Search

This thesis is essentially a combination of my papers [74] and [75]. We study two closely related topics in symplectic topology: Koszul duality and categorical dynamics. Koszul duality is a duality phenomenon between homologically smooth and proper A∞-algebras. When applied in the geometric context, it requires that the wrapped Fukaya category W(M) of a Liouville manifold to be recoverable from its full subcategory F(M) of compact Lagrangian submanifolds, which in particular implies the isomorphism HH*(W(M)) ≅ HH*(F(M)) between the Hochschild cohomologies. Categorical dynamics, on the other hand, studies (infinitesimal) symmetries on the Fukaya category F(M). When such symmetries have geometric origins, they define cohomology classes in the symplectic cohomology SH*(M), i.e. Hochschild cohomology HH*W(M) of the wrapped Fukaya category. In particular, the restriction morphism HH*(W(M)) → HH*(F(M)) is non-vanishing on the classes defining the symmetries on F(M). Thus both of the theories can be regarded as imposing restrictions on the symplectic topology of M so that the Fukaya categories F(M) and W(M) are closely related. To explore the relationships between these two theories, we show that the method of constructing Liouville manifolds whose Fukaya categories F(M) carry dilating C*-actions can also be applied to obtain examples of Liouville manifolds whose Fukaya categories are related by A∞-Koszul duality. On the other hand, we show that for many Liouville manifolds whose Fukaya categories are Koszul dual, there is a generalized theory of categorical dynamics, so that infinitesimal symmetries exist on A∞-subcategory of F(M) consisting of mutually orthogonal spherical objects, and these infinitesimal symmetries can be patched together to give rise to a cyclic homology class in HC-n+1(W(M)). This is the so-called exact Calabi-Yau structure to be mentioned below. As a consequence, this thesis is naturally divided into two parts. In the first part of this thesis (Chapters 3, 4, 5, 6), we consider a family of 6-dimensional Milnor fibers Mp,q,r which are affine hypersurfaces in C4. They are Milnor fibers of stabilizations of cusp and simple elliptic singularities in C3. Explicit computations enable us to identify their compact Fukaya categories F(Mp,q,r) with the cyclic completions of certain directed quiver algebras Ap,q,r, and their wrapped Fukaya categories W(Mp,q,r) with the Calabi-Yau completions of the same quiver algebras Ap,q,r, therefore showing that the Fukaya categories F(Mp,q,r) and W(Mp,q,r) are related to each other via A∞-Koszul duality. As applications, we prove the formality of the Fukaya A∞-algebra of a basis of vanishing cycles in Mp,q,r, and show that the compact Fukaya category F(Mp,q,r) is split-generated by vanishing cycles. In particular, the manifolds Mp,q,r provide interesting examples of Liouville manifolds whose wrapped Fukaya categories are exact Calabi-Yau in the sense of Keller. An exact Calabi-Yau structure on a homologically smooth A∞-category, being the key notion for our study in the second part of this thesis (Chapters 7, 8, 9), is a special kind of smooth Calabi-Yau structures in the sense of Kontsevich-Vlassopoulos [69]. For a Weinstein manifold M, the existence of an exact Calabi-Yau structure on its wrapped Fukaya category W(M) imposes strong restrictions on its symplectic topology. Under the cyclic open-closed map constructed by Ganatra [43], an exact Calabi-Yau structure on W(M) induces a class ̃b in the degree one equivariant symplectic cohomology SH1S1(M). Using ̃b ∈ SH1 S1(M), we construct an endomorphism on the Floer cohomology HF*(L, L) of a Lagrangian sphere L €⊂ M with dimension n ≥ 3, which acts trivially on HF0(L, L), and non-trivially on HFn(L, L). This enables us to prove that for many Weinstein manifolds with exact Calabi-Yau wrapped Fukaya categories, there is an upper bound on the number of disjoint Lagrangian spheres, and the rational homology classes of these spheres are linearly independent. These results generalize those of Seidel in [96] since any Weinstein manifold admitting a quasi-dilation has an exact Calabi-Yau wrapped Fukaya category. Finally, using Koszul duality, we prove that there are examples of Weinstein manifolds whose wrapped Fukaya categories are exact Calabi-Yau, but which do not admit quasi-dilations.

