In the archaeological literature, mountains are often portrayed as the boundaries between inhabited spaces. Yet occupying high elevations may have been an adaptive choice for ancient peoples, as rapidly changing elevations also offer variation in climate and resources over a relatively small area. So what happens, instead, if we put mountain landscapes at the center of our analyses of prehistoric seasonal rounds and ecological adaptation? This Ph.D. argues that, in order to understand any landscape that includes mountains, from the Alps to the Andes, one must include the ecology and archaeology of the highest elevations. Specifically, I base my findings on new fieldwork and lithic collections from the Absaroka and Beartooth Mountains in the Greater Yellowstone Ecosystem (GYE) of the Rocky Mountains, which was a vital crossroads of prehistoric cultures for more than 11,000 years. I include five interlocking analyses. First, I consider the impacts of anthropogenic climate change on high elevation cultural resources, focusing on the diminishing resiliency of ancient high elevation ice patches and the loss of the organic artifacts and paleobiological materials they contain. Second, I create a dichotomous key for chronologically typing projectile points, suggesting a methodological improvement for typological dating in the GYE and for surface archaeology more broadly. Third, I use obsidian source data to consider whether mountain people were a single, unified group or were represented by a variety of peoples with different zones of land tenure. Fourth, I consider high elevation occupation in both mountain ranges as part of the seasonal round, using indices of diversity in tool types and raw material to study how the duration of those occupations changed through time. And, finally, I test the common contention that ancient people primarily used mountains as refugia from extreme climatic pressure at lower elevations. Ultimately, I find that, in both mountain ranges, increased high elevation activity is most highly correlated with increased population, not with hot, dry climatic conditions. In other words, the mountains were more than simply refugia for plains or basin people to occupy when pressured by climatic hardship. In addition, between the Absarokas and the Beartooths the evidence suggests two different patterns of occupation, not a monolithic pan-mountain adaptation. These results demonstrate the potential contributions of surface archaeology to our understanding of prehistory, and have important implications for the way we think about mountain landscapes as peopled spaces in relation to adjacent lower-elevation areas.

Problems involving optimal sequential making in uncertain dynamic systems arise in domains such as engineering, science and economics. Such problems can often be cast in the framework of Markov Decision Process (MDP). Solving an MDP requires computing the optimal value function and the optimal policy. The idea of dynamic programming (DP) and the Bellman equation (BE) are at the heart of solution methods. The three important exact DP methods are value iteration, policy iteration and linear programming. The exact DP methods compute the optimal value function and the optimal policy. However, the exact DP methods are inadequate in practice because the state space is often large and in practice, one might have to resort to approximate methods that compute sub-optimal policies. Further, in certain cases, the system observations are known only in the form of noisy samples and we need to design algorithms that learn from these samples. In this thesis we study interesting theoretical questions pertaining to approximate and learning algorithms, and also present an interesting application of MDPs in the domain of crowd sourcing. Approximate Dynamic Programming (ADP) methods handle the issue of large state space by computing an approximate value function and/or a sub-optimal policy. In this thesis, we are concerned with conditions that result in provably good policies. Motivated by the limitations of the PBE in the conventional linear algebra, we study the PBE in the (min, +) linear algebra. It is a well known fact that deterministic optimal control problems with cost/reward criterion are (min, +)/(max, +) linear and ADP methods have been developed for such systems in literature. However, it is straightforward to show that infinite horizon discounted reward/cost MDPs are neither (min, +) nor (max, +) linear. We develop novel ADP schemes namely the Approximate Q Iteration (AQI) and Variational Approximate Q Iteration (VAQI), where the approximate solution is a (min, +) linear combination of a set of basis functions whose span constitutes a subsemimodule. We show that the new ADP methods are convergent and we present a bound on the performance of the sub-optimal policy. The Approximate Linear Program (ALP) makes use of linear function approximation (LFA) and offers