12023 research outputs found
Sort by
Synthesis of Organic Building Blocks and Synthetic Strategies Toward Aleutianamine
Get PDFDescribed herein are the development of the palladium-catalyzed decarboxylative asymmetric allylic alkylation of medicinally relevant 5- and 7-membered diazaheterocycles and efforts toward the total synthesis of the cytotoxic pyrroloiminoquinone marine alkaloid aleutianamine. The former methodology provides a new tactic to incorporate Csp3 structural complexity into future lead compounds containing diazepane and imidazolidine moieties. The latter project is ongoing.
Additionally, Chapter 3 discusses preliminary attempts to improve the synthetic accessibility of minimally substituted corroles, which were conducted during a research internship in the laboratory of Prof. Zeev Gross at the Technion. During the course of this research, the first example of a β-unsubstituted free base monoazaporphyrin was isolated, and its cobalt complex was characterized by x-ray crystallography.
Finally, Appendix 8 presents a series of cationic and radical-mediated fragmentations of a derivative of (+)-3-Carene, a chiral pool material. These experiments led to the observation and mechanistic study of an unexpected rearrangement.</p
Visual and Spatial Representation Learning with Applications in Ecology
Get PDFMachine learning has the potential to empower scientists, physicians, and other human experts working to solve problems of societal importance. To realize this goal, we need algorithms that can distill useful knowledge from real-world data. However, most machine learning research focuses on benchmarks that seldom reflect real-world challenges, such as learning from limited, noisy, or weak supervision. This thesis develops new benchmarks, algorithms, and problem settings that link fundamental machine learning research to impactful applications in ecology. In Part I, we provide context and motivation for our work. How and why should machine learning researchers work with domain experts on real-world problems? What is the appeal of ecology specifically Part II focuses on visual representation learning with an emphasis on label efficiency. We discuss the strengths and limitations of self-supervised learning, the relationship between concept specificity and representation learning, and multi-label learning with minimal labeled data. Part III covers our work in the emerging field on spatial representation learning. In particular, we consider the problem of modeling the spatial distribution of plant and animal species. We review this important ecological problem from a machine learning perspective before showing how deep learning can transform the way these models are applied (using spatial models to assist image classifiers) and developed (jointly learning spatial distributions and representations). Finally, Part IV concludes and highlights opportunities for future work
MFeS Clusters as Models for Complex Multimetallic Systems
Get PDFThe impressive chemistry of the FeMco-factor of nitrogenase is still under intense study with many remaining mechanistic questions about the important transformation of N2 to NH3, including the role of the synthetically interesting carbide ligand and the complex structure of the octanuclear active site cluster. This thesis describes efforts to incorporate bridging carbon based ligands into synthetic MFeS clusters, which remains a significant synthetic challenge in inorganic chemistry. The subsequent characterization and reactivity studies of such clusters provide insight into how cluster structural properties, including ligand identity, metal identity, and cluster composition and geometry affect cluster electronic properties and reactivity.
Chapter II describes the synthesis of an unprecedented WFe3S3(µ3-carbyne) cluster and comparisons with a WFe3S3(µ3-S) cluster provide insight into the role of the bridging carbide in FeMco. The generality of the developed method for synthesis of MFe3S3(µ3-carbyne) clusters is also demonstrated. Characterization of the various carbyne clusters are discussed. Cleavage of the C-Si bond of MFe3S3(µ3-CSiMe3) clusters towards bridging methylidyne clusters is also demonstrated.
Chapter III discusses the synthesis of novel octanuclear (CHn) bridged Mo2Fe6S6 clusters and describes the reactivity of these bridges in the context of organometallic and nitrogenase chemistry.
