Long noncoding RNA (lncRNA) have been identified in all eukaryotes but the functions and mechanisms of many lncRNA are unknown, and remain challenging to study in vivo. Mammalian X-Chromosome Inactivation (XCI) is regulated by lncRNA and has emerged as a model system in which to study lncRNA function. The "master regulator" of XCI is lncRNA Xist, which coordinates epigenetic silencing one of the two X chromosomes in females. While the primary Xist activating factor has been under debate, the lncRNA Jpx has emerged as a proposed activator. However, Jpx's function and mechanism have been unclear due to conflicting findings in mouse embryonic stem cell models.

I hypothesized that Jpx activates Xist expression in vivo, and that it would do so using both trans and cis genetic mechanisms. To test this, I utilized two transgenes, Tg(Jpx) and Tg(Jpx, Xist), to develop a novel Jpx transgenic mouse model and study the function of Jpx in mouse XCI. I found that transgenic Jpx mice were viable and fertile, although transgenic male viability was lower in certain Tg(Jpx) lines. Using RT-qPCR and Fluorescence in situ Hybridization (FISH) in mouse embryonic fibroblasts (mEFs) and post-implantation embryos, I observed a dose-dependent relationship between Jpx copy number and Xist expression in transgenic mice. This suggests that Jpx is sufficient to activate Xist expression in vivo. Further, I analyzed Jpx's allelic origin with RNA and DNA FISH and observed Jpx using both trans and cis mechanisms to activate Xist expression. RT-qPCR showed reduced X-linked gene expression in male and female Tg(Jpx) mEFs, suggesting that ectopic Xist can induce XCI. The presence of XCI in the male may explain the reduction in transgenic male viability. Overall, I have demonstrated that Jpx is a trans- and cis-acting Xist activator in mice. In the discussion, I have provided suggestions for future studies, including live cell imaging of the Jpx transcript and development of a Jpx knockout mouse model. The work presented here provides a framework for lncRNA functional studies in mice, and will help us understand how lncRNA regulate eukaryotic gene expression.

If we are to understand how nature formulates complex animal behavior, we must develop the tools that have the capabilities to capture these processes in their full capacity. I have created Transverse-Sheet Illumination Microscopy (TranSIM), a highly customizable, single objective microscope, with micron level resolution (0.7 um, 1.1 um, 1.6 um in x, y, and z, respectively) with the ability to image extended three dimensional volumes (750 um x 560 um x 160 um) at 200 volumes per second. For Danio rerio, approximately 10^5 neurons were imaged in 3D, opening an unparalleled view into the animal's complex processes at near millisecond temporal resolution. Fast intra-planar and inter-planar processes were observed signaling that utility of TranSIM. Overall, a 10-fold increase in the total volumetric imaging rate is achieved compared to any system currently available, and still with ample room for improvement.

Inflation forecasts of the Federal Reserve systematically under-predicted inflation before Volcker and systematically over-predicted it afterward. Furthermore, under quadratic loss, commercial forecasts have information not contained in those forecasts. To investigate the cause, this paper recovers the loss function implied by Federal Reserve’s forecasts. It finds that the cost of having inflation above an implicit time-varying target was larger than the cost of having inflation below it for the period since Volcker, and that the opposite was true for the pre-Volcker era. Once these asymmetries are taken into account, the Federal Reserve is found to be rational. (JEL C53, E52)

This dissertation analyzes the changes that the tobacco trade between southeastern Europe and Germany underwent from the late nineteenth century, until the beginning of World War II. Such changes affected the distribution of economic activities across geographic space, the types of actors involved in these activities, and the labor processes that were necessary at each node of the commercial chain. I argue that these developments were concrete manifestations of two broader, inter-related historical processes that took place in the 1920s and 1930s: the expansion of Greek state authorities into new areas of economic life, and the integration of Greece into a trading bloc with Germany at its center. I also argue that these historical processes had far-reaching effects on the lives of Greek peasants, urban workers, and merchants, as well as on the relationship between the Greek state and its population. By focusing on a specific transnational commodity, this dissertation adds to our understanding of how the processes of economic integration and internal expansion of state power manifest themselves in the material and spatial dimensions of production and trade.

The evidence presented here has been drawn from archival material produced by banks, state agencies, and tobacco companies. There are also abundant references to Greek legislation, as well as published sources such as press articles and journals specialized in the tobacco industry. The evidence has been weaved into a coherent narrative through the use of conceptual tools drawn from the literature on commodity chains and value chains.

We propose a novel training approach for neural networks based on switching among an array of feedback controllers (FC). Traditionally, the neural network training implemented in reinforcement learning problems can be achieved with observable kinematic variables (KV). In this work, our training approach takes a step further by using the metrics-only (MO) or LTM inputs into the networks where each metric and FC fulfills a design specification. Alongside, the reinforcement learning algorithm is a hierarchical control architecture for switching among multiple FCs and its modes. With the designed FCs based on Lyapunov functions or stochastic optimal control, we show that the MO training has a faster convergence with less variations than the one that is based on KV. These results are important for applications requiring large number of controllers. We provide results for a pendulum control problem, a bicycle, and a tricycle navigation problem. In the case of the tricycle problem, we also show that the trained neural network can be applied beyond numerically simulated control problems. The results of this work are illustrated by numerical and virtual reality simulations.