Many large scale simulations now produce datasets that are signi cantly larger than can typically be stored in memory on a visualization system. Visualization algorithms then become ine ective and stall since the data must be paged to disk. Recently, in-situ visualization has received renewed attention for visualizing large datasets that are distributed among many processors during a simulation. It takes advantage of the fact that the full dataset is already in main memory, distributed among multiple processors. Visualization in this environment then requires communication which can be more expensive than disk access. The goal of this thesis was to develop an in-situ visualization technique using ray tracing that employs ray caches to reduce communication overhead. Ray caches attempt to replace a communication operation with a less expensive cache search operation. A prototype implemented on Sharcnet shows ray caching can signi cantly improve overall performance at a small cost to image quality. / UOIT

The difficulty of clustering and the variety of clustering methods suggest the need for a theoretical study of clustering. Using the idea of a standard statistical framework, we propose a new framework for clustering. For a well-defined clustering goal we assume that the data to be clustered come from an underlying distribution and we aim to find a high-density cluster tree. We regard this tree as a parameter of interest for the underlying distribution. However, it is not obvious how to determine a connected subset in a discrete distribution whose support is located in a Euclidean space. Building a cluster tree for such a distribution is an open problem and presents interesting conceptual and computational challenges. We solve this problem using graph-based approaches and further parameterize clustering using the high-density cluster tree and its extension. Motivated by the connection between clustering outcomes and graphs, we propose a graph family framework. This framework plays an important role in our clustering framework. A direct application of the graph family framework is a new cluster-tree distance measure. This distance measure can be written as an inner product or kernel. It makes our clustering framework able to perform statistical assessment of clustering via simulation. Other applications such as a method for integrating partitions into a cluster tree and methods for cluster tree averaging and bagging are also derived from the graph family framework.

The search for clusters in empirical data is an important and often encountered research problem. Numerous algorithms exist that are able to render groups of objects or individuals. Of course each algorithm has its strengths and weaknesses. In order to identify these crucial points artificial data was generated - based primarily on experience with structures of empirical data - and used as benchmark for evaluating the results of numerous cluster algorithms. This work is an addition to SFB Working Paper No. 7 where hard competitive learning (HCL), neural gas (NGAS), k-means and self organizing maps (SOMs) were compared. Since the artificial data scenarios and the evaluation criteria used remained the same, they are not explained in this work, where the results of five additional algorithms are evaluated. (author's abstract) / Series: Working Papers SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

Social scientists confronted with the problem of segmenting individuals into plausible subgroups usually encounter two main problems: First: there is very little indication about the correct choice of the number of clusters to search for. Second: different cluster algorithms and even multiple replications of the same algorithm result in different solutions due to random initializations and stochastic learning methods. In the worst case numerous solutions are found which all seem plausible as far as interpretation is concerned. The consequence is, that in the end clusters are postulated that are in fact "chosen" by the researcher, as he or she makes decisions on the number of clusters and the solution chosen as the "final" one. In this paper we concentrate on the power and stability of several popular clustering algorithms under the condition that the correct number of clusters is known. Artificial data sets modeled to mimic typical situations from tourism marketing are constructed. The structure of these data sets is described in several scenarios, and artificial binary data are generated accordingly. These data, ranging from very simple to more complex, real-data-like structures, enable us to systematically analyze the "behavior" of the cluster methods. Section 3 gives an overview of all cluster methods under investigation. Section 4 describes our experimental results, comparing first all scenarios and then all cluster methods. To accomplish this task, several evaluation criteria for cluster methods are proposed. Finally: Sections 5 and 6 draw some conclusions and give an outlook on future research. (author's abstract) / Series: Working Papers SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

The difficulty of clustering and the variety of clustering methods suggest the need for a theoretical study of clustering. Using the idea of a standard statistical framework, we propose a new framework for clustering. For a well-defined clustering goal we assume that the data to be clustered come from an underlying distribution and we aim to find a high-density cluster tree. We regard this tree as a parameter of interest for the underlying distribution. However, it is not obvious how to determine a connected subset in a discrete distribution whose support is located in a Euclidean space. Building a cluster tree for such a distribution is an open problem and presents interesting conceptual and computational challenges. We solve this problem using graph-based approaches and further parameterize clustering using the high-density cluster tree and its extension. Motivated by the connection between clustering outcomes and graphs, we propose a graph family framework. This framework plays an important role in our clustering framework. A direct application of the graph family framework is a new cluster-tree distance measure. This distance measure can be written as an inner product or kernel. It makes our clustering framework able to perform statistical assessment of clustering via simulation. Other applications such as a method for integrating partitions into a cluster tree and methods for cluster tree averaging and bagging are also derived from the graph family framework.

The organometallic chemistry of C60 has attracted much attention concerning the effect of metal coordination on the properties of C60 since the discovery and macroscopic synthesis of C60. In our study, we try to synthesize two analogous ligands which contain two phosphines. And reaction of the new ligands and metal carbonyl clusters will produce new-type of metal complexes. Addition of Ph2(Li)PBH3, prepared by n-BuLi-deprotonation of Ph2(H)PBH3 in THF, to toluene solution of C60 took place to give the adduct C60H2(Ph2PBH3)2 (2) after quenching with HCl in ethyl acetate. Reaction of C60 and sodium 1-propanethiolate in acetonitrile produces C602-, and then adding PPh2Cl to C602- solution to afford C60(PPh2)2 (8). Treatment of the borane complex 2 with diazabicyclo[2.2.2]octane (DABCO) in toluene removes the borane group to give the phosphine C60H2(PPh2)2 (3). Reaction of 3 and Os3(CO)10(NCMe)2 at room temperature produces Os3(CO)10(£g,£b3-(PPh2)C60H) (5) and C60H2(PPh2)2(Os3(CO)10)2 (6). Furthermore, reaction of 3 and Ru3(CO)12 at 85 ¢J produces Ru3(CO)10(£g,£b3-(PPh2)C60H) (7). The resulting compounds are characterized by NMR, IR, Mass, X-ray and EA.