Large bowel cancer, or colorectal cancer (CRC) is the third most common cause of cancer worldwide and the fourth biggest cause of cancer mortality. Twin studies have shown that the heritable contribution is ~35%, with ~5% of cases due to rare, high-penetrance mutations. In the last decade, the use of genome-wide association studies on large, well-characterised case-control cohorts of CRC has facilitated the identification of over 25 common genetic variants that carry with them an increased predisposition to colorectal cancer, invoking the common-disease common variant paradigm. As almost all of these variants lie within non-coding regions, the underlying causal mechanism is to-date poorly understood for the majority of these loci, and it is thought that they mediate risk by influencing gene expression levels. To test this hypothesis, an agnostic approach that utilises expression quantitative trait loci (eQTL) analysis was first carried on 115 normal colorectal mucosa samples and 59 peripheral blood mononuclear cells (PBMC). As these heritable variation on gene expression are likely to be subtle, there is a strong emphasis on the technical methodology to minimise experimentally-induced non-biological variations, including the extraction of high-quality RNA from primary tissue, the selection and validation of reference genes for normalisation of gene expression quantification, as well as internal validation of the samples and data processing. Thereafter, the association between the 25 CRC risk variants and the expression of their cis-genes were examined systematically, demonstrating that ten of these variants are also tissue-specific eQTLs. This intermediate phenotype strongly suggests that they confer risk, at least in part, by modifying regulatory mechanisms. One of the best eQTL associations (Xp22.2) is investigated in further detail to reveal a novel indel polymorphism (Indel24) at the distal promoter region of target gene SHROOM2 that influenced both transcript abundance and CRC risk more than the original tagging SNP. Functional verification with gene reporter assays indicated that Indel24 displays differential allelic control over transcriptional activity. Further in silico analysis and mutations to the reporter gene constructs provided evidence that Indel24 modulates transcription by modifying the spacing between CCAAT motifs and the consequent binding affinity of NF-Y transcription factor. siRNA depletion of NF-Y was associated with a reduction in transcriptional activity of the Indel24 gene construct as well as endogenous SHROOM2, which is strongly supportive of the interaction between Indel24 and NF-Y in the transcriptional activation of SHROOM2. Preliminary evidence is suggestive of SHROOM2 being expressed at the top of the intestinal epithelial crypt and playing a role in cell cycle regulation. Hypothesis-driven approaches can also be of utility in demonstrating functionality of CRC risk variants, complementing the hypothesis-free approach of eQTL analysis. Guided by a recently discovered gene-environment interaction between the 16q22.1 risk variant and circulating vitamin D levels, the influence of the rs9929218 SNP on CDH1 gene expression was examined, in relation to the expression of putative regulatory genes derived from in silico analysis and studies of other target genes. Although there was no direct association between rs9929218 and CDH1 expression, there were multiple two-way interactions that were together suggestive of rs9929218 influencing the VDR/FOXO4 regulation of CDH1. This provides functional support for the mechanism underlying the epidemiological observation of the gene-environment interaction between 16q22.1 and vitamin D, and demonstrates a candidate-based approach in deciphering the link between genetic locus and CRC susceptibility. In summary, the research presented in this thesis has validated the experimental rationale of utilising expression studies of normal colorectal mucosa to hone in on the molecular mechanisms and susceptibility genes underlying the association between common genetic variation and CRC risk.

This work is concerned with the application of the principles and techniques or risk management in the Chinese construction industry. While the benefits of risk management have been increasingly recognized in the Western construction industry, this study suggests that Chinese construction managers have little knowledge or understanding of this important area of management. In recognition of the existence of this failing and the need to overcome it, this study introduces principles, procedures, approaches, models and concepts in risk management. The study concentrates on the consideration of those risks associated with the various elements of construction activities and environment. In the quantitative studies, the main elements considered are the duration and cost of the activities of a project. The main emphasis is placed upon the contribution which can be made to the avoidance of cost overrun. A risk management model has been developed using the Monte Carlo simulation. Although some network-based cost models already exist, this item-based model offers an approach which can predict construction cost rapidly, and enable the consequences of alternative construction methods or management strategies to be evaluated quickly and realistically before tendering and other decisions such as the procurement of materials and equipment are made. While the item-base cost model can be structured to include very detailed cost items, the model in this study assumes a simple structure, and describes construction cost in three general levels - budget cost, primary estimate and planning cost. The computer programs for the model have been developed as a part of the study.

