Browsing by Author "Karlander, William"

Now showing 1 - 3 of 3

Geometrisk numerisk integrering av differentialekvationer
(2022-07-08) Karlander, William; Kyhn, Georg; Sahlin, Erik; University of Gothenburg/Department of Mathematical Science; Göteborgs universitet/Institutionen för matematiska vetenskaper
Rapporten studerar fr¨amst Eulers symplektiska och St¨ormer-Verlets metoder applicerade p˚a hamiltonska problem. Metoderna appliceras numeriskt p˚a tre olika hamiltonska problem och j¨amf¨ors med andra numeriska metoder i form av Eulers explicita metod och Eulers implicita metod. Det f¨orsta hamiltonska problemet som studeras ¨ar en ideal pendel. Det andra ¨ar ett hamiltonskt system best˚aende av tv˚a himlakroppar. Och det tredje ¨ar ett molekyldynamikproblem best˚aende av tv˚a atomer. I exemplen uppvisas egenskaper som sedan definieras och bevisas konkret f¨or generella fall. Av dessa egenskaper ing˚ar definiering av invarianter och symplektiska avbildningar. Bevis av metodernas bevarande av invarianter och bevis av att vissa stegmetoder ¨ar symplektiska genomf¨ors. Det visas ¨aven vilka problem som har symplektiska l¨osningar Keplers problem ¨ar ett av exemplen som granskas, d¨ar modelleras tv˚a himlakroppar med den ena som kretsar runt den andra. De fyra stegmetoderna (Eulers explicita, implicita och symplektiska metoder samt St¨ormer-Verlets metod) appliceras och avvikelser fr˚an den exakta l¨osningen j¨amf¨ors, specifikt visas omloppsbana, avvikelse av energi, vinkelmoment och position skapat av de numeriska metoderna. Rapporten visar att alla hamiltonska system har en symplektisk avbildning. Ytterligare visar rapporten att Eulers symplektiska metod och St¨ormer-Verlets metod ¨ar symplektiska. Det visas ¨aven att symplektiska numeriska metoder ¨ar n¨astan energibevarande p˚a hamiltonska problem. Detta g¨ors genom att utf¨ora simuleringar som numeriskt visar att Eulers symplektiska och St¨ormer-Verlets metoder bevarar energin inom ett begr¨ansat intervall ¨over exponentiellt l˚ang tid. Bakl¨anges felanalys introduceras sedan som omr˚ade f¨or att f¨orklara varf¨or symplektiska numeriska metoder har denna egenskap.
I prompt, therefore I know: Making sense of expertise in an AI world
(2025-06-24) Dan, Pavel; Karlander, William; University of Gothenburg/Graduate School; Göteborgs universitet/Graduate School
The widespread use of generative artificial intelligence (AI) such as large language models is challenging traditional notions of expertise. While AI tools can increase efficiency and aid creativity, they are often prone to hallucinations, might miss contexts and raise challenges around accountability. To better understand how to approach AI, this study explores how AI experts and non-AI experts make sense of AI. Drawing on sensemaking theory and concepts of expertise, this study investigates how individuals with different levels of expertise in AI enact, select and retain information regarding AI. By exploring how the two groups of experts make sense, it is possible to reveal broader implications of AI for different professionals. Through 28 semi-structured interviews with professionals across various industries in Sweden, the study identifies a notable difference between the two groups of experts. Our findings show that AI experts engage with AI tools in a critical, intentional manner, emphasising approaches that mitigate potential drawbacks, such as verifying the material produced with AI and reflecting on limitations of AI. The non-AI experts on the other hand tend to reflect less and act more intuitively, often after limited experiences with AI. These differences in sensemaking approaches indicate that expertise is not only enacted but also situationally and relationally shaped by how these experts interact with the technology in a specific context. The study contributes to management research by providing a guideline for how to categorise AI expertise, insight into what factors affect sensemaking of AI and insight into how organisations can implement AI responsibly for different groups of experts.
Is the EU ETS effective? Examining the impact of the EU ETS on methane emissions
(2023-07-03) Lindsjö, Carl Otto; Karlander, William; University of Gothenburg/Department of Economics; Göteborgs universitet/Institutionen för nationalekonomi med statistik; University of Gothenburg/Department of Business Administration; Göteborgs universitet/Företagsekonomiska institutionen
The European Union Emission Trading System (EU ETS) is the world's largest trading system for emissions. It has a crucial role to play in meeting the goals of the Paris Agreement. However, there has been debate over its effectiveness. This thesis examines how the EU ETS impacts methane emissions. The study uses Swedish emission data from 2005 to 2021 and employs a fixed effect model with two exogenous and two endogenous variables. While the test produced significant results for carbon dioxide emissions, it could not do so for methane emissions. The thesis concludes that EUA prices have a negative effect on carbon dioxide emissions, while electricity prices have a positive effect. The impact on methane emissions could not be determined. Previous research is divided, suggesting that financial solutions like a cap and trade system can have varying effects depending on the sector and perspective. A new model with interaction variables, more control variables, or a different data set could yield significant results.