In recent years, deep learning technologies have received huge attention and interest in the field of high-performance material design. This is primarily because deep learning algorithms in nature have huge advantages over the conventional machine learning models in processing massive amounts of unstructured data with high performance. Besides, deep learning models are capable of recognizing the hidden patterns among unstructured data in an automatic fashion without relying on excessive human domain knowledge. Nevertheless, constructing a robust deep learning model for assessing materials' structure-property relationships remains a non-trivial task due to highly flexible model architecture and the challenge of selecting appropriate material representation methods. In this regard, we develop advanced deep-learning models and implement them for predicting the quantum-chemical calculated properties (i.e., formation energy) for an enormous number of crystal systems. Chapter 1 briefly introduces some fundamental theory of deep learning models (i.e., CNN, GNN) and advanced analysis methods (i.e., saliency map). In Chapter 2, the convolutional neural network (CNN) model is established to find the correlation between the physically intuitive partial electronic density of state (PDOS) and the formation energies of crystals. Importantly, advanced machine learning analysis methods (i.e., salience mapping analysis) are utilized to shed light on underlying physical factors governing the energy properties. In Chapter 3, we introduce the methodology of implementing the cutting-edge graph neural networks (GNN) models for learning an enormous number of crystal structures for the desired properties. / Master of Science / Machine learning technologies, particularly deep learning, have demonstrated remarkable progress in facilitating the high-throughput materials discovery process. In essence, machine learning algorithms have the ability to uncover the hidden patterns of data and make appropriate decisions without being explicitly programmed. Nevertheless, implementing machine learning models in the field of material design remains a challenging task. One of the biggest limitations is our insufficient knowledge about the structure-property relationships for material systems. As the performance of machine learning models is to a large degree determined by the underlying material representation method, which typically requires the experts to have in-depth knowledge of the material systems. Thus, designing effective feature representation methods is always the most crucial aspect for machine learning model development and the process takes a significant amount of manual effort. Even though tremendous efforts have been made in recent years, the research process for robust feature representation methods is still slow. In this regard, we attempt to automate the feature engineering process with the assistance of advanced deep learning algorithms. Unlike the conventional machine learning models, our deep learning models (i.e., convolutional neural networks, graph neural networks) are capable of processing massive amounts of structured data such as spectrum and crystal graphs. Specifically, the deep learning models are explicitly designed to learn the hidden latent variables that are contained in crystal structures in an automatic fashion and provide accurate prediction results. We believe the deep learning models have huge potential to simplify the machine learning modeling process and facilitate the discovery of promising functional materials.

Convective heat transfer during hot pressing in wood-based composite panel manufacturing is widely accepted as the most important means of heat transport for resin curing. The rate of convective heat transfer to the panel core is controlled by its permeability. Permeability in the plane of the panel also controls the flow of vapor to the panel edges, thereby influencing the potential for panel "blowing". This research considers how flake thickness, flake alignment and changing mat density during hot-pressing influences OSB mat permeability, through its thickness and in the plane of the panel. Some previous research exists but it fails to address the affects of horizontal and vertical density gradients as well as flake alignment. An apparatus was designed to allow cold pressing of aligned flakes to desired densities while enabling permeability measurements through the mat thickness. An additional apparatus was designed to allow the measuring of permeability in the plane of the mat. These designs permitted permeability measurements in mats that had no vertical density gradient, allowing for the direct study of permeability versus density (compaction ratio). Superficial permeability was determined using Darcy's law and for each sample, multiple readings were made at five different pressure differentials. Permeability through the mat thickness was highly dependent on compaction ratio and to a lesser extent flake thickness. As the compaction ratio is increased, the initial reduction in permeability is severe, once higher compaction ratios are achieved the reduction in permeability is less pronounced. Permeability decreased with decreasing flake thickness. Permeability in the plane of the mat decreases with increasing compaction ratio but in a less severe manner than through the mat thickness. In this case, the permeability-compaction ratio relationship appears linear in nature. Again, permeability decreases with decreasing flake thickness. / Master of Science

Current carbon and bioenergy markets shifted the focus of typical forest attribute estimation from volume to biomass. We used multiple linear regression and the dataset collected as part of the National Scale Volume and Biomass modeling effort to develop biomass prediction models for Pinus taeda L., Pinus elliottii Engelm. var. elliottii, Pinus echinata Mill., and Pinus palustris Mill. In addition to utilizing traditional forest measurements such as diameter at breast height and total tree height, biomass was estimated as functions of volume, latitude, and longitude. We also evaluated the differences in wood density by geographic location for these species. The best results were obtained when models were fitted using the combined dataset and a log transformed model. Wood density estimates were improved by including latitude and longitude in the model. These findings will be useful to managers seeking improved biomass yield estimates and density by geographic regions.

No / We used biased random-dot dynamic test stimuli to measure the strength of the motion aftereffect (MAE) to evaluate the usefulness of this technique as a measure of motion adaptation strength. The stimuli consisted of noise dots whose individual directions were random and of signal dots moving in a unique direction. All dots moved at the same speed. For each condition, the nulling percentage (percentage of signal dots needed to perceptually null the MAE) was scaled with respect to the coherence threshold (percentage needed to perceive the coherent motion of signal dots without prior adaptation). The increase of these scaled values with the density of dots in the test stimulus suggests that MAE strength is underestimated when measured with low densities. We show that previous reports of high nulling percentages at slow speeds do not reflect strong MAEs, but are actually due to spatio-temporal aliasing, which dramatically increases coherence thresholds. We further show that MAE strength at slow speed increases with eccentricity. These findings are consistent with the idea that using this dynamic test stimulus preferentially reveals the adaptation of a population of high-speed motion units whose activity is independent of adapted low-speed motion units.