California Assembly Bill 32 (AB32) sets the goal to reduce greenhouse gas (GHG) emissions to a level 80% below 1990 levels by 2050. This deep decarbonization target requires major technology advancement and energy structure transition to a renewable and sustainable future. The environmental aspects of such a transition should be evaluated carefully before major investments in infrastructure are made. Biogas is a promising renewable energy resource in California that shares many similarities with natural gas but with the advantage of being carbon neutral since it generates energy from organic waste. However, biogas contains trace levels of numerous chemical compounds that depend on the feedstock and production process. The air quality implications of using biogas in different situations should be examined carefully before widespread adoption across California. The second chapter of this thesis characterizes the chemical and biological composition of raw biogas produced at five facilities using different feedstocks. The toxicity of combusted biogas is tested under fresh and photo-chemically aged conditions. Results find no strong evidence of potential occupational health risk from the five California biogas sites. Results also show no obvious differences between the toxicity of different biogas combustion exhaust after atmospheric dilution and aging. The third chapter of this thesis examines the emissions from the combustion of upgraded biogas that has CO2 removed and CH4 concentrated to be qualified as renewable natural gas (RNG). A light-duty cargo van was tested with CNG and two RNG blends on a chassis dynamometer to compare the toxicity of the resulting exhaust. CNG vehicle engine exhaust showed a higher or similar level of various toxicity responses, and photochemical reactions did not seem to alter the observed trend. These preliminary results suggest that utilizing biogas for direct heat and electricity generation or as vehicle fuel after upgrading could be useful strategies to reduce carbon intensity without negatively impacting air quality or public health. The fourth chapter of this thesis extends the scope by modeling air pollutant emissions from all California socio-economic sectors under different energy scenarios in the year 2050. To study the air quality implications of some key resources and technologies in the decarbonization transition, a total of six different scenarios were analyzed for various particulate and gaseous pollutant emissions. These scenarios include: 1) a business-as-usual future reference scenario "BAU", 2) a partial GHG reduction scenario that constrains only through 2030 with 40% reduction "CAP30", 3) a climate-friendly 80% GHG reduction scenario featuring deep penetration of advanced technologies and renewable energies "GHGAi", a same 80% GHG reduction scenario with the deployment of biomass carbon capture and sequestration technology "CCS", and two variation scenarios on GHGAi that examine the effect of using more natural gas in built environment "NGB" and power generation "NGT". Results show that major air quality benefits are expected from the GHGAi scenario, which includes aggressive decarbonization of electricity supply, electrification of most end-use appliances, improvement of appliances efficiency, and deployment of low-carbon transportation fuels and technologies. Bio-CCS technology holds promise as a shortcut to GHG mitigation and the utilization of natural gas bridges the transition from traditional to renewable energy systems, but neither of these technologies appear to be optimal from a future air quality management perspective. Adoption of biogas as an energy source plays a small but constructive role in the overall transition of California’s energy system towards a low carbon future.

Abstract: This thesis specializes in statistical issues involving crossover designs, a very popular design in clinical trials for comparing non-curative treatments for their efficacy. The popularity stems from the fact that each experimental subject receives a sequence of trial treatments rather than one single treatment as in parallel designs, and thereby requires fewer experimental subjects. Further, it reduces variability in treatment comparisons because subjects serve as their own controls and between-subject variations are eliminated. One distinct feature in crossover designs is that the treatment assigned to subjects may have lasting effects, called carryover effects, on their responses to treatments in subsequent applications. Crossover designs are well-deliberated for its controversy involving the non-orthogonal key parameters of direct and carryover treatments, which leads to completely different experimental designs depending on which is the primary interest. There are several issues that we address in this thesis. First, when building optimal designs, there are often competing objectives that the investigator desires to optimize. These multiple objectives can include two or more parameters or some functionals, ultimately requiring simultaneous considerations. We revisit the controversy from the point of view of constrained and compound designs for better understanding. Second, we focus on the construction of optimal designs to that of individual- based designs. Typically, designs were constructed to optimize the average subjects and not ideal in clinical and medical applications. N-of-1 trials are randomized multi-crossover experiments using two or more treatments on a single patient. They provide evidence and information on an individual patient, thus optimizing the management of the individual’s chronic illness. We build one sequence N-of-1 universally optimal designs. We also construct optimal N-of-1 designs for two treatments. Then, we discuss the extension to optimal aggregated N-of-1 designs, which will be optimal for an overall treatment effect. Third, we extend the response adaptive allocation strategy for continuous responses to construct those for binary responses with the goal of allocating more patients to better treatment sequences without sacrificing much estimation precision. Although design efficiency in terms of mean squared error may drop sharply, increase in allocated patients to the treatment with beneficial effect is evident. We show a balance can be achieved between various competing multiple objectives. Fourth, we advocate the convex optimization techniques to construct optimal crossover designs where analytic solutions are not feasible. Upon identifying the unique problems and conditions for constructing optimal designs to that of the convex optimization problem, we apply them to find optimal designs relatively simply.

