Spelling suggestions: "subject:"agradient."" "subject:"9gradient.""
171 |
DETERMINING HEAT PRODUCTION OF BLACK SOLDERI FLY LARVAE, <em>HERMETIA ILLUCENS</em>, TO DESIGN REARING STRUCTURES AT LIVESTOCK FACILITIESMcEachern, Travis 01 January 2018 (has links)
Due to their small size and ectothermic biology, the heat production of insects and insect larvae is hard to quantify. However, knowing the amount of heat production, as well as ammonia production of insects may be beneficial for commercial production of valuable insect species. Black soldier fly larvae (BSFL) are of interest in the agricultural industry because they quickly consume organic waste and have high amounts of protein and fat in their bodies. It has been proposed that BSFL be used to manage livestock waste, while serving as a high-protein feed source for livestock animals. To efficiently rear BSFL, it is necessary to design rearing facilities, which maintain optimal conditions for the larvae. To design such a facility, it is necessary to know the amount of heat and ammonia that BSFL produce.
A gradient calorimeter was used to measure the heat and ammonia production rates of black soldier fly larvae. The study determined that BSFL heat production changes significantly with the age and weight of the larvae. Aggregations produce the most total heat when larvae are older and larger. The study also found that larvae produce less heat per individual and per gram of body weight as they grow. Larvae also produce significantly different amounts of heat depending on the size of the groups they are in, and do not produce consistent amounts of heat per individual or per gram of body weight, even if maintained at a consistent population density. Larvae in group sizes of 100, 300, and 500 produced an average and standard deviation of 0.00107±0.000295, 0.00067±0.00014, and 0.00049±0.00020 W/larva, respectively. Likewise, larvae in groups of 100, 300 and 500 produced an average of 0.01826±0.00010, 0.01023±0.00565, and 0.00575±0.00371 W/g, respectively. The differences in heat produced per individual and per gram is troublesome when trying to estimate a total heat production for large populations.
The largest heat production rate observed in this study was 0.407 W, and was produced by a group of 500 BSFL. Frass analysis indicated that between 4.80 and 7.79 lbs of ammoniacal-nitrogen is emitted for every ton of frass produced. These data could be used to estimate the total heat and ammonia produced from a larger population of BSFL being reared inside a closed facility, allowing engineers to design HVAC systems to keep the larvae at their optimal growing condition year-round. Placing BSFL rearing accommodations at livestock facilities could be beneficial to livestock, poultry, and fishery producers, because BSFL can be used to dispose of animal wastes and are also a good source of protein-rich animal feed.
|
172 |
Projected Wirtinger gradient descent for spectral compressed sensingLiu, Suhui 01 August 2017 (has links)
In modern data and signal acquisition, one main challenge arises from the growing scale of data. The data acquisition devices, however, are often limited by physical and hardware constraints, precluding sampling with the desired rate and precision. It is thus of great interest to reduce the sensing complexity while retaining recovery resolution. And that is why we are interested in reconstructing a signal from a small number of randomly observed time domain samples. The main contributions of this thesis are as follows.
First, we consider reconstructing a one-dimensional (1-D) spectrally sparse signal from a small number of randomly observed time-domain samples. The signal of interest is a linear combination of complex sinusoids at R distinct frequencies. The frequencies can assume any continuous values in the normalized frequency domain [0, 1). After converting the spectrally sparse signal into a low-rank Hankel structured matrix completion problem, we propose an efficient feasible point approach, named projected Wirtinger gradient descent (PWGD) algorithm, to efficiently solve this structured matrix completion problem. We give the convergence analysis of our proposed algorithms. We then apply this algorithm to a different formulation of structured matrix recovery: Hankel and Toeplitz mosaic structured matrix. The algorithms provide better recovery performance; and faster signal recovery than existing algorithms including atomic norm minimization (ANM) and Enhanced Matrix Completion (EMaC). We further accelerate our proposed algorithm by a scheme inspired by FISTA. Extensive numerical experiments are provided to illustrate the efficiency of our proposed algorithms. Different from earlier approaches, our algorithm can solve problems of very large dimensions very efficiently. Moreover, we extend our algorithms to signal recovery from noisy samples. Finally, we aim to reconstruct a two-dimension (2-D) spectrally sparse signal from a small size of randomly observed time-domain samples. We extend our algorithms to high-dimensional signal recovery from noisy samples and multivariate frequencies.
|
173 |
The effects of human development on avian diversity along an urban-rural gradient in Iowa City, IowaMcCurdy, Jason D. 01 August 2016 (has links)
The human population is rapidly urbanizing. While this will undoubtedly present challenges for humans it will also place pressure on birds in these areas. To better manage these spaces in a manner that promotes avian biodiversity, we must first come to understand how human development affects the distribution and abundance of bird species and guilds within cities and if patterns observed previously in large cities can be extended to smaller ones.
