampling rates, and we will be able ability to interrogate larger classes of samples that were previously too beam sensitive to survive the experiment. For example consider an in-situ experiment that takes 1 minute. With traditional sensing, we might collect 5 images per second for a total of 300 images. With compressive sensing, each of those 300 images can be expanded into 10 more images, making the collection rate 50 images per second, and the decompressed data a total of 3000 images [3]. But, what are the implications, in terms of data, for this new methodology? Acquisition of compressed data will require downstream reconstruction to be useful. The reconstructed data will be much larger than traditional data, we will need space to store the reconstructions during analysis, and the computational demands for analysis will be higher. Moreover, there will be time costs associated with reconstruction. Deep learning [5] is an approach to address these problems. Deep learning is a hierarchical approach to find useful (for a particular task) representations of data. Each layer of the hierarchy is intended to represent higher levels of abstraction. For example, a deep model of faces might have sinusoids, edges and gradients in the first layer; eyes, noses, and mouths in the second layer, and faces in the third layer. There has been significant effort recently in deep learning algorithms for tasks beyond image classification such as compressive reconstruction [6] and image segmentation [7]. A drawback of deep learning, however, is that training the model requires large datasets and dedicated computational resources (to reduce training time to a few days). A second issue is that deep learning is not user-friendly and the meaning behind the results is usually not interpretable. We have shown it is possible to reduce the data set size while maintaining model quality [8] and developed interpretable models for image classification [9], but the demands are still significant. The key to addressing these problems is to NOT reconstruct the data. Instead, we should design computational sensors that give answers to specific problems. A simple version of this idea is compressive classification [10], where the goal is to classify signal type from a small number of compressed measurements. Classification is a much simpler problem than reconstruction, so 1) much fewer measurements will be necessary, and 2) these measurements will probably not be useful for reconstruction. Other simple examples of computational sensing include determining object volume or the number of objects present in the field of view [11].

Stevens, Andrew↗

Search for dark matter produced in association with one or two top quarks in proton-proton collisions at $\sqrt{\text{s}}$ = 13 TeV

A search is performed for dark matter (DM) produced in association with a single top quark or a pair of top quarks using the data collected with the CMS detector at the LHC from proton-proton collisions at a center-of-mass energy of 13 TeV, corresponding to 138 fb −1 of integrated luminosity. An excess of events with a large imbalance of transverse momentum is searched for across 0, 1 and 2 lepton final states. Novel multivariate techniques are used to take advantage of the differences in kinematic properties between the two DM production mechanisms. No significant deviations with respect to the standard model predictions are observed. The results are interpreted considering a simplified model in which the mediator is either a scalar or pseudoscalar particle and couples to top quarks and to DM fermions. Axion-like particles that are coupled to top quarks and DM fermions are also considered. Expected exclusion limits of 410 and 380 GeV for scalar and pseudoscalar mediator masses, respectively, are set at the 95% confidence level. A DM particle mass of 1 GeV is assumed, with mediator couplings to fermions and DM particles set to unity. A small signal-like excess is observed in data, with the largest local significance observed to be 1.9 standard deviations for the 150 GeV pseudoscalar mediator hypothesis. Because of this excess, mediator masses are only excluded below 310 (320) GeV for the scalar (pseudoscalar) mediator. The results are also translated into model-independent 95% confidence level upper limits on the visible cross section of DM production in association with top quarks, ranging from 1 pb to 0.02 pb.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Search for new phenomena in final states with large jet multiplicities and missing transverse momentum using $\sqrt {s}$ = 7 TeV pp collisions with the ATLAS detector

Results are presented of a search for any particle(s) decaying to six or more jets in association with missing transverse momentum. The search is performed using 1.34 fb -1 of $\sqrt {s}$ = 7 TeV proton-proton collisions recorded by the ATLAS detector during 2011. Data-driven techniques are used to determine the backgrounds in kinematic regions that require at least six, seven or eight jets, well beyond the multiplicities required in previous analyses. No evidence is found for physics beyond the Standard Model. The results are interpreted in the context of a supersymmetry model (MSUGRA/CMSSM) where they extend previous constraints.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Computation of Alfvèn eigenmode stability and saturation through a reduced fast ion transport model in the TRANSP tokamak transport code