Chapter IV describes spectroscopic data of an extensive family of MFeS cubane clusters that is currently lacking in the literature and provides important structure property benchmarking information for synthetic and biological FeS clusters. Chapter V describes the synthesis of well-defined high nuclearity Fe13 clusters resulting from studies of WFeS clusters. These clusters exhibit high spin states and undergo well-defined reactivity with small molecules and thus provide an unprecedented atomically resolved study of reactivities of high-nuclearity transition metal clusters.</p
Dismantling The Help: The Hollywoodization of The Civil Rights Movement in 1963
No full text[Introduction] The Help directed by Tate Taylor and based off Kathryn Stockett’s novel portrays an incomplete narrative of Black women’s domestic work in white households through a white feminist lens under the backdrop of the 1963 Civil Rights Movement. While the movie seeks to show the perspective of these Black women often ignored throughout history, the movie instead focuses largely on an upcoming white woman writer named Skeeter. To shape Skeeter’s
ultimate writing success story, the movie utilizes the Civil Rights Movement and the Black domestic workers, mainly Aibileen and Minny, to jumpstart Skeeter’s writing career, subsequently leaving the Black community to deal with the aftermath of the publication of Skeeter’s novel. Based in Jackson, Mississippi, The Help details the racial turmoil of the time through the grossly comic portrayal of outdoor bathrooms for the domestic workers, the unrealistic arrest of Yule Mae Davis, and the turbulent assassination of Medgar Evers. While Medgar Evers killing was historically accurate, the oversimplification of white allyship, Black
resistance, and the softening of segregation and discrimination allow the audience to feel a false sense of accomplishment at the end of the movie which would realistically end in bloodshed. Although The Help builds its story around various key aspects of the Civil Rights movement in 1963, the movie substantially underplays the intensity of discrimination against Black people by heavily filtering the movement to craft an agreeable Hollywood narrative
Manufacturing 3-D Lithium-Ion Batteries with Interpenetrating Lattice Electrodes
Get PDF3-D lithium-ion batteries have been proven to exhibit a higher energy density while minimizing power loss compared to the standard, layer-by-layer constructed 2-D lithium-ion batteries. This thesis explores the implementation of additive manufacturing in the process of constructing the proposed 3-D battery due to its capability of architecting materials with high accuracy and tunability. A 3-D lithium-ion battery backbone was created using a 2-step process, in which the first step 3-D printed the overall structure as a polymer, and the second step sputtered gold onto the polymer for conductive properties. The 3-D printed battery backbone consisted of two interpenetrating lattices made of post-cured PR48 resin that would serve as the anode and cathode, while the electrolyte would fill the space between the two electrodes. During the sputtering process, the polymer structure was rotated 6 times to guarantee that the sputtering will be conformal throughout the lattice. Electrodeposition was used to generate a LiCoO2 anode and a Li cathode. The electrodeposition of the lithium cobalt oxide cathode onto the lattice structure was proven to be unsuccessful due to the low thermal stability and high reactivity of the 3-D printed polymer when submerged into the electrolyte, consisting of KOH at 260 °C. Results indicate that the uniform electrodeposition of the lithium anode onto the lattice structure was successful using a 1s on, 1s off pulse current for a 60-minute duration. Using a titanium and gold layer proved to increase the uniformity of the coating. However, due to the failure of the lithium cobalt oxide electrodeposition, a different backbone structure may need to be considered. Having two separate structures serving as the anode and cathode (and later combining them into one structure) as opposed to both electrodes being on one structure may be beneficial. This allows for the cathode to be altered without altering both electrodes, allowing more flexibility to coat the structure with lithium cobalt oxide
Gut Microbiota Modulation of Host Feeding Behavior
Get PDFThe rich, diverse community of microorganisms in the gastrointestinal tract of animals, or gut microbiota, regulates aspects of host metabolism, immunity, and neural function, with resulting effects on the expression of complex behaviors, including feeding.