theoretical performance guarantees. Nevertheless, the ALP is difficult to solve due to the presence of a large number of constraints and in practice, a reduced linear program (RLP) is solved instead. The RLP has a tractable number of constraints sampled from the original constraints of the ALP. Though the RLP is known to perform well in experiments, theoretical guarantees are available only for a specific RLP obtained under idealized assumptions. In this thesis, we generalize the RLP to define a generalized reduced linear program (GRLP) which has a tractable number of constraints that are obtained as positive linear combinations of the original constraints of the ALP. The main contribution here is the novel theoretical framework developed to obtain error bounds for any given GRLP. Reinforcement Learning (RL) algorithms can be viewed as sample trajectory based solution methods for solving MDPs. Typically, RL algorithms that make use of stochastic approximation (SA) are iterative schemes taking small steps towards the desired value at each iteration. Actor-Critic algorithms form an important sub-class of RL algorithms, wherein, the critic is responsible for policy evaluation and the actor is responsible for policy improvement. The actor and critic iterations have deferent step-size schedules, in particular, the step-sizes used by the actor updates have to be generally much smaller than those used by the critic updates. Such SA schemes that use deferent step-size schedules for deferent sets of iterates are known as multitimescale stochastic approximation schemes. One of the most important conditions required to ensure the convergence of the iterates of a multi-timescale SA scheme is that the iterates need to be stable, i.e., they should be uniformly bounded almost surely. However, the conditions that imply the stability of the iterates in a multi-timescale SA scheme have not been well established. In this thesis, we provide veritable conditions that imply stability of two timescale stochastic approximation schemes. As an example, we also demonstrate that the stability of a widely used actor-critic RL algorithm follows from our analysis. Crowd sourcing (crowd) is a new mode of organizing work in multiple groups of smaller chunks of tasks and outsourcing them to a distributed and large group of people in the form of an open call. Recently, crowd sourcing has become a major pool for human intelligence tasks (HITs) such as image labeling, form digitization, natural language processing, machine translation evaluation and user surveys. Large organizations/requesters are increasingly interested in crowd sourcing the HITs generated out of their internal requirements. Task starvation leads to huge variation in the completion times of the tasks posted on to the crowd. This is an issue for frequent requesters desiring predictability in the completion times of tasks specified in terms of percentage of tasks completed within a stipulated amount of time. An important task attribute that affects the completion time of a task is its price. However, a pricing policy that does not take the dynamics of the crowd into account might fail to achieve the desired predictability in completion times. Here, we make use of the MDP framework to compute a pricing policy that achieves predictable completion times in simulations as well as real world experiments.

Die Entwicklung von Technologien sowie die Veränderung von rechtlichen Rahmenbedingungen haben die Handlungsmöglichkeiten von Industrieunternehmen im Zusammenhang mit der Energiebereitstellung seit der Jahrtausendwende deutlich erhöht: Neben dem klassischen Einkauf, d. h. dem Fremdbezug elektrischer Energie, können Industrieunternehmen eigene Kraftwerkskapazitäten aufbauen, um elektrische Energie selbst zu erzeugen. Ferner schafft die Liberalisierung des Energiemarktes die Voraussetzungen dafür, dass diese Unternehmen ihre selbst erzeugte elektrische Energie auch verkaufen bzw. ins öffentliche Netz einspeisen können. Zukünftig ist außerdem zu erwarten, dass Speichertechnologien eine immer größere Rolle spielen, um die Versorgungssicherheit unabhängig vom öffentlichen Stromnetz zumindest kurzzeitig aufrecht erhalten oder Lastspitzen abfangen zu können. Mit den skizzierten Entwicklungen geht eine zunehmende Komplexität der von Industrieunternehmen im Rahmen der Energiebereitstellung zu treffenden Entscheidungen einher. Die vorliegende Arbeit strukturiert diese Entscheidungen, arbeitet Handlungsalternativen und daraus zu bildende Handlungsbündel systematisch heraus und entwickelt schließlich ein Bewertungskonzept, mit dem auf Basis eines sukzessiven Vorgehens eine unter monetären Gesichtspunkten vorteilhafte Bereitstellungsalternative identifiziert werden kann. Das Bewertungskonzept stützt sich dabei auf Methoden der Investitionsrechnung und erfasst differenziert Produktions- sowie Transaktionskosten der jeweiligen Bereitstellungsalternativen.