In this thesis, two parts of practical issues in the GPRAM system design are included. The first part is the coding part. The sum-product decoding algorithm of LDPC codes has been refined to fit for the GPRAM hardware implementation. As we all know, communication channel has noise. The noise in telecom system is different from that in GPRAM systems. So the noise should be handled well in the GPRAM design. A noise look-up table was created for FPGA and those noises in the table are quantized. The second part of the thesis is to convert perfect images in video stream to those similar to the coarse images in human vision. GPRAM is an animal like robot in which coarse images are needed more than the fine images in order for us to understand how to GPRAM progresses those images to generate as clear image as we experienced. We use three steps, Point Spread function, inserting Poisson Noise, and introducing Eye fixation movements to mimic the coarse images seen merely from our eyes at the retinal photo-receptor level, i.e., without any brain processing.

Recently, there have been studies on the projection from the entopeduncular nucleus (EP) to the lateral habenula (LHb) and its relationship to the reward and disappointment systems. Specifically, the projection from the EP to the LHb have been found to be excitatory and behaviorally aversive in rats. However, the response to pain in the projection from the EP to the LHb has not yet been studied or identified. Here, we find that many neurons are initially excited, then inhibited in both the EP and LHb, but only the EP has a large proportion of cells that are only inhibited after the shock. We also found that the majority of inhibited EP cells had high baseline firing rates, whereas the majority of LHb neurons with high baseline firing rates were excited by tailshock. These results suggest that there at least two populations of cells in the EP that project to the LHb and are activated by shock. One population sends excitatory projections to the LHb and is excited initially, then inhibited by shock. The other population sends inhibitory projections to the LHb and is only inhibited by shock

The G protein-coupled receptor (GPCR) family is composed of hundreds of members and is expressed in eukaryotes. Each GPCR has seven transmembrane domains and is in charge of sensing changes from the environment, transducing signals, and activating a series of biological responses. The signal transduction pathway of the receptor starts from sensing outside signal and then activates G proteins. This signaling requires a tight control for activation without which impaired cellular function leads to pathology. We have used the pheromone alpha-factor receptor (Ste2p) of the yeast Saccharomyces cerevisiae as a model system to understand ligand binding, receptor activation, and G protein interaction. One method we have used to study ligand binding is to incorporate the photo-reactive crosslinker p-benzoyl-L-phenylalanine (Bpa) into Ste2p to capture alpha-factor. This powerful tool requires the incorporation of Bpa, an unnatural amino acid, into Ste2p by a special genetic manipulation designed in the lab of Peter Schulz (Scripps Institute) and adapted by our lab for Ste2p. Another method to study ligand binding that we have adapted for use in our system is to incorporate a chemical crosslinker [3,4-dihydroxylphenylacetyl (DHPA)] into alpha-factor for periodate-mediated crosslinking to Ste2p. The interacting domain between alpha-factor and transmembrane domain 2 to 3 of Ste2p was identified after DOPAC crosslinking, cyanogen bromide digestion and MALDI-TOF mass spectrometry. After ligand binding, signal transduction is mediated by the interaction of activated Ste2p with its G protein (Gpa1p). We studied this interaction by replacing natural residues in the intracellular loop 3 of Ste2p and C-terminal end of Gpa1p with cysteine and then determining disulfide crosslinking between Ste2p and Gpa1p. Some residues were found to be in close proximity and displayed different interacting patterns due to conformational changes of the receptor upon ligand binding. The information we gathered here allows us to understand more about the physical interactions of alpha-factor, Ste2p, and Gpa1p and provides us insights about the initiation and activation of the signal transduction pathway of a peptide ligand receptor.