Osteoporosis is a condition involving decreased bone mineral density (BMD) and increased fragility of the skeletal system. Osteoporosis affects ~75 million individuals in the United States, Europe, and Japan. In the United States alone, hip fractures affect 500,000 individuals per year, and annual healthcare costs for osteoporotic fractures are approximately $14 billion. A high peak BMD can prevent or delay the onset of osteoporosis and its complications. Exercise and diet may affect peak BMD by as much as 20 to 40% each and have been identified as the two most important controllable factors determining BMD. The current study investigated the effect of a calcium, vitamin D, and vitamin K supplement combination during initial military cadet exercise training on: BMD, stress fracture occurrence, hormones associated with BMD, and biochemical markers of bone turnover. Significant changes in BMD, either between the supplemented group or the unsupplemented group or across time for both groups were not found. The majority of participants (n = 22) had unexpectedly high levels of physical activity prior to enrollment, and the initial military exercise training program included only moderate levels of activity. Therefore, the exercise stimulus to bone was likely insufficient to promote gains in BMD, regardless of the nutrient supplement status. Serum insulin-like growth factor-1 and osteocalcin significantly increased over time (p < 0.05 and p < 0.001, respectively), irrespective of treatment group. Significant decreases were found in dietary intake of calories (p < 0.01), carbohydrate (p < 0.05), protein (p < 0.0001), and fat (p < 0.01) over time. Decreases in reported dietary intake were likely due to less variety of foods eaten, and diminished compliance with food records. Significant differences were not found between groups or across time in dietary intakes of calcium, vitamin D, or vitamin K. Low dose supplementation with a calcium, vitamin D, vitamin K supplement during initial military training in young-adult cadets did not change BMD or alter stress fracture occurrence. / Master of Science

Piezoelectricity in two-dimensional (2D) transition metal dichalcogenides (TMDs) has attracted significant attention due to their unique crystal structure and the lack of inversion centers when the bulk TMDs thin down to monolayer. Although the piezoelectricity effect in atomic-thickness TMDs has been demonstrated, they are not scalable. Herein, we demonstrate a piezoelectric effect from large-scale, sputtered MoS2 and WS2 using a robust defect-engineering based on the thermal-solvent annealing and solvent immersion process. This yields a higher piezoelectric output over 20 times after annealing or solvent immersion. Indeed, the piezoelectric responses are strengthened with the increases of defect density. Moreover, the MoS2 or WS2 piezoelectric device array shows an exceptional piezoelectric sensitivity with a high-level uniformity and excellent environmental stability under ambient conditions. A detailed study of the sulfur vacancy-dependent property and its resultant asymmetric structure-induced piezoelectricity is reported. The proposed approach is scalable and can produce advanced materials for flexible piezoelectric devices to be used in emerging bioinspired robotics and biomedical applications.

Call number: LD2668 .T4 1985 M367 / Master of Architecture

ITC/USA 2005 Conference Proceedings / The Forty-First Annual International Telemetering Conference and Technical Exhibition / October 24-27, 2005 / Riviera Hotel & Convention Center, Las Vegas, Nevada / This paper will present techniques for accurately measuring the power flux density (PFD) of interfering signals at telemetry receiving stations. The solar power flux density is measured daily by radio astronomers and will be used as a calibration signal. The electromagnetic spectrum is being used more intensely as time marches on so being familiar with interference measurement techniques is becoming more important because more interfering signals are present.

published_or_final_version / Physics / Doctoral / Doctor of Philosophy

Let (L,h) → (X,ω) be a compact toric polarized Kahler manifold of complex dimension n. For each k ε N, the fibre-wise Hermitian metric hk on Lk induces a natural inner product on the vector space C∞(X,Lk) of smooth global sections of Lk by integration with respect to the volume form ωn /n! . The orthogonal projection Pk : C∞(X,Lk) → H0(X,Lk) onto the space H0(X,Lk) of global holomorphic sections of Lk is represented by an integral kernel Bk which is called the Bergman kernel (with parameter k ε N). The restriction ρk : X → R of the norm of Bk to the diagonal in X × X is called the density function of Bk. On a dense subset of X, we describe a method for computing the coefficients of the asymptotic expansion of ρk as k → ∞ in this toric setting. We also provide a direct proof of a result which illuminates the off-diagonal decay behaviour of toric Bergman kernels. We fix a parameter l ε N and consider the projection Pl,k from C∞(X,Lk) onto those global holomorphic sections of Lk that vanish to order at least lk along some toric submanifold of X. There exists an associated toric partial Bergman kernel Bl,k giving rise to a toric partial density function ρl,k : X → R. For such toric partial density functions, we determine new asymptotic expansions over certain subsets of X as k → ∞. Euler-Maclaurin sums and Laplace’s method are utilized as important tools for this. We discuss the case of a polarization of CPn in detail and also investigate the non-compact Bargmann-Fock model with imposed vanishing at the origin. We then discuss the relationship between the slope inequality and the asymptotics of Bergman kernels with vanishing and study how a version of Song and Zelditch’s toric localization of sums result generalizes to arbitrary polarized Kahler manifolds. Finally, we construct families of induced metrics on blow-ups of polarized Kahler manifolds. We relate those metrics to partial density functions and study their properties for a specific blow-up of Cn and CPn in more detail.