Companies are under constant pressure to adapt and improve their processes to staycompetitive. Since most of their processes are handled by software, it also needs toconstantly change. Those improvements and changes add up over time and increase thecomplexity of the system, which in turn prevents the company from further adaption.In order to change and improve existing business processes and their implementation withinsoftware, several stakeholders have to go through a long process. Current IT methodologies arenot suitable for such a dynamic environment. The analysis of this change process shows thatfour software characteristics are important to speed it up. They are: transparency, adaptability,testability and reparability. Transparency refers to the users capability to understand what thesystem is doing, where and why. Adaptability is a mainly technical characteristic that indicatesthe capability of the system to evolve or change. Testability allows automated testing andvalidation for correctness without requiring manual checks. The last characteristic is reparability,which describes the possibility to bring the system back into a consistent and correct state, evenif erroneous software was deployed.An architecture and software development patterns are evaluated to build an overall programmingmodel that provides the software characteristics. The overall architecture is basedon microservices, which facilitates decoupling and maintainability for the software as well asorganizations. Command Query Responsibility Segregation decouples read from write operationsand makes data changes explicit. With Event Sourcing, the system stores not only the currentstate, but all historic events. It provides a built-in audit trail and is able to reproduce differentscenarios for troubleshooting and testing.A demo process is defined and implemented within multiple prototypes. The design of theprototype is based on the programming model. It is built in Javascript and implements Microservices,CQRS and Event Sourcing. The prototypes show and validate how the programmingmodel provides the software characteristics. Software built with the programming model allowscompanies to iterate faster at scale. Since the programming model is suited for complex processes,the main limitation is that the validation is based on a demo process that is simpler and thebenefits are hard to quantify. / ör att fortsatt vara konkurrenskraftiga är företag under konstant press att anpassa ochförbättra sina processer. Eftersom de flesta processer hanteras av programvara, behöveräven de ständigt förändras. Övertiden leder dessa förbättringar och förändringar till ökadsystemkomplexitet, vilket i sin tur hindrar företaget från ytterligare anpassningar. För attförändra och förbättra befintliga affärsprocesser och dess programvara, måste idag typiskt fleraaktörer vara en del av en lång och tidskrävande process. Nuvarande metoder är inte lämpade fören sådan dynamisk miljö. Detta arbete har fokuserat på fyra programvaruegenskaper som ärviktiga för att underlätta förändringsprocesser. Dessa fyra egenskaper är: öppenhet, anpassningsförmåga,testbarhet och reparerbarhet. Öppenhet, hänvisar till förmågan att förstå varför, var ochvad systemet gör. Anpassningsbarhet är huvudsakligen en teknisk egenskap som fokuserar påsystemets förmåga att utvecklas och förändras. Testbarhet strävar efter automatisk testning ochvalidering av korrekthet som kräver ingen eller lite manuell kontroll. Den sista egenskapen ärreparerbarhet, som beskriver möjligheten att återhämta systemet till ett konsekvent och korrekttillstånd, även om felaktig programvara har använts. En programmeringsmodell som rustarprogramvara med de ovan beskrivna programegenskaperna är utvecklad i detta examensarbete.Programmeringsmodellens arkitektur är baserad på diverse micro-tjänster, vilka ger brafrånkopplings- och underhållsförmåga för en programvara, samt användarorganisationerna.Command Query Responsibility Segregation (CQRS) frånkopplar läsoperationer från skrivoperationeroch gör ändringar i data explicita. Med Event Sourcing lagrar systemet inte endastdet nuvarande tillståndet, utan alla historiska händelser. Modellen förser användarna medett inbyggt revisionsspår och kan reproducera olika scenarion för felsökning och testning. Endemoprocess är definierad och implementerad i tre olika prototyper. Designen av prototypernaär baserad på den föreslagna programmeringsmodellen. Vilken är byggd i Javascript och implementerarmicro-tjänster, CQRS och Event Sourcing. Prototyperna visar och validerar hurprogrammeringsmodellen ger programvaran rätt egenskaper. Programvara byggd med dennaprogrammeringsmodell tillåter företag att iterera snabbare. De huvudsakliga begränsningarna iarbetet är att valideringen är baserad på en enklare demoprocess och att dess fördelar är svåraatt kvantifiera.

