Sem Sælands vei 24
How did two physicists end up working at Norway’s largest bank? Milan completed his PhD in experimental physics at the University of Western Australia in 2007, while Vilde did her master degree in computational physics at the University of Oslo in 2018.
This Wednesdays Karl Henrik will talk about the programming language Julia and exciting things which can be done with this programming language.
Karl Henrik is working on Machine learning applied to quantum mechanical problems for his MSc thesis.
Marcos Caballero: Integrating computation in American high schools - a tale of resourcing, federalism, and equity
Tor Ole Odden: Everything can be a vector! An approach to teaching machine learning to early physics students?
Joao Inacio is MSc student CS:Physics 1st year. After the seminar we will have pizza.
This week we have the pleasure of having Anne Marthine Rustad and Signe Riemer-Sørensen from SINTEF and the Mathematics and Cybernetics department.
SINTEF is a very important employer in our field of study and we look forward to hear about exciting research and job possibilities. Earlier this semester we forwarded several summer job options from SINTEF.
Philip Sørli Niane is master of Science student in CS:physics, 2nd year. Pizza thereafter and plenty of time for discussions. This is also a topic of interest for potential master of science thesis projects.
Statistics courses typically come in two different flavors: 1) A first type of course seeks deep understanding based on mathematical reasoning and proofs. 2) A second type of course focuses on practical applications, primarily based on fixed recipes without delving into underlying details. 3) I will here propose a third way, which seeks understanding of fundamental concepts, but through programming and simulation instead of mathematical proofs. This third way is thus geared towards students that have a stronger background in programming than mathematics, and that seek to understand the fundamentals as a basis for developing statistical analyses and data science methodology. