This paper illustrates how classroom small talk between a teacher and students constitutes a distinct interaction pattern which varies significantly from pedagogical discourse of an institutional nature such as the initiation/response/feedback (IRF) pattern described in previous literature (Mehan, 1979; Sinclair & Coulthard, 1975). By presenting a piece of extended small talk in an ESL secondary classroom in Hong Kong and contrasting it with a piece of typical teacher-orchestrated institutional classroom talk, I show how the teacher and students demonstrate more dynamic and less asymmetrical roles during small talk with clear evidence of active contributions to the exchange by the students in terms of topic setting, turn initiation, turn development, and negotiation of meaning. Features of the small talk resemble everyday social discourse. Implications of this kind of classroom talk on the learners’ L2 language development are explored.

This dissertation comprises of three chapters on macroeconomics, monetary policy, and Bayesian econometrics. The first and third chapters make use of financial time series and traded market price data that explicitly represent allocation choices and beliefs of rational capital market agents, and apply Bayesian modeling and estimation to answer questions about monetary policy and optimal asset allocation. The second chapter examines the concept of negative borrowing rates in light of central bank actions taken during the COVID crisis using a theoretical model. A common thread of the three essays is the exploration of financial markets and the global macroeconomic concepts of the day.

Chapter 1 is an empirical analysis of the influence of market expectations on the Federal Reserve’s monetary policy decisions. There is a well-established literature using the unanticipated component of fed funds target changes to identify monetary policy shocks. However, the dimensions through which ex ante market expectations affect Fed policy decisions themselves have not been carefully studied. I explore their effects on 127 FOMC target announcements between 1994 and 2008 using Bayesian estimation of a two-equation ordered probit model. I find strong evidence that market expectations have a meaningful influence on FOMC outcomes. These findings are robust across a variety of model specifications and empirically show the magnitude of financial markets’ sway on FOMC policy decisions.

The second chapter examines the efficacy of implementing dual interest rates as a novel unconventional monetary policy tool. Global central banks have typically responded to the zero lower bound with unconventional monetary policies such as forward guidance and quantitative easing. During the COVID-19 pandemic, the European Central Bank offered term lending programs at minus 1 percent, which is below that of their deposit facility, implicitly introducing a novel tool – a separately targeted economy-wide benchmark lending rate. Using a financial accelerator model, I analyze the effects of removing the ZLB on lending rates while deposit rates are kept floored at zero and show that the added ability to support asset prices vis-à-vis aggregate net worth helps greatly diminish output shortfalls due to the ZLB constraint.

The last chapter introduces a novel, unifying structural dynamic factor model of asset returns. Principal component analysis is applied on a large breadth of macroeconomic indicators to estimate latent factors representing economic activity, inflation, interest rates, trade, and volatility. The returns of six asset classes are jointly modeled using the macro factors in a time-varying parameters with stochastic volatility model and estimated using Bayesian techniques. Monthly unconditional covariance matrices are generated from October 1993 to September 2022 and used to produce cross-asset correlation estimates. Out-of-sample testing is conducted on a minimum variance asset allocation strategy based on the model covariance matrix estimator and shown to produce superior Sharpe ratio performance over other benchmark portfolio optimization heuristics. As an additional empirical application of the model, I apply k-means clustering on the macro factors to characterize shifting cross-asset correlations across different market regimes.