In this thesis, we sought to characterize gut microbiota influences on the behavioral response to palatable foods in mice. We discover that binge-like consumption of palatable foods, including high-sucrose pellets and a high-fat diet, is exacerbated in mice in the absence of a gut microbiota. Furthermore, using automated feeding dispensers and video analysis, we find that microbiota depletion with oral antibiotics results in elongated feeding bouts and conserved changes in the dynamics of palatable food intake. We show the hyperphagic phenotype of antibiotic-treated mice is reversible upon microbiota reconstitution via fecal microbiota transplant. Operant conditioning tests reveal that the motivation to pursue high-sucrose rewards is augmented in microbiota-depleted mice. The mesolimbic brain region activity induced upon high-sucrose pellet consumption is elevated in antibiotic-treated mice. Gut bacteria from the family S24-7 and genus Lactobacillus were identified by differential antibiotic treatment and fecal microbiota transplants as correlating with reduction of high-sucrose pellet consumption. Indeed, colonization of vancomycin-treated mice with a mixture of S24-7 and Lactobacillus johnsonii reduces overconsumption of high-sucrose pellets in a limited-access binge-eating model. The work in this thesis comprehensively demonstrates that the gut microbiota regulates feeding induced in response to palatable foods in mice.</p
Physics-Informed Neural Approaches for Multiscale Molecular Modeling and Design
Get PDFChemical processes in nature span multiple characteristic length and time scales, and the computational simulation for systems at the intersection of different scales is highly challenging with far-reaching implications for numerous scientific and industrial problems. To facilitate the computational modeling and design for large molecular systems and address the cost-resolution tradeoffs in conventional strategies, in this dissertation we introduce a series of physics-informed machine learning methods for the efficient computational modeling of chemical systems and the accurate prediction of their properties such as energetics, structures, and dynamics. In Chapters 2-3, we introduce a family of orbital-based geometric deep learning methods for the prediction of quantum chemical properties while adhering to the scaling and symmetry constraints of electronic structure theory. The presented methods achieve a chemical accuracy on community-wide benchmarks for molecular property prediction, and are shown to be transferable among diverse main-group molecular systems. In Chapter 4, we introduce a method for the prediction of protein-ligand complex structures based on a finite-time stochastic process parameterized by deep equivariant neural networks. The presented method achieves improved structure prediction accuracy against existing approaches, and is able to rapidly sample protein structures for folding landscapes that are modulated by inter-molecular interactions.</p
Graph Modeling for Genomics and Epidemiology
Get PDFThe last decades have seen great leaps made in the development of RNA sequencing technologies, yielding lower cost and greater throughput of experiments, to the point where the scale of the data produced on a daily basis is staggering. While computational hardware is also continuously improving, famously (or perhaps infamously) described by Gordon Moore (Moore, 1965), the rate at which data are produced eclipses advances on the hardware front. Over the last few years, many new methods have been proposed for bridging that ever-widening chasm, more than a few of which harness the latent graphical structure of genomic data to reduce the number of calculations required and pack the data tighter in memory. This body of work continues this development on three different, but related, fronts. Firstly, I present developments that greatly improve upon the efficiency of state-of-the-art methods for the quantification of RNA-seq reads, and describe a method that improves the accuracy of quantification without substantially increasing the computational over- head. Secondly, I introduce a procedure for the discovery of associations between novel gene isoforms and phenotypes, without prior knowledge of those isoforms. Lastly, I present the largest reconstruction of the transmission tree of a viral outbreak to date, modeled from viral genome sequences, contact tracing, and symptom data. I then use the reconstructed transmission tree to assess the efficacy of different vaccination strategies
Agency Problems in Political Science
Get PDFThis dissertation consists of three chapters analyzing agency problems in political science. More specifically the role of different/additional information available to the principal or the agent.