Breeding birds were surveyed along a gradient of increasing urbanization in Iowa City, Iowa, during the summers of 2014 and 2015. Study areas included a forested park, recreational park, low density residential area, medium density residential area, high density/mixed-use area, and urban core. Birds were censused a total of four times at each site using variable circular plot counts. Landscape characteristics were measured using a high resolution land cover dataset and tree canopy model. Regression models were developed to investigate relationships between the bird community and land cover characteristics.
Bird species richness, diversity, and evenness all decreased with increasing urbanization, while biomass and the number of individuals peaked in the urban core. The community shifted from non-native, resident, granivorous, multi-brooding building nesters in highly developed areas to native, migrant, invertivorous, single-brooding, tree and tree cavity nesters at the least developed sites. Regression models indicated varied relationships among landscape characteristics and species richness and community prevalence of functional guilds. Native, migratory, invertivorous, tree cavity-nesting, and single-brooding species showed negative relationships to variables measuring the built environment, while non-native, resident, granivorous, building nesting, and multi-brooding species showed positive relationships to these measures.
Overall, the response of avian functional guilds to varying levels of urban intensity in Iowa City were remarkably similar to the results of previous studies. This suggests that much of what has been learned previously concerning avian responses to human development can be extended to planning and implementing conservation strategies in smaller cities.
|
174 |
Phenology of a Southern Population of Mountain Pine Beetle (<i>Dendroctonus ponderosae</i>)McManis, Anne Elise 01 May 2018 (has links)
Mountain pine beetle (Dendroctonus ponderosae, Hopkins) is a major disturbance agent in pine ecosystems of western North America. Adaptation to local climates has resulted in primarily univoltine (one generation per year) generation timing across a thermally diverse latitudinal gradient. We hypothesized that this pattern in total development time is shaped by selection for slower developmental rates, altered developmental thresholds, or oviposition rates in southern populations inhabiting warmer climates. To investigate traits responsible for latitudinal differences we measured lifestage-specific development of southern mountain pine beetle eggs, larvae and pupae across a range of temperatures. We also describe and model oviposition of southern US MPB. Using a novel technique that included frequent X-ray imaging, oviposition rate and fecundity were estimated separately and shown to both be significant and independent sources of variation. When compared with previously collected data for a northern MPB population, total oviposition time predicted for southern MPB at a constant 20°C was slightly longer than that of northern MPB, but the delay was too small to account for significant differences between the populations in total development time.
Developmental rate curves for eggs, larvae, and pupae were fit using maximum posterior likelihood estimation with a Bayesian prior to improve fit stability. When compared to previously published data for a northern population (Régnière et al. 2012), observed developmental rates of the southern and northern populations were similar across all studied lifestages at 20 and 25°C, although southern individuals were generally faster at temperature extremes (10 and 27°C). These findings were inconsistent with our hypothesis that southern individuals would have consistently slower rates. Optimal development of southern individuals occurred at higher temperatures, with higher development thresholds, as compared with northern individuals. Our results suggest that evolved traits in the remaining unstudied lifestage, teneral (i.e., pre-emergent) adult, likely influence latitudinal differences in mountain pine beetle generation time.