Alfvénic instabilities (AEs) are well known as a potential cause of enhanced fast ion transport in fusion devices. Given a specific plasma scenario, quantitative predictions of (i) expected unstable AE spectrum and (ii) resulting fast ion transport are required to prevent or mitigate the AE-induced degradation in fusion performance. Reduced models are becoming an attractive tool to analyze existing scenarios as well as for scenario prediction in time-dependent simulations. Here, in this work, a neutral beam heated NSTX discharge is used as reference to illustrate the potential of a reduced fast ion transport model, known as kick model, that has been recently implemented for interpretive and predictive analysis within the framework of the time-dependent tokamak transport code TRANSP. Predictive capabilities for AE stability and saturation amplitude are first assessed, based on given thermal plasma profiles only. Predictions are then compared to experimental results, and the interpretive capabilities of the model further discussed. Overall, the reduced model captures the main properties of the instabilities and associated effects on the fast ion population. Finally, additional information from the actual experiment enables further tuning of the model's parameters to achieve a close match with measurements.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Computation of Alfvèn eigenmode stability and saturation through a reduced fast ion transport model in the TRANSP tokamak transport code

Alfvénic instabilities (AEs) are well known as a potential cause of enhanced fast ion transport in fusion devices. Given a specific plasma scenario, quantitative predictions of (i) expected unstable AE spectrum and (ii) resulting fast ion transport are required to prevent or mitigate the AE- induced degradation in fusion performance. Reduced models are becoming an attractive tool to analyze existing scenarios as well as for scenario prediction in time-dependent simulations. In this work, a neutral beam heated NSTX discharge is used as reference to illustrate the potential of a reduced fast ion transport model, known as kick model, that has been recently implemented for interpretive and predictive analysis within the framework of the time-dependent tokamak transport code TRANSP. Predictive capabilities for AE stability and saturation amplitude are first assessed, based on given thermal plasma profiles only. Predictions are then compared to experimental results, and the interpretive capabilities of the model further discussed. Overall, the reduced model captures the main properties of the instabilities and associated effects on the fast ion population. Additional information from the actual experiment enables further tuning of the model’s parameters to achieve a close match with measurements.

Alfven instabilities↗

Search for direct top squark pair production in events with a Z boson, b-jets and missing transverse momentum in sqrt(s)=8 TeV pp collisions with the ATLAS detector

CERN-LHC. A search is presented for direct top squark pair production using events with at least two leptons including a same-flavour opposite-sign pair with invariant mass consistent with the Z boson mass, jets tagged as originating from b-quarks and missing transverse momentum. The analysis is performed with proton-proton collision data at sqrt(s) = 8 TeV collected with the ATLAS detector at the LHC in 2012 corresponding to an integrated luminosity of 20.3 fb^-1. No excess beyond the Standard Model expectation is observed. Interpretations of the results are provided in models based on the direct pair production of the heavier top squark state (t2) followed by the decay to the lighter top squark state (t1) via t2 --> Z t1, and for t1 pair production in natural gauge-mediated supersymmetry breaking scenarios where the neutralino (chi1^0) is the next-to-lightest supersymmetric particle and decays producing a Z boson and a gravitino (G) via the chi1^0 --> Z G process.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Interpreting space-based trends in carbon monoxide with multiple models

Here, we use a series of chemical transport model and chemistry climate model simulations to investigate the observed negative trends in MOPITT CO over several regions of the world, and to examine the consistency of time-dependent emission inventories with observations. We also found that simulations driven by the MACCity inventory, used for the Chemistry Climate Modeling Initiative (CCMI), reproduce the negative trends in the CO column observed by MOPITT for 2000–2010 over the eastern United States and Europe. However, the simulations have positive trends over eastern China, in contrast to the negative trends observed by MOPITT. The model bias in CO, after applying MOPITT averaging kernels, contributes to the model–observation discrepancy in the trend over eastern China. This demonstrates that biases in a model's average concentrations can influence the interpretation of the temporal trend compared to satellite observations. The total ozone column plays a role in determining the simulated tropospheric CO trends. A large positive anomaly in the simulated total ozone column in 2010 leads to a negative anomaly in OH and hence a positive anomaly in CO, contributing to the positive trend in simulated CO. Our results demonstrate that accurately simulating variability in the ozone column is important for simulating and interpreting trends in CO.