In Chapter 2 we analyze the effect of the politician's knowledge of the external shock on his policy decisions on unrelated issues. Elected politicians cannot control some external shocks, even if they can still anticipate their occurrence better than the general public. How can politicians use these types of anticipated external shocks to their benefit? How do they change their pandering incentives? And how does a rational voter incorporate these seemingly irrelevant external shocks in her voting decision? We build on the political accountability model of Canes-Wrone Herron, Shotts (2001), adding the ability to the voter to observe her utility, which is affected by an external shock. The shock is observed by the incumbent politician but not by the voter. Our analyses show that for high or low enough magnitude external shocks, a politician's ability to anticipate them eliminates his pandering incentives in equilibrium. For medium negative shocks, pandering could be a "gamble for resurrection," while for medium positive shocks, it acts as an "insurance" to guarantee the reelection. We show that both of these pandering regions emerge in equilibrium. The politician's knowledge of the shock, overall, decreases the voter's welfare in equilibrium.
In Chapter 3 we endogenize the information acquisition for the voter to study what types of policy decisions voters pay attention to, and why, and how rational voter attention affects the behavior of politicians in office. We extend the Canes-Wrone, Herron, Shotts (2001) model of electoral agency to allow the voter to rationally choose when to ``pay attention'' to an incumbent's policy choice by expending costly effort to learn its consequences. When attention is moderately costly the voter generally pays more of it after the ex-ante unpopular policy than the ex-ante popular one. Rational attention may improve accountability by encouraging the politician to be truthful. In some cases, it may also severely harm accountability both by inducing a strong incumbent to ``play it safe'' with a policy that avoids attention, or a weak incumbent to ``gamble for resurrection'' with a policy that draws it. Finally, rational attention can induce or worsen pandering but never ``fake leadership''.
Chapter 4 analyzes delegated information acquisition with a biased agent who also has private information about the state of the world. The information acquired is public and its informativeness increases with costly effort. Equilibria are characterized for two cases: when the agent decides an effort level and when a principal imposes formal requirements on it. The analysis demonstrates that even when the principal cannot commit to an arbitrary decision rule, he benefits from imposing formal requirements by getting as much public information as possible and correctly aligning the biased agent's incentives. In the optimal mechanism, the principal incentivizes the low-type agent to truthfully reveal her private information by requiring a relatively low amount of costly effort, while the high private report has to be followed by the maximum effort in the public signal.</p
Modeling and Programming Shape-Morphing Structured Media
Get PDFShape-morphing and self-propelled locomotion are examples of mechanical behaviors that can be "programmed" in structured media by designing geometric features at micro- and mesostructural length scales. This programmability is possible because the small-scale geometry often imposes local kinematic modes that are strongly favored over other deformations. In turn, global behaviors are influenced by local kinematic preferences over the extent of the structured medium and by the kinematic compatibility (or incompatibility) between neighboring regions of the domain. This considerably expands the design space for effective mechanical properties, since objects made of the same bulk material but with different internal geometry will generally display very different behaviors. This motivates pursuing a mechanistic understanding of the connection between small-scale geometry and global kinematic behaviors. This thesis addresses challenges pertaining to the modeling and design of structured media that undergo large deformations.
The first part of the thesis focuses on the relation between micro- or mesoscale patterning and energetically favored modes of deformation. This is first discussed within the context of twisted bulk metallic glass ribbons whose edges display periodic undulations. The undulations cause twist concentrations in the narrower regions of the structural element, delaying the onset of material failure and permitting the design of structures whose deployment and compaction emerge from the ribbons' chirality. Following this discussion of a periodic system, we study sheets with non-uniform cut patterns that buckle out-of-plane. Motivated by computational challenges associated with the presence of geometric features at disparate length scales, we construct an effective continuum model for these non-periodic systems, allowing us to simulate their post-buckling behavior efficiently and with good accuracy.
The second part of the thesis discusses ways to leverage the connection between micro/mesoscale geometry and energetically favorable local kinematics to create "programmable matter" that undergo prescribed shape changes or self-propelled locomotion when exposed to an environmental stimulus. We first demonstrate the capabilities of an inverse design method that automates the design of structured plates that morph into target 3D geometries over time-dependent actuation paths. Finally, we present devices made of 3D-printed liquid crystal elastomer (LCE) hinges that change shape and self-propel when heated.</p