The Woodland-period (ca. 1000 B.C. to A.D. 1050) fisher-hunter-gatherers of the Crystal River drainage on Florida’s Big Bend Coast are well known among southeastern archaeologists for their elaborate shell mound architecture, maritime lifeway, and exotic exchange goods. Recent archaeological investigations at the Crystal River site have employed high-resolution topographic mapping, geophysical surveys, trench excavations, and coring to model the temporality of mound construction and occupation at the site; this work has set the stage for subsequent research focusing on community structure, resource extraction, and human-ecosystem dynamics. However, like many central and north peninsular Gulf Coast sites, our understanding of Crystal River lacks robust environmental context. Various geologists and archaeologists have proposed that several major sea-level oscillations transpired during the late Holocene at scales that would have heavily impacted coastal habitation; however, the paleoenvironmental reconstructions available for the Gulf Coast disagree dramatically about the timing, extent, and general pattern of sea-level history, which discourages the use of non-localized (or averaged) reconstructions for archaeological interpretation. This study investigates palynological remains recovered from midden and wetland core samples taken at Crystal River site to provide a record of vegetation response to both externally driven environmental changes (i.e. sea-level and climate flux) and anthropogenic landscape alterations throughout the first millennium A.D. Results show that stratified midden deposits and wetland soils at Crystal River site contain intact fossil pollen assemblages that span the site’s occupational history and extend into pre- and post-abandonment periods. The pollen data suggest that local communities experienced substantial environmental changes that reorganized the composition and geographic distribution of coastal ecosystems. These transformations align well with broadly recognized climatic episodes (i.e. the Roman Warm Period, Vandal Minimum, and Medieval Warm Period). In contrast, the Crystal River drainage experienced a unique pattern of sea-level flux that does not approximate oft-cited sea-level records developed elsewhere on Florida’s Gulf Coast, or the averaged sea-level curves for the Gulf Coast. Additionally, the microbotanical remains preserved within midden soils at Crystal River site may speak to the roles of particular plants as subsistence, technological, medicinal, and/or ceremonial resources.

Master thesis describes the design of the project of a new service creation for the com-pany, which is engaged in offering services through active telemarketing. The method-ology of project management is used in the thesis.

This thesis focuses on mapping the service department at Bystronic Czech Republic s.r.o. and optimizing it to improve future process conditions. Optimization will include a solution design to improve the functionality of the department so that business goals can be achieved. The evaluation is based on a comparison of data of past periods using a process approach. The introduction describes the analytical and optimization tools from which the data will be used to design the most appropriate solution.

The theory of multidimensional persistence was introduced in a paper by G. Carlsson and A. Zomorodian as an extension to persistent homology. The central object in multidimensional persistence is the persistence module, which represents the homology of a multi filtered space. In this thesis, a novel algorithm for computing the persistence module is described in the case where the homology is computed with coefficients in a field. An algorithm for computing the feature counting invariant, introduced by Chachólski et al., is investigated. It is shown that its computation is in general NP-hard, but some special cases for which it can be computed efficiently are presented. In addition, a generalization of the barcode for persistent homology is defined and conditions for when it can be constructed uniquely are studied. Finally, a new topology is investigated, defined for fields of characteristic zero which, via the feature counting invariant, leads to a unique denoising of a tame and compact functor. / Teorin om multidimensionell persistens introduserades i en artikel av G. Carlsson och A. Zomorodian som en generalisering av persistent homologi. Det centrala objektet i multidimensionell persistens är persistensmodulen, som representerar homologin av ett multifilterat rum. I denna uppsats beskrivs en ny algoritm för beräkning av persistensmodulen i fallet där homologin beräknas med koefficienter i en kropp. En algoritm för beräkning av karaktäristik-räknings-invarianten, som introducerade av Chachólski et al., utforskas och det visar sig att dess beräkning i allmänhet är NP-svår. Några specialfall för vilka den kan beräknas effektivt presenteras. Vidare definieras en generalisering av stäckkoden för persistent homologi och kraven för när den kan konstrueras unikt studeras. Slutligen undersöks en ny topologi, definierad för kroppar av karaktäristik noll, som via karaktäristik-räknings-invarianten leder till en unik avbränning.