54 ENVIRONMENTAL SCIENCES↗

Reinforcement learning based hybrid bond-order coarse-grained interatomic potentials for exploring mesoscale aggregation in liquid–liquid mixtures

Exploring mesoscopic physical phenomena has always been a challenge for brute-force all-atom molecular dynamics simulations. Although recent advances in computing hardware have improved the accessible length scales, reaching mesoscopic timescales is still a significant bottleneck. Coarse-graining of all-atom models allows robust investigation of mesoscale physics with a reduced spatial and temporal resolution but preserves desired structural features of molecules, unlike continuum-based methods. Here, we present a hybrid bond-order coarse-grained forcefield (HyCG) for modeling mesoscale aggregation phenomena in liquid–liquid mixtures. The intuitive hybrid functional form of the potential offers interpretability to our model, unlike many machine learning based interatomic potentials. We parameterize the potential with the continuous action Monte Carlo Tree Search (cMCTS) algorithm, a reinforcement learning (RL) based global optimizing scheme, using training data from all-atom simulations. The resulting RL-HyCG correctly describes mesoscale critical fluctuations in binary liquid–liquid extraction systems. cMCTS, the RL algorithm, accurately captures the mean behavior of various geometrical properties of the molecule of interest, which were excluded from the training set. The developed potential model along with the RL-based training workflow could be applied to explore a variety of other mesoscale physical phenomena that are typically inaccessible to all-atom molecular dynamics simulations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

STAR FORMATION AND YOUNG POPULATION OF THE H II COMPLEX Sh2-294

The Sh2-294 H II region ionized by a single B0V star features several infrared excess sources, a photodissociation region, and also a group of reddened stars at its border. The star formation scenario in this region seems to be quite complex. In this paper, we present follow-up results of Sh2-294 H II region at 3.6, 4.5, 5.8, and 8.0 {mu}m observed with the Spitzer Space Telescope Infrared Array Camera (IRAC), coupled with H{sub 2} (2.12 {mu}m) observation, to characterize the young population of the region and to understand its star formation history. We identified 36 young stellar object (YSO, Class I, Class II, and Class I/II) candidates using IRAC color-color diagrams. It is found that Class I sources are preferentially located at the outskirts of the H II region and associated with enhanced H{sub 2} emission; none of them are located near the central cluster. Combining the optical to mid-infrared (MIR) photometry of the YSO candidates and using the spectral energy distribution fitting models, we constrained stellar parameters and the evolutionary status of 33 YSO candidates. Most of them are interpreted by the model as low-mass (<4 M{sub Sun }) YSOs; however, we also detected a massive YSO ({approx}9 M{sub Sun }) of Class I nature, embedded in a cloud of visual extinction of {approx}24 mag. Present analysis suggests that the Class I sources are indeed a younger population of the region relative to Class II sources (age {approx} 4.5 Multiplication-Sign 10{sup 6} yr). We suggest that the majority of the Class I sources, including the massive YSOs, are second-generation stars of the region whose formation is possibly induced by the expansion of the H II region powered by a {approx}4 Multiplication-Sign 10{sup 6} yr B0 main-sequence star.

79 ASTRONOMY AND ASTROPHYSICS↗

Combination of searches for singly produced vectorlike top quarks in 𝑝⁢𝑝 collisions at $\sqrt{𝑠}$ =13 TeV with the ATLAS detector

A combination of searches for the single production of vectorlike top quarks (𝑇) is presented. These analyses are based on proton-proton collisions at $\sqrt{𝑠}$ = 13 TeV recorded in 2015–2018 with the ATLAS detector at the Large Hadron Collider, corresponding to an integrated luminosity of 139 fb −1 . The 𝑇 decay modes considered in this combination are into a top quark and either a Standard Model Higgs boson or a 𝑍 boson (𝑇 → 𝐻⁡𝑡 and 𝑇 → 𝑍⁢𝑡). The individual searches used in the combination are differentiated by the number of leptons (𝑒, 𝜇) in the final state. The observed data are found to be in good agreement with the Standard Model background prediction. Interpretations are provided for a range of masses and couplings of the vectorlike top quark for benchmark models and generalized representations in terms of 95% confidence level limits. For a benchmark signal prediction of a vectorlike top quark SU(2) singlet with electroweak coupling, 𝜅, of 0.5, masses below 2.1 TeV are excluded, resulting in the most restrictive limits to date.

Composite models↗

A search for long-lived particles that stop in the CMS detector and decay to muons

A search for long-lived particles that are produced in proton-proton collisions at the CERN LHC, come to rest in the CMS detector, and decay to muons is presented. The decays of the stopped particles could be observed during the intervals between LHC beam crossings, at times that are well separated from any proton-proton collisions. The analysis uses 19.7 1/fb of 8 TeV data collected by CMS in 2012, during a search interval of 293 hours of trigger livetime. Massive, long-lived particles do not exist in the Standard Model, and so any sign of them would be an indication of new physics. The results are interpreted with a model that predicts a long-lived particle that has a charge of twice the electron charge and that behaves like a lepton. Cross section limits are set for each long-lived particle mass as a function of lifetime, for lifetimes between 100 ns and 10 days. These are the first limits for long-lived stopped particles that decay to muons.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

The right conditions for high-precision dynamic temperature and heat capacity measurement via pyrometry and conductivity

The pursuit of accurate bulk temperature T under extreme conditions has been a long-standing goal of the high pressure science community, complicated by a lack of data to inform models. To reach these extremely high-pressure, high-temperature (high P − T) conditions, a combination of dynamic and heated static experiments (e.g., diamond or gem anvil cel experiments) are used. For example, in a diamond anvil cell (DAC) experiment, a sample placed in the DAC is first pressurized. Following pressurization, the sample T is increased either by heating the entire DAC (usually using resistive heating, and limited to ∼1000K) or by applying intense laser power to the sample surfaces. In a dynamic experiment, the process of pressurizing the sample also heats it. In the case of shock physics experiments, such heating is substantial, easily reaching thousands of Kelvin; in our work we have seen T ∼17000K. Most methods of measuring temperature at ambient are not compatible with experiments under these high-pressure, high-temperature conditions: thermocouples break, melt, or have conductivity properties that differ from ambient where they are calibrated; thermometers would melt; both are too slow. As a result most methods are based on non-contact techniques such as x-ray diffraction broadening, neutron scattering, or optical methods. Of these, optical methods using the visible and near-infrared region of the spectrum are the most commonly used as the sources and detectors are readily available. In the case of optical methods the optical depth, and therefore the measurement location, is limited to the surface. When a window or anvil material is used, heat flows from the sample into the window/anvil. Likewise, if the sample undergoes a change in thermodynamic state, such as expansion upon release, different T may be expected. As a result, the surface or apparent temperature T app measurement will differ from the bulk or interior temperature that is desired. This surface measurement must be related to the bulk measurement using thermal transport models and material models. While it is tempting to conclude that one should just use x-ray methods that directly probe the interior, even these methods have been shown to depend on thermal transport and material models. Regardless of the method used to create the high P − T condition, therefore, we must understand the role of thermal transport and material models upon our interpretation of the T measurement, as well as the errors and uncertainties associated with the choice of models used in the analysis. This is a substantial area of research and this paper is by no means a complete survey of the relevant sources of uncertainty. For example, we have yet to begin to address alternate transport models in a detailed manner (e.g., Tan-Ahrens), or the many models that use additional layers to approximate melting, turbulence, or epitaxial phenomena). Likewise, we have not explored the impact upon uncertainty of thermal models that use temperature-dependent thermal transport coefficients, or the wide range of material models that can be applied. Instead, this paper focuses on using one simple model, the Urtiew-Grover model, to understand the sources of error in T measurement so that we may identify how best to focus future research efforts to return the best improvements and avoid working on over-optimizing a single type of measurement. To this end, we work through some of the best and worst case scenarios for T measurement.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Replication Data for: Measurement of the mean number of muons with energies above 500 GeV in air showers detected with the IceCube Neutrino Observatory