|
175 |
Elaboration et caractérisation de composites Alumine/Zircone à vocation orthopédique / Elaboration and characterization of alumina/zirconia composites for orthopedic applicationsBiotteau, Katia 10 September 2012 (has links)
Ce travail de thèse a pour objectif l’élaboration et la caractérisation de composites Alumine/Zircone obtenus par voies conventionnelles, et dédiés à un usage orthopédique. Ces composites présentent une biocompatibilité prouvée, d’excellentes propriétés mécaniques ainsi qu’une grande stabilité. Ils sont plus résistants, plus fiables que l’alumine ou la zircone seules et permettent d’envisager des composants de tailles et formes plus exigeantes mécaniquement. Actuellement ces composites semblent les plus adaptés pour la réalisation de prothèses orthopédiques mais peuvent encore être optimisés via la modification des microstructures. La première partie de ce travail a concerné l’étude de la réalisation industrielle de composants de grande taille à partir d’une poudre. Les différentes étapes de l’élaboration sont traitées : pressage des composants, frittage et usinage. Cette première partie est majoritairement consacrée à l’étude des gradients thermique dans une sphère lors du frittage. Nous montrons qu’il est possible de modéliser et de mesurer les gradients thermiques dans le matériau de manière très réaliste, ainsi que d’obtenir des ordres de grandeur des contraintes mécaniques. On pourra ainsi envisager de tester numériquement les cycles de frittage en fonction de la géométrie des pièces frittées. Nous étudions par ailleurs la possibilité de réaliser un usinage des composants après un traitement de préfrittage, qui permettrait de diminuer les coûts et simplifier l’élaboration de composants de grande taille. La seconde partie de ce mémoire a permis de montrer que différents types de microstructures, présentant des propriétés mécaniques différentes, peuvent être obtenues par simple mélange de poudre. Ceci est possible par l’utilisation d’un traitement thermique adapté, la variation du taux de zircone et grâce à l’ajout de dopants (Si, Ca et Mg) jouant sur la mobilité des joints de grains d’alumine. Lors de l’utilisation de Ca ou Mg, le taux de zircone et la température ont un effet prépondérant sur l’aspect des microstructures, permettant d’obtenir des micro/micro-composites (< 16vol% de zircone et >1500°C) et nano/nano-composites (25vol% de zircone et T < 1500°C). Seul l’ajout conjoint de Si et de Ca pour des échantillons contenant 2,5vol% de zircone permet de conduire à des micro/nano-composites avec une grande proportion de zircone intragranulaire. Les observations sur des composites avec un taux de zircone proche du taux de percolation (16vol%) permettent de mettre en évidence l’ensemble des types de renforcement observés dans la littérature, en fonction de la température et des dopants utilisés. La variété des microstructures obtenues permet de progresser dans le contrôle des microstructures des composites alumine-zircone, mais aussi d’envisager d’autres applications de ces composites en fonction des mécanismes de renforcement observés et de leurs propriétés mécaniques et structurales. / The aim of this work was to elaborate and characterize zirconia toughened alumina composites with different microstructures, using a simple process. These composites are obtained by colloidal process and are dedicated to orthopedic application. ZTA composites offer both higher strength and toughness than alumina, a lower sensitivity to ageing than zirconia, and also a proven biocompatibility. They open the door to component designs not reachable with other, more brittle materials. Nowadays, these composites are the safest for orthopedic implants application, but can still be improved. The first part of our study is dedicated to a numerical modeling of a large femoral head during sintering. It is so possible to obtain a realistic model of thermal and mechanical strain gradient. However, the modeling should be enhanced by a thorough study of the elastic-viscous-plastic behavior of the composite at high temperature. Then some experiments of sintering with various load or pressure and speed should be practice to determine precisely the sintering related strain. The possibility of machining in the pre-sintered state, with the aim of reducing machining costs and simplify the process of large components, is investigated. The second part is focused on the development of various microstructures with specific mechanical properties and reinforcement behaviors. Such structures were achieved by adjusting the amount of zirconia, controlling the grain growth with dopants to improve or inhibit the alumina grain growth (Si, Ca and Mg), and by adjusting the sintering thermal treatment. The use of calcium associated or not with magnesium seems useless as compared to the predominant influence of the zirconia content. Adjusting zirconia amount led to micro-composites (< 16vol% of zirconia and >1500°C) and nano-composites (25vol% of zirconia and < 1500°C). Only Si/Ca co-doped ZTA composites with small amount of zirconia (2.5vol %) leads to almost homogeneous micro/nano composites with a large proportion of intragranular zirconia particles at high temperature. Around the percolation threshold (16vol% of zirconia) all types of reinforcement mechanisms that could be observed in ZTA composites (referred to literature) can be observed, depending on the thermal treatment and the dopants used. The range of microstructures obtained in this study leads us to investigate other applications for these composites depending of its reinforcement behavior and its mechanical and structural properties.
|
176 |
Policy-Gradient Algorithms for Partially Observable Markov Decision ProcessesAberdeen, Douglas Alexander, doug.aberdeen@anu.edu.au January 2003 (has links)
Partially observable Markov decision processes are interesting because
of their ability to model most conceivable real-world learning
problems, for example, robot navigation, driving a car, speech
recognition, stock trading, and playing games. The downside of this
generality is that exact algorithms are computationally
intractable. Such computational complexity motivates approximate
approaches. One such class of algorithms are the so-called
policy-gradient methods from reinforcement learning. They seek to
adjust the parameters of an agent in the direction that maximises the
long-term average of a reward signal. Policy-gradient methods are
attractive as a \emph{scalable} approach for controlling partially
observable Markov decision processes (POMDPs).
¶
In the most general case POMDP policies require some form of internal
state, or memory, in order to act optimally. Policy-gradient methods
have shown promise for problems admitting memory-less policies but have
been less successful when memory is required. This thesis develops
several improved algorithms for learning policies with memory in an
infinite-horizon setting. Directly, when the dynamics of the world
are known, and via Monte-Carlo methods otherwise.
The algorithms simultaneously learn how to act and what to remember.
¶
Monte-Carlo policy-gradient approaches tend to produce gradient
estimates with high variance. Two novel methods for reducing variance
are introduced. The first uses high-order filters to replace the
eligibility trace of the gradient estimator. The second uses a
low-variance value-function method to learn a subset of the parameters
and a policy-gradient method to learn the remainder.
¶
The algorithms are applied to large domains including a simulated
robot navigation scenario, a multi-agent scenario with 21,000 states,
and the complex real-world task of large vocabulary continuous speech
recognition. To the best of the author's knowledge, no other policy-gradient
algorithms have performed well at such tasks.
¶
The high variance of Monte-Carlo methods requires lengthy simulation
and hence a super-computer to train agents within a reasonable time. The ANU
``Bunyip'' Linux cluster was built with such tasks in mind. It was
used for several of the experimental results presented here. One
chapter of this thesis describes an application written for the Bunyip
cluster that won the international Gordon-Bell prize for
price/performance in 2001.
|
177 |
Policy Gradient Methods: Variance Reduction and Stochastic ConvergenceGreensmith, Evan, evan.greensmith@gmail.com January 2005 (has links)
In a reinforcement learning task an agent must learn a policy for performing actions so as to perform well in a given environment. Policy gradient methods consider a parameterized class of policies, and using a policy from the class, and a trajectory through the environment taken by the agent using this policy, estimate the performance of the policy with respect to the parameters. Policy gradient methods avoid some of the problems of value function methods, such as policy degradation, where inaccuracy in the value function leads to the choice of a poor policy. However, the estimates produced by policy gradient methods can have high variance.¶
In Part I of this thesis we study the estimation variance of policy gradient algorithms, in particular, when augmenting the estimate with a baseline, a common method for reducing estimation variance, and when using actor-critic methods. A baseline adjusts the reward signal supplied by the environment, and can be used to reduce the variance of a policy gradient estimate without adding any bias. We find the baseline that minimizes the variance. We also consider the class of constant baselines, and find the constant baseline that minimizes the variance. We compare this to the common technique of adjusting the rewards by an estimate of the performance measure. Actor-critic methods usually attempt to learn a value function accurate enough to be used in a gradient estimate without adding much bias. In this thesis we propose that in learning the value function we should also consider the variance. We show how considering the variance of the gradient estimate when learning a value function can be beneficial, and we introduce a new optimization criterion for selecting a value function.¶
In Part II of this thesis we consider online versions of policy gradient algorithms, where we update our policy for selecting actions at each step in time, and study the convergence of the these online algorithms. For such online gradient-based algorithms, convergence results aim to show that the gradient of the performance measure approaches zero. Such a result has been shown for an algorithm which is based on observing trajectories between visits to a special state of the environment. However, the algorithm is not suitable in a partially observable setting, where we are unable to access the full state of the environment, and its variance depends on the time between visits to the special state, which may be large even when only few samples are needed to estimate the gradient. To date, convergence results for algorithms that do not rely on a special state are weaker. We show that, for a certain algorithm that does not rely on a special state, the gradient of the performance measure approaches zero. We show that this continues to hold when using certain baseline algorithms suggested by the results of Part I.