The convective cells are observed in the auroral ionosphere and they could play an important role in the nonlinear interaction of Alfv {e}n waves and disrupt the kinetic Alfv {e}n wave (KAW) turbulence. Zonal fields, which are analogous to convective cells, are generated by microturbulence and regulate microturbulence inside toroidally confined plasmas. It is important to understand the role of convective cells in the nonlinear interaction of KAW leading to perpendicular cascade of spectral energy. A nonlinear gyrokinetic particle simulation has been developed to study the perpendicular spectral cascade of kinetic Alfv {e}n wave. However, convective cells were excluded in the study. In this thesis project, we have modified the formulation to implement the convective cells to study their role in the nonlinear interactions of KAW. This thesis contains detail description of the code formulation and convergence tests performed, and the simulation results on the role of convective cells in the nonlinear interactions of KAW. In the single KAW pump wave simulations, we observed the pump wave energy cascades to waves with shorter wavelengths, with three of them as dominant daughter waves. Convective cells are among those dominant daughter waves and they enhance the rate of energy transfer from pump to daughter waves. When zonal fields are present, the growth rates of the dominant daughter waves are doubled. The convective cell (zonal flow) of the zonal fields is shown to play a major role in the nonlinear wave interaction, while the linear zonal vector potential has little effects. The growth rates of the daughter waves linearly depends on the pump wave amplitude and the square of perpendicular wavenumber. On the other hand, the growth rates do not depend on the parallel wavenumber in the limit where the parallel wavenumber is much smaller than the perpendicular wavenumber. The nonlinear wave interactions with various perpendicular wavenumbers are also studied in this work. When convective cells are excluded, the nonlinear wave interactions show exponential growth on the daughter waves, but at a rate about half of that of the wave interactions with convective cells. In the two pump wave simulations, six daughter waves dominate in the energy cascade process, and three of them are convective cells. The growth rates of the daughter waves are doubled compared with the growth rates of the daughter waves generated in single KAW pump wave simulation. The relationship between the growth rates of the daughter waves and pump wave parameters are studied. The growth rates of the daughter waves have a linear relationship with both pump wave amplitudes and the square of perpendicular wavenumber of the pump waves. On the other hand, the growth rates do not depend on the parallel wavenumber of the pump waves in the limit where the parallel wavenumber is much smaller than the perpendicular wavenumber. The growth rate dependence on one of the two pump waves shows that the time variation on the pump wave amplitudes must be considered.

Studies of fundamental interactions have played a principle role in the development of our understanding of particle physics, but the \ac{SM} remains incomplete.The phenomenon of neutrino oscillation lies outside of the prescriptions of the \ac{SM}, driving current experimental efforts. The \acf{DUNE} is being built to provide a comprehensive study of neutrino oscillation and to search for \ac{BSM} physics. To accomplish this, \ac{DUNE} utilizes a massive-scale far detector, a sophisticated near detector, and a high-intensity, broad-spectrum neutrino beam. The near detector adopts a new design of \ac{LArTPC} with pixelated charge readout and a modular structure to resolve events in the high-intensity neutrino beam. A prototype of this detector design was successfully operated at the University of Bern, Switzerland in 2021. From the cosmic-ray data collected with this prototype, a study of muon capture on argon has been performed, resulting in the first measurement of the Huff factor for argon ($R_H = 1.29 \pm 0.15$) and measurements of the muon capture rate ($\lambda_c = 1.53\pm0.19~\mu$s$^{-1}$) and muon disappearance rate ($\lambda_d = 2.11\pm0.24~\mu$s$^{-1}$) in argon. These results can help to reduce the uncertainty in neutrino-nucleus cross-section modelling at low $Q^2$.

The Daya Bay reactor neutrino experiment measures the oscillation of electron antineutrinos produced by the Daya Bay and Ling Ao-I & II Power Plants (totaling 17.4 GW$_\mathrm{th}$) in southern China. The fluxes and spectra of the antineutrinos are observed using eight functionally identical gadolinium-doped liquid scintillator detectors, with a total target mass of 160 tons, divided among two near and one far experimental halls, resulting in baselines of order 500 m (near) and 2 km (far). The oscillation phenomenon appears as a deficit of electron antineutrinos at the far hall relative to the yield at the near halls.