The perspective of procurement and supply chain management is changing dramatically; traditionally, it was seen as a support function; however, the procurement function is receiving increased attention and investment as an essential contributor to the strategic success and a business enabler. While an end-to-end digital supply chain is an opportunity as it unleashes the next level of strategic growth and involves minimal investment in infrastructure, it is still a challenge to optimize and transform. Furthermore, the recent pandemics and geopolitical disruptions of Covid-19, the Ukraine-Russian war, Brexit and the US-China trade war; have structurally changed the global economy and revealed a new risk assessment that will result in the re-introduction of buffers, boundaries across industries and a partial return to regionalization with sort of de-globalization in which existing just-in-time getting replaced by just-in-case strategy.

Artificial Intelligence technology has become increasingly important from a business perspective. In strategic sourcing, the technology has not been explored much. However, 67% of CPO:s in a survey showed that AI is one of their top priorities the next 10 years. AI can be used to identify patterns, predict prices and provide support in decision making. A qualitative case study has been performed in a strategic sourcing function at a large size global industrial company where the purpose has been to investigate how applicable AI is in the strategic sourcing process at The Case Company. In order to achieve the purpose of this study, it has been important to understand the strategic sourcing process and understand what AI technology is and what it is capable of in strategic sourcing. Based on the empirical data collection combined with literature, opportunities of applying AI in strategic sourcing have been identified and key areas for an implementation have been suggested. These include Forecasting, Spend Analysis & Savings Tracking, Supplier Risk Management, Supplier Identification & Selection, RFQ process, Negotiation process, Contract Management and Supplier Performance Management. These key areas have followed the framework identified in the literature study while identifying and adding new factors. It also seemed important to consider factors such as challenges and risks, readiness and maturity as well as factors that seems to be important to consider in order to enable an implementation. To assess how mature and ready the strategic sourcing function is for an implementation, some of the previous digital projects including AI technologies have been mapped and analysed. Based on the identified key areas of opportunities of applying AI, use cases and corresponding benefits of applying AI have been suggested. A guideline including important factors to consider if applying the technology has also been provided. However, it has been concluded that there might be beneficial to start with a smaller use case and then scale it up. Also as the strategic sourcing function has been establishing a spend analytics platform for the indirect team, there might be a good start to evaluate that project and then apply AI on top of the existing solution. Other factors to consider are ensuring data quality and security, align with top management as well as demonstrate the advantages AI can provide in terms of increased efficiency and cost savings. The entire strategic sourcing function should be involved in an AI project and the focus should not only be on technological aspect but also on soft factors including change management and working agile in order to successfully apply AI in strategic sourcing. / Artificiell Intelligens har blivit allt viktigare ur ett affärsperspektiv. När det gäller strategiskt inköp har tekniken inte undersökts lika mycket tidigare. Hursomhelst, 67% av alla tillfrågade CPO:er i en enkät ansåg att AI är en av deras topprioriteringar de kommande tio åren. AI kan exempelvis identifiera mönster, förutspå priser samt ge support inom beslutsfattning. En kvalitativ fallstudie har utförts i en strategisk inköpsfunktion hos ett globalt industriföretag där syftet har varit att undersöka hur tillämpbart AI är i strategiskt inköp hos Case-Företaget. För att uppnå syftet med denna studie har det varit viktigt att