This report accounts the experience of a faculty member’s intention of creating an innovative interactive learning simulation in the field of pharmacokinetics to support the faculty member’s teaching and addresses his students’ learning needs. The report also describes the collaboration process between the faculty member and the instructional technology support units through the different phases of design, development, implementation and assessment on the simulation. It also discusses a faculty member’s role in using technology to enhance teaching and learning under university context.

Abstract: Exponential Random Graph Models (ERGMs) have been developed for fitting social network data on both static and dynamic levels. However, the lack of large sample asymptotic properties makes it inadequate in assessing the goodness-of-fit of these ERGMs. Simulation-based goodness-of-fit plots were proposed by Hunter et al (2006), comparing the structured statistics of observed network with those of corresponding simulated networks. In this research, we propose an improved approach to assess the goodness of fit of ERGMs. Our method is shown to improve the existing graphical techniques. We also propose a simulation based test statistic with which the model comparison can be easily achieved.

Service-oriented architectures (SOA), and in particular Web services, have quickly become a popular paradigm to develop distributed applications. Nowadays, more and more organizations shift their core business to the Web services platform within which various interactions between the autonomous services occur. One of the widely accepted standards in the Web services platform is Business Process Execution Language for Web Services (BPEL4WS, or BPEL for short). BPEL defines a language to integrate Web services by creating composite Web services in the form of business processes following the service orchestration paradigm, and it enables organizations to focus on core competence and mission-critical operations while outsource every-thing else to reduce costs and time to market. However BPEL is deficient in privacy issues. The facts are: (1) service requestors’ personal information is fundamental to enable business processes (e.g., the mortgage approval business process); (2) privacy concerns have become one of the most important issues in Information Technology and has received increasing attention from organizations, consumers and legislators; (3) most organizations have recognized that dealing correctly and honestly with customers’ privacy concerns can have beneficial returns for their businesses, not only in terms of being compliant with laws and regulations but also in terms of reputation and potential business opportunities. If not addressed properly, privacy concerns may become an impediment to the widespread adoption of BPEL. Privacy issues have many aspects, the privacy concerns of potential service requestor (i.e., client) and the privacy concerns of service provider (i.e., organization) are two of them. Service requestor specifies his/her privacy concerns as privacy preference, while service provider defines and publishes its privacy policy to specify its privacy promises. Before requestor accesses certain service, he/she likes to know whether the service provider will respect his/her privacy preference. Otherwise, the requestor may seek the desired service from somewhere else. On the other hand, even though most organizations publish their privacy promises, it will be more convincing if customers are assured that such privacy promises are actually kept within the organizations. In this thesis, we propose a privacy enforcement framework for business processes. In particular, we focus on those that are automated using BPEL. The framework consists of two parts. One focuses on the service requestors’ perspective of privacy, the other concentrates on the privacy concerns of the business process owner (i.e., the service provider). More specifically, the first part of the framework is based on description logic, and allows to represent privacy concepts and perform some reasoning about these concepts. The reasoning engine will check requestor's privacy preference against the service provider's published privacy promises before the requestor accesses the desired service. The second part of the framework facilitates the service provider to enforce its privacy policy within all its business processes throughout the life cycle of personal data. The privacy enforcement can be achieved step by step: privacy inspection, privacy verification and privacy obligation management. The first step, privacy inspection, aims to identify which activity needs the involvement of what personal data. The second step, privacy verification, is to verify the correctness of designed BPEL business processes in terms of privacy. The third step is to enforce the privacy by managing the fulfillment of the obligation during the execution of business process. The privacy enforcement framework presented in the thesis has been implemented. The first part of the framework is implemented in the Privacy Match Engine prototype. For the second part of the framework, as different parts of the privacy policy need to be enforced at different stages of the life cycle of business processes, the implementation consists of a privacy verification tool and a privacy obligation management system.