<b>Measurement of the mean number of muons with energies above 500 GeV in air showers detected with the IceCube Neutrino Observatory</b> <br><br> This data release accompanies results submitted to Physical Review D describing the measurement of the average multiplicity of TeV muons with IceCube. It contains the data necessary to reproduce the main plots from the paper (Figs. 7 and 9), i.e. the numerical results for the average number of muons with energies above 500 GeV as a function of primary cosmic ray energy. <br><br> For any questions about this data release, please write to analysis@icecube.wisc.edu. <br><br> Files included in this release: <ul> <li>A README file <li>Files including data to reproduce the results plots from the paper (see below for details) <li>An example python script showing how to read and plot the data </ul> <br> <u>What is in the files icecube_Nmu500_X_Y.txt:</u> <br> Y indicates wether the file contains values obtained from experimental data (Y="data") or air-shower simulations (Y="MC"). <br> X indicates the hadronic interaction model for which the plot is made. If Y="data", this means that the experimental data was interpreted using this model. If Y="MC", it means that the simulations were performed with this model. The three models included are Sibyll 2.1, QGSJet-II.04, and EPOS-LHC (see paper for references). The file with X="modelaverage" gives the average over the three individual results with the deviations from the average included in the systematic uncertainties. <br><br> Please see the README file for details on how the data is structured in the files.

Astroparticle Physics↗

Explainable AI (XAI)-driven vibration sensing scheme for surface quality monitoring in a smart surface grinding process

Local Interpretable and Model-agnostic Explanation (LIME), an explainable artificial intelligence (XAI) approach is adapted to identify the globally important time-frequency bands for predicting average surface roughness (Ra) in a smart grinding process. The smart grinding setup consisted of a Supertech CNC precision surface grinding machine, instrumented with a Dytran piezoelectric accelerometer attached to the tailstock quill along the tangential direction (Y-axis). For every grinding pass, vibration signatures were captured, and the ground truth surface roughness values were recorded using a Mahr Marsurf M300C portable surface roughness profilometer. The roughness values ranged from 0.06 to 0.14 microns over the complete set of experiments. Time-frequency domain spectrogram frames were extracted for each of the vibration signals collected during the grinding process. Convolutional Neural Networks (CNNs) were modeled to predict the surface roughness based on these spectrogram frames and their image augmentations. The best CNN model was able to predict the roughness values with an overall R2-score of 0.95, training R2-score of 0.99, and testing R2-score of 0.81 with only 80 sets of vibration signals corresponding to 4 experiments with 20 trials each. Although the data size is not large enough to guarantee such performance metrics in real-world scenarios, one can extract statistically consistent explanations underlying the relationships these complex deep learning models capture. Further, the LIME methodology was implemented on the developed surface roughness CNN model to identify the important time-frequency bands (i.e., the superpixels of a spectrogram) influencing the predictions. Based on the identified important regions on the spectrogram frames, the corresponding frequency characteristics were determined that influence the surface roughness predictions. The important frequency range based on LIME results was approximately 11.7 to 19.1 kHz. The power of XAI was demonstrated by cutting down the sampling rate from 160 kHz to 30, 20, 10, and 5 kHz based on the important frequency range and considering Nyquist criteria. Separate CNN models were developed for these ranges by only extracting time-frequency contents below their corresponding Nyquist cut-offs. A proper data acquisition strategy is proposed by comparing the model performances to argue the selection of a sufficient sampling rate to capture the grinding process successfully and robustly.

42 ENGINEERING↗

Reprint of: Explainable AI (XAI)-driven vibration sensing scheme for surface quality monitoring in a smart surface grinding process

Local Interpretable and Model-agnostic Explanation (LIME), an explainable artificial intelligence (XAI) approach is adapted to identify the globally important time-frequency bands for predicting average surface roughness (Ra) in a smart grinding process. The smart grinding setup consisted of a Supertech CNC precision surface grinding machine, instrumented with a Dytran piezoelectric accelerometer attached to the tailstock quill along the tangential direction (Y-axis). For every grinding pass, vibration signatures were captured, and the ground truth surface roughness values were recorded using a Mahr Marsurf M300C portable surface roughness profilometer. The roughness values ranged from 0.06 to 0.14 microns over the complete set of experiments. Time-frequency domain spectrogram frames were extracted for each of the vibration signals collected during the grinding process. Convolutional Neural Networks (CNNs) were modeled to predict the surface roughness based on these spectrogram frames and their image augmentations. The best CNN model was able to predict the roughness values with an overall R2-score of 0.95, training R2-score of 0.99, and testing R2-score of 0.81 with only 80 sets of vibration signals corresponding to 4 experiments with 20 trials each. Although the data size is not large enough to guarantee such performance metrics in real-world scenarios, one can extract statistically consistent explanations underlying the relationships these complex deep learning models capture. Further, the LIME methodology was implemented on the developed surface roughness CNN model to identify the important time-frequency bands (i.e., the superpixels of a spectrogram) influencing the predictions. Based on the identified important regions on the spectrogram frames, the corresponding frequency characteristics were determined that influence the surface roughness predictions. The important frequency range based on LIME results was approximately 11.7 to 19.1 kHz. The power of XAI was demonstrated by cutting down the sampling rate from 160 kHz to 30, 20, 10, and 5 kHz based on the important frequency range and considering Nyquist criteria. Separate CNN models were developed for these ranges by only extracting time-frequency contents below their corresponding Nyquist cut-offs. A proper data acquisition strategy is proposed by comparing the model performances to argue the selection of a sufficient sampling rate to capture the grinding process successfully and robustly.