förstå vad den strategiska inköpsprocessen omfattas av samt vad AI-teknologi är och vad den är kapabel till inom strategiskt inköp. Därför har litteraturstudien gjorts för att undersöka hur man använt AI inom strategiskt inköp tidigare och vilka fördelar som finns. Baserat på empirisk datainsamling kombinerat med litteratur har nyckelområden för att applicera AI inom strategiskt inköp föreslagits inkluderat forecasting, spendanalys & besparingsspårning, riskhantering av leverantörer, leverantörsidentifikation och val, RFQ-processen, förhandlingsprocessen, kontrakthantering samt uppföljning av leverantörsprestation. Dessa nyckelområden har följt det ramverk som skapats i litteraturstudien samtidigt som nya faktorer har identifierats och lagts till då de ansetts som viktiga. För att tillämpa AI i strategiska inköpsprocessen måste Case-Företaget överväga andra aspekter än var i inköpsprocessen de kan dra nytta av AI mest. Faktorer som utmaningar och risker, beredskap och mognad samt faktorer som ansetts viktiga att beakta för att möjliggöra en implementering har identifierats. För att bedöma hur mogen och redo den strategiska inköpsfunktionen hos Case-Företaget är för en implementering har några av de tidigare digitala projekten inklusive AI-teknik kartlagts och analyserats. Det har emellertid konstaterats att det kan vara fördelaktigt för strategiskt inköp att börja med ett mindre användningsområde och sedan skala upp det. Eftersom strategiska inköpsfunktionen har implementerat en spendanalys plattform kan det vara en bra start att utvärdera det projektet och sedan tillämpa AI ovanpå den befintliga lösningen. Andra faktorer att beakta är att försäkra datakvalitet och säkerhet, involvera ledningen samt lyfta vilka fördelar AI kan ge i form av ökad effektivitet och kostnadsbesparingar. Därtill är det viktigt att inkludera hela strategiska inköps-funktionen samt att inte endast beakta den tekniska aspekten utan också mjuka faktorer så som change management och agila metoder.

本研究旨在探討研發取得策略，情境因子，研發人力資本，與公司績效之間的關聯性。藉由理論模型的推導與實務訪談研發中心主管來發展假說，進而以混合資料模型(pooled data models)與橫斷面資料模型來進行實證分析。主要結果為：當公司的研發多樣化程度較高，專利權數目較少，公司較可能採取內外部研發並進策略(R&D hybrid strategy)，而非完全內部研發策略(make strategy)。最重要地，過度/低度的外部研發取得(under-/over-external R&D sourcing)對於公司績效有負/正向影響。相較於會計績效，此效果對市場績效的影響尤其顯著。同時，研發取得策略對於公司績效的影響，取決於公司的研發人力資本。相較於採取完全內部研發策略的公司，為了吸收外來的異質技術，採取內外部研發並進策略的公司較可能聘任不同研發種類的研發人員；而如此的研發取得策略與研發人力資本契合將進而改善會計與市場績效。本研究的發現與交易成本理論與supermodular理論一致。如同預期，相較於橫斷面資料模型，本研究的實證分析在混合資料模型較為顯著。 / The association between performance and R&D sourcing strategy in relation to contextual variables and R&D human capital is determined by the analytical model coupled with field interviews with directors or managers in R&D centers. Capitalizing on a unique database of the 2001-2003 Taiwan Industry R&D Investment Survey containing more detailed information available on the R&D activities to verify this association, the researcher tests the empirical results by cross-sectional and pooled data models. The findings are that an innovating firm will prefer to implement the R&D hybrid strategy when the higher degrees of R&D diversity and fewer counts of patents are exhibited. Perhaps most importantly, this study shows compelling evidence that over-/under-external R&D sourcing affects negatively/positively a firm’s performance. This effect is more significant in the market-based performance (Tobin’s q and average two-year Tobin’s q) than accounting-based performance (ROS and ROA). Moreover, the associations between R&D sourcing strategy and a firm’s performance are contingent on the use of R&D human capital. Innovative firms with the R&D hybrid/make strategy are more/less prone to employee diverse types of R&D experts to absorb the coming external knowledge, and such alignment between R&D sourcing strategy and R&D human capital thus improves both accounting- and market-based performance. The results are consistent with both transaction cost paradigms that discriminating alignment of transactions with strategy leads to more efficient outcomes, and the supermodularity model that a firm’s performance is a function of coherent alignment between strategy and structural elements of an organization. As predicted, these effects are noticeable and more pronounced in the pooled data model than in the cross-sectional design.