One of the major challenges in Big Data is the accurate and meaningful assessment of system performance. Unlike other systems, minor differences in efficiency can escalate to large differences in costs and power consumption. While there are several tools on the marketplace for measuring the performance of Big Data systems, few of them have been explored in-depth. This report investigated the interfacing, data generation and workload implementations of two Big Data benchmarking suites, BigDataBench and Hibench. The purpose of the study was to establish the capabilities of each tool with regards to interfacing, data generation and workload implementation. An exploratory and qualitative approach was used to gather information and analyze each benchmarking tool. Source code, documentation, and reports published by the developers were used as information sources. The results showed that BigDataBench and HiBench were designed similarly with regards to interfacing and data flow during the execution of a workload with the exception of streaming workloads. BigDataBench provided for more realistic data generation while the data generation for HiBench was easier to control. With regards to workload design, the workloads in BigDataBench were designed to be applicable to multiple frameworks while the workloads in HiBench were focused on the Hadoop family. In conclusion, neither of benchmarking suites was superior to the other. They were both designed for different purposes and should be applied on a case-by-case basis. / En av de stora utmaningarna i Big Data är den exakta och meningsfulla bedömningen av systemprestanda. Till skillnad från andra system kan mindre skillnader i effektivitet eskalera till stora skillnader i kostnader och strömförbrukning. Medan det finns flera verktyg på marknaden för att mäta prestanda för Big Data-system, har få av dem undersökts djupgående. I denna rapport undersöktes gränssnittet, datagenereringen och arbetsbelastningen av två Big Data benchmarking-sviter, BigDataBench och HiBench. Syftet med studien var att fastställa varje verktygs kapacitet med hänsyn till de givna kriterierna. Ett utforskande och kvalitativt tillvägagångssätt användes för att samla information och analysera varje benchmarking verktyg. Källkod, dokumentation och rapporter som hade skrivits och publicerats av utvecklarna användes som informationskällor. Resultaten visade att BigDataBench och HiBench utformades på samma sätt med avseende på gränssnitt och dataflöde under utförandet av en arbetsbelastning med undantag för strömmande arbetsbelastningar. BigDataBench tillhandahöll mer realistisk datagenerering medan datagenerering för HiBench var lättare att styra. När det gäller arbetsbelastningsdesign var arbetsbelastningen i BigDataBench utformad för att kunna tillämpas på flera ramar, medan arbetsbelastningen i HiBench var inriktad på Hadoop-familjen. Sammanfattningsvis var ingen av benchmarkingssuperna överlägsen den andra. De var båda utformade för olika ändamål och bör tillämpas från fall till fall.

Due to the unpredictability of the stock market,forecasting stock prices is a challenging task. In this project,we will investigate the performance of the machine learningalgorithm LSTM for stock market prediction. The algorithmwill be based only on historical numerical data and technicalindicators for IBM and FORD. Furthermore, the denoising anddimension reduction algorithm, PCA, is applied to the stockdata, to examine if the performance of forecasting the stockprice is greater than the initial model. A second method, transferlearning, is applied by training the model on the IBM datasetand then applying it on the FORD dataset, and vice versa, toevaluate if the results will improve. The results show that whenthe PCA algorithm is applied to the dataset separately, and incombination with transfer learning, the performance is greater incomparison to the initial model. Moreover, the transfer learningmodel is inconsistent as the performance is worse for FORD inrespect to the initial model, but better for IBM. Thus, concerningthe results when forecasting stock prices using related tools, it issuggested to use trial and error to identify which of the modelsthat performs the optimally. / Att förutse aktiekurser är en utmanande uppgift. Detta beror på aktiemarknadens oförutsägbarhet. Därför kommer vi i detta projekt att undersöka prestandan för maskininlärnings algoritmen LSTMs prognosförmåga för aktie priser. Algoritmen baseras endast på historisk numerisk data och tekniska indikatorer for företagen IBM och FORD. Vidare tillämpas brus minskande och dimension reducerande algorithmen, PCA, på aktiedata för att undersöka om prestandan för att förutse aktie priser är bättre än den ursprungliga modellen. En andra metod, transfer learning, tillämpas genom att träna modellen på IBM data och sedan använda den på FORD data, och vice versa, för att utvärdera om resultaten kommer att förbättras. Resultaten visar, när PCA-algoritmen tillämpas på aktiedata separat, och i kombination med transfer learning är prestandan bättre jämfört med bas modellen. Vidare kan vi inte dra slutsatser om transfer learning då prestandan är sämre för FORD med avseende på bas modellen, men bättre för IBM. I hänsyn till resultaten så föreslås det att man tillämpar modellerna för att identifiera vilken som är mest optimal när man arbetar i ett relaterat ämnesområde. / Kandidatexjobb i elektroteknik 2020, KTH, Stockholm