|
178 |
Accélération d'ions par interaction laser-matière en régime de ultra haut contraste laserLévy, Anna 26 September 2008 (has links) (PDF)
Grâce au développement récent des lasers de puissance, le domaine de l'interaction laser-matière a beaucoup évolué, ouvrant depuis quelques années, entre autres, la possibilité de générer des faisceaux d'ions à des énergies de plusieurs MeV. Soumise à de très fortes intensités laser (> 1018 W/cm2), une cible solide est rapidement transformée en plasma. Une fraction de sa population électronique est accélérée à des vitesses relativistes et à l'origine de la génération d'ions à des énergies très élevées. Les faisceaux de particules ainsi produits affichent des caractéristiques particulièrement intéressantes pour différentes applications (diagnostics de plasmas denses, fusion inertielle, protonthérapie...). Cependant, malgré la montée en puissance des installations laser, pour nombre d'entre elles, les énergies atteintes restent aujourd'hui insuffisantes, suscitant un important effort de recherche visant à optimiser les processus d'accélération. En se plaçant dans un régime d'interaction de ultra-haut contraste, nous avons étudié l'influence de l'épaisseur de la cible sur l'accélération des ions et dégager l'intérêt de l'interaction avec des feuilles minces. Au-delà d'un accroissement de l'énergie et du nombre d'ions accélérés par réduction de l'épaisseur de la cible, nous avons mis en évidence les caractéristiques de chaque étape de l'interaction. De plus, en tirant parti de ces conditions optimales d'interaction, nous avons étudié l'influence d'un gradient ionique en face arrière de la cible sur l'accélération des ions, conditions dans lesquelles intervient un mécanisme fin de déferlement ionique.
|
179 |
Modélisation de l'endommagement en fatigue des superalliages monocristallins pour aubes de turbines en zone de concentration de contrainteKaminski, Myriam 23 November 2007 (has links) (PDF)
Les aubes de turbine Haute Pression des moteurs d'avion, constituées du superalliage monocristallin AM1, sont refroidies par un circuit complexe de microcanalisations, situé en bord d'attaque et près du bord de fuite. Il peut constituer des sites privilégiés d'endommagement et d'amorçage de fissures, qu'il est indispensable de prendre en compte dans le dimensionnement en fatigue des aubes de turbines. Ce travail a consisté, dans un premier temps, à réaliser une étude expérimentale sur des éprouvettes perforées de trous de différents diamètres. Elle a permis de mettre en évidence qualitativement et quantitativement l'effet du gradient de contrainte sur l'amorçage de fissure. Cependant, la contrainte locale maximale (ou la déformation maximale) ne peut suffire dans un critère de rupture, car elle surestime le risque de rupture, ne prenant pas en compte les effets d'échelle ou de géométrie. Une méthode de moyenne volumique a alors été proposée pour prendre en compte le gradient de contrainte dans le calcul de durée de vie et a ainsi permis d'améliorer les prévisions de durée de vie.Parallèlement, un modèle d'endommagement de fatigue anisotrope, qui couple plasticité et endommagement, en s'attachant particulièrement à décrire la phase de micro-amorçage, a été développé. L'identification et la validation du modèle ont été réalisées à partir d'une base de données expérimentales conséquente sur l'AM1 et sur les essais sur éprouvettes perforées réalisés au cours de ce travail. Afin de prendre en compte les effets de gradient dans les zones à forte concentration de contrainte, la méthode de moyenne volumique a été appliquée avec ce modèle et a fourni des résultats encourageants.
|
180 |
Upper gradients and Sobolev spaces on metric spacesFärm, David January 2006 (has links)
<p>The Laplace equation and the related p-Laplace equation are closely associated with Sobolev spaces. During the last 15 years people have been exploring the possibility of solving partial differential equations in general metric spaces by generalizing the concept of Sobolev spaces. One such generalization is the Newtonian space where one uses upper gradients to compensate for the lack of a derivative.</p><p>All papers on this topic are written for an audience of fellow researchers and people with graduate level mathematical skills. In this thesis we give an introduction to the Newtonian spaces accessible also for senior undergraduate students with only basic knowledge of functional analysis. We also give an introduction to the tools needed to deal with the Newtonian spaces. This includes measure theory and curves in general metric spaces.</p><p>Many of the properties of ordinary Sobolev spaces also apply in the generalized setting of the Newtonian spaces. This thesis includes proofs of the fact that the Newtonian spaces are Banach spaces and that under mild additional assumptions Lipschitz functions are dense there. To make them more accessible, the proofs have been extended with comments and details previously omitted. Examples are given to illustrate new concepts.</p><p>This thesis also includes my own result on the capacity associated with Newtonian spaces. This is the theorem that if a set has p-capacity zero, then the capacity of that set is zero for all smaller values of p.</p>
|
Page generated in 0.0572 seconds