47 OTHER INSTRUMENTATION↗

Interpreting machine learning prediction of fire emissions and comparison with FireMIP process-based models

Annual burned areas in the United States have increased 2-fold during the past decades. With more large fires resulting in more emissions of fine particulate matter, an accurate prediction of fire emissions is critical for quantifying the impacts of fires on air quality, human health, and climate. This study aims to construct a machine learning (ML) model with game-theory interpretation to predict monthly fire emissions over the contiguous US (CONUS) and to understand the controlling factors of fire emissions. The optimized ML model is used to diagnose the process-based models in the Fire Modeling Intercomparison Project (FireMIP) to inform future development. Results show promising performance for the ML model, Community Land Model (CLM), and Joint UK Land Environment Simulator-Interactive Fire And Emission Algorithm For Natural Environments (JULES-INFERNO) in reproducing the spatial distributions, seasonality, and interannual variability of fire emissions over the CONUS. Regional analysis shows that only the ML model and CLM simulate the realistic interannual variability of fire emissions for most of the subregions (r >0.95 for ML and r =0.14~0.70 for CLM), except for Mediterranean California, where all the models perform poorly (r =0.74 for ML and r <0.30 for the FireMIP models). Regarding seasonality, most models capture the peak emission in July over the western US. However, all models except for the ML model fail to reproduce the bimodal peaks in July and October over Mediterranean California, which may be explained by the smaller wind speeds of the atmospheric forcing data during Santa Ana wind events and limitations in model parameterizations for capturing the effects of Santa Ana winds on fire activity. Furthermore, most models struggle to capture the spring peak in emissions in the southeastern US, probably due to underrepresentation of human effects and the influences of winter dryness on fires in the models. As for extreme events, both the ML model and CLM successfully reproduce the frequency map of extreme emission occurrence but overestimate the number of months with extremely large fire emissions. Comparing the fire PM 2.5 emissions from the ML model with process-based fire models highlights their strengths and uncertainties for regional analysis and prediction and provides useful insights into future directions for model improvements.

54 ENVIRONMENTAL SCIENCES↗

Stellar mass and color dependence of the three-point correlation function of galaxies in the local universe

The three-point correlation function (3PCF) for galaxies provides an opportunity to measure the non-Gaussianity generated from nonlinear structure formation and also probes information about galaxy formation and evolution that is generally not available from the two-point correlation function (2PCF). We measure the 3PCF of the Sloan Digital Sky Survey DR7 main sample galaxies in both redshift and projected spaces on scales up to 40 h {sup –1} Mpc. We explore the dependence of the 3PCF on galaxy stellar mass and color in order to constrain the formation and evolution for galaxies of different properties. The study of the dependence on these properties also helps better constrain the relation between galaxy stellar mass and color and the properties of their hosting dark-matter halos. We focus on the study of the reduced 3PCF, Q, defined as the ratio between the 3PCF and the sum of the products of the 2PCFs. We find a very weak stellar-mass dependence of Q in both redshift and projected spaces. On small scales, more massive galaxies tend to have slightly higher amplitudes of Q. The shape dependence of Q is also weak on these small scales, regardless of stellar mass and color. The reduced 3PCF has a strong color dependence for the low-mass galaxies, while no significant dependence on color is found for the high-mass galaxies. Low-mass red galaxies have higher amplitudes and stronger shape dependence of the reduced 3PCF than the blue galaxies, implying that these low-mass red galaxies tend to populate filamentary structures. The linear galaxy bias model fails to interpret the color dependence of Q, emphasizing the importance of a nonvanishing quadratic bias parameter in the correct modeling of the galaxy color distribution.

79 ASTRONOMY AND ASTROPHYSICS↗