This dissertation examines econometric properties of foreign exchange markets under conditions which approximate actual market conditions more closely than studies done in existing literature. This study uses continuous data for spot exchange rates for the months of April, May, and June in 1989 and daily data from 1973 to 1990. The continuous data was obtained from Reuters Ltd from a direct data feed of the Reuters FXFX page. The currencies chosen are the German Mark, the Pound Sterling, and the Japanese Yen, all quoted bilaterally with respect to the U.S. dollar The first chapter of this dissertation discusses the efficient market hypothesis. The second chapter carries out a detailed analysis of the behavior of spot exchange rates by testing for the presence of three, two and one unit roots using various recently developed methodology. It was found that for high frequencies of 30 seconds or one minute the assumption of an unit root was not valid on quite a few occasions. After this, the study uses temporal aggregation to decrease the frequency of the data to see if there is any change in the results with lower frequencies. At lower frequencies, like 5 minutes, 15 minutes, 30 minutes, or one hour, the hypothesis of the presence of unit roots is accepted. Diagnostic tests indicate the presence of heteroskedasticity and high kurtosis in the data at high frequencies. Therefore, the problem of time-varying variance and selection of an appropriate distribution to reduce the kurtosis was dealt with by fitting ARCH and GARCH models to the data Another test of market efficiency is to test whether forward rates are unbiased and efficient estimators of future spot rates. This is carried out using daily, weekly, and monthly data for the Deutschemark, French Franc, and the British Pound against the U.S. Dollar for the period 1977 to 1990. We find that spot and forward rates are cointegrated for daily and weekly data but not for monthly data. Additional coefficient restriction necessary for the unbiasedness hypothesis are not satisfied for any currency; and, therefore, we conclude that the unbiasedness hypothesis does not hold / acase@tulane.edu

Prototyping has been proposed as an alternative to the traditional waterfall approach for designing software systems. The corresponding methodology consists of an ad-hoc mixture of language, tools, and methods. Although there are a few systems available for prototyping concurrent programs, they lack some important features. One problem is that some of these systems are limited to paper designs; there is no implementation available. A larger problem is that these systems lack a formal basis for their language and translation method definitions A programming system, Ripple, is proposed for prototyping parallel and distributed programs. The proposed environment to support the prototyping methodology consists of a hierarchy of three languages (TPL, SPL, IPL), a set of tools, and a hierarchical method. TPL is a process language designed to express the structure of the prototype without describing the details of functions. SPL is a specification language designed to describe the functions in an abstract specification. IPL is a programming language that can be implemented on a given system. It will be shown how TPL can be formally converted into SPL and in turn be translated into IPL The Ripple languages have been designed with the two important features found lacking in previous systems. The formality of the Ripple languages and the translation from one language to the other implies that the meanings of TPL programs are preserved throughout the hierarchy. In addition, this formal basis provides a framework for various static analysis tasks. Finally, a compiler was built to show the feasibility of translating a kernel of TPL to both SPL and IPL / acase@tulane.edu

acase@tulane.edu

In this body of work, studies of folding kinetics and thermodynamic stability of Pseudomonas aeruginosa azurin, a copper-binding beta-barrel protein, are presented. To uncover the role of the copper cofactor in determining the folding pathway of azurin, interactions of the unfolded protein with copper were investigated using a mutagenic approach and a model-peptide system. Based on the findings, the coordination sphere of copper in the unfolded protein is proposed to involve at least two native-state ligands: histidine-117 and cysteine-112. A small peptide corresponding to the carboxy-terminal part of full-length azurin is found to adopt beta-sheet-like structure upon copper binding, Formation of this beta-hairpin induced by the copper ion in the unfolded protein was originally suggested as a mechanism for folding nucleation. However, detailed folding-kinetic studies demonstrate that the folding speed of azurin is not increased in the presence of copper. It is concluded that formation of the beta-hairpin induced by copper binding is not rate limiting for folding The topology of the native state is the primary determinant of the folding rate for azurin, as demonstrated by the study of two active-site azurin mutants. His46Gly and His117Gly azurins, in their apo (without cofactor) forms. The decreases in the thermodynamic stability of the mutants, as compared to wild-type azurin, originate from the acceleration of unfolding, However, the folding rates for the mutants are almost identical to what is observed for wild-type apo-azurin / acase@tulane.edu

acase@tulane.edu

acase@tulane.edu

acase@tulane.edu

acase@tulane.edu

acase@tulane.edu

acase@tulane.edu