This work describes a spectral measurement of the neutrino mixing parameters $\sin^2 2\theta_{13}$ and $\Delta m^2_{ee}$ using a six-year sample of over 2 million inverse beta-decay (IBD) events identified via the outgoing neutron's subsequent capture on gadolinium. A defining feature of this analysis is its ability to be efficiently repeated for arbitrary choices of the IBD cuts, that is, the criteria used for selecting IBD events. This required the implementation of a flexible and efficient IBD selection, which was coupled to a parallelized upgrade of the official Daya Bay oscillation fitter developed at Lawrence Berkeley National Laboratory. The generalization of the analysis to arbitrary IBD cuts (beyond the two nominal IBD cuts used in official Daya Bay results) involved studies of various efficiencies and backgrounds and their dependence on the cuts, which we describe here.

With this infrastructure, we are able to explore how the best-fit oscillation parameters and their measured uncertainties vary as functions of the cuts, enabling, first, the characterization of the systematic uncertainty associated with the freedom to vary the cuts, and second, the exploration of whether the nominal IBD cuts should be modified in order to gain a reduced uncertainty on the oscillation parameters. We find that, compared to the two nominal cuts, there is no alternative IBD cut that provides a substantial reduction in uncertainty. The cut-variation study suggests the assessment of an additional systematic uncertainty of 0.0006 on $\sin^2 2\theta_{13}$ and $1.9 \times 10^{-5}$ eV$^2$ on $\Delta m^2_{ee}$, leading to a 2% increase in the total uncertainty of $\sin^2 2\theta_{13}$ and a 4% increase in that of $\Delta m^2_{ee}$. Our final measurement is $\sin^2 2\theta_{13} = 0.0850 \pm 0.0030$ and $\Delta m^2_{ee} = (2.5010 \pm 0.0072) \times 10^{-3}$ eV$^2$ in the standard three-flavor model.

The methods developed here serve to demonstrate that our oscillation fit is both robust and optimal with respect to variations in the cuts. Given the overall similarity between our oscillation analysis and those used in official Daya Bay results, similar conclusions can be drawn for the latter. Our findings thus validate past Daya Bay results, while future results will be able to benefit from this groundwork, enabling additional validation of the results and broadened assessment of their systematic uncertainties.

A major goal of in vivo imaging is to obtain individualized structural, functional, and molecular information to provide personalized medicine. In particular, optical imaging uses non-ionizing radiation to provide functional information such as hemoglobin concentration and also visualize exogenous contrast agents as well as molecular and functional markers. Indeed, in vivo optical imaging extends across a wide range of applications, from cellular to organ levels. At high end of the spectrum, diffuse optical tomography (DOT) can penetrate up to 10 centimeters but only offer low-resolution images (> 5mm) due to highly scattering nature of the tissue.

Significant effort has been spent on multi-modality imaging techniques to improve the resolution of DOT. While combining DOT with spatial information defined by a separate anatomical imaging modality is promising, there are many challenges and inaccuracies that arise with co-registration. However this can be overcome with a novel multi-modality imaging technique, Photo-Magnetic Imaging (PMI) which uses MRI as a detector to provide both high resolution anatomical and optical information. PMI takes advantage of the 3D measurement capabilities of MR thermometry (MRT) to non-invasively measure the temperature increase of the medium induced by a laser to acquire a temperature map of the entire volume. As DOT measures the photon flux only from the boundary, the major advantage of PMI is that high absorbing regions can be resolved directly from the high spatial resolution temperature map. These measurements can then be converted to obtain the optical absorption properties of the tissue using a reconstruction algorithm to model the light propagation and heat transfer in tissue.

This thesis will present the development of the first preclinical in vivo small animal PMI system prototype using the safety standards set by the American National Standard Institution. To optimize the system for in vivo studies, a fast PMI reconstruction method was also developed to accelerate the original PMI reconstruction method ~1000 times faster. These promising results validated the practicality of PMI for preclinical studies and showed the potential of PMI for clinical studies. This led to the development of the first human PMI prototype for clinical breast studies.