As the rises of the market and industry cluster of China, the importance of the Taiwan¡¦s cluster effect to multinational corporations¡]MNCs¡^will have the changes. And the changes of cluster effect between Taiwan and China have had the impact on MNC Taiwan subsidiary¡¦s role. This thesis focuses of the cluster effect between Taiwan and China that changes the MNC¡¦s Taiwan subsidiary¡¦s role. With regard to the factors affecting cluster of industries, previous studies have based mainly on Porter¡¦s Diamond Model, which includes factor conditions, demand conditions, related and supporting industries and firm strategy, structure and rivalry. Besides, according to the series research by Birkinshaw, the HQ assignment and the subsidiary¡¦s behavior will also have the impact on subsidiary¡¦s role. This thesis will study the changes between these three dimensions and MNC¡¦s Taiwan subsidiary¡¦s role. This thesis uses the case study in order to support the formation of research structure and hypothesis, which also survey to collect data. The population is the list of foreign enterprises in Taiwan published by Dun and Bradstreet, 2000. The subsidiaries belong to manufacturing industry and service industry. The subsidiaries are owned by foreign MNC which must operate over one year in Taiwan. We use multi-regression to examine the relationships between the dependent variables and the changes of MNC Taiwan subsidiary¡¦s role. After analyzing 60 MNC¡¦s Taiwan subsidiaries, the results reveal that HQ assignment, the subsidiary¡¦s initiative behavior and the changes of cluster effect between Taiwan and China will have the influence on MNC¡¦s Taiwan subsidiaries¡¦ roles. Therefore, these dissertation results respond to Birkinshaw¡¦s perspective. The subsidiary behavior is simultaneously influenced by HQ, subsidiary and local environment.

House Price in Taiwan is very volatile during the past few decades. As Taiwan go into enormous boom, more and more amount of money invest in the house market. Although house investment is considered as a good investment tool with low risk and inflation hedge properties, its risk can not be underestimated. Therefore, by using the GARCH model, this paper tries to analyze volatilities of investment return in the Taipei housing market from 1973 to 2002. For existing housing, we are not able to use GARCH to model investment volatility because of uncorrelated term risks. On the contrary, pre-sale housing contains correlated term risk. We adopt ARMA(4,4)-GARCH(1,1) to model the investment volatility of pre-sale housing. The investment risk of pre-sale housing is not constant but is time-varying. When an unexpected event happened, the shock will persist but decay from 86 percent in the next term to 40 percent in the sixth term. And we can observe volatility cluster phenomenon from the graph of conditional variance. During 1973 to 1975¡B1979 to 1983 and 1987 to 1990, the risks are higher than other period. Because previous studies commonly suggest some structural changes in the Taiwan housing market, we also control the risk premium affected by the structural changes in our model. We found ARMA(4,4)-GARCH(1,1) can still model the investment volatility process of pre-sale housing, but there is no evidence of risk premium caused by structural changes.

In order to create win-win solutions, outstanding multinational companies usually look for a well-performed cluster to join. On the one hand, the local cluster can upgrade their competitiveness by leveraging the resources of the MNCs, such as technologies, specialized materials, human resources, and also learning from their management know-how. On the other hand, MNCs can enjoy benefits, such as lower cost, more effective logistics, and approaching the local market. Both the development of the Export Processing Zone and the success of high-tech industry of Taiwan are good examples. The TFT-LCD industry is the newly arisen one after semi-conductor industry in Taiwan, and the government has positively pushed TFT-LCD industry to be the core competitive-advantage industry, and expect to create 1 trillion NT dollars in 2006 to become worldwide FDP research and development country. Though the local companies work hard for years, they build up the whole supply-chain as a competitive cluster. This research focuses on the MNCs effects on the development of the TFT-LCD cluster, and uses multi-case methodology to analyze the MNC suppliers Corning, Asahi and the local player PVU in the TFT glass industry. The conclusion of the research shows that the strategy evolution of the MNC TFT glass suppliers played the vital role of the upgrading of whole industry becoming an embedded cluster.

Abstract We propose to research the interaction between users and web servers in Internet , and we call that ¡§Session - Based model¡§ . Then we add some policies to the session-based model , we define that is ¡§Session - Based Management¡§ . For the explosive growth of Internet service¡]eg.,e-commerce¡K¡^, we consider about what users want and tracing users¡¦ behaviors , those are what we want to research and analysis . We using ¡§cookie¡§ , what is the technique to use on the Client-Server model of Internet . This make server directly and easily know some information about users . So server supply quality of service to user what they have identified . Then we trace technologies of cookie which are used in some popular web sites¡]eg., eBay , ubid , kimo , openfind¡^, and analyze the impact of users about those technologies . For example , we classify the Internet service to ¡V shopping cart¡Bsearch engine¡K Finally , we construct our e-commerce web site to parsing every packets through our site , getting the information what we want from those packets , and then we define our some policies into our Session-Based model . The infrastructure of our implementation environment is in Server Cluster Architecture , which is the most popular one this time . More and more ISPs¡]Internet Service Provider¡^or ICPs¡]Internet Content Provider¡^construct their web sites in this kind of structure . In our cluster system , our distributor will analyze all kinds of packets from all heterogeneous servers . Using our technique will make distributor to learn how to know the user information and identify the users , so distributor will know how to supply users¡¦ session-based priority¡Bdifferentiated service and so on . We actually implement our mechanisms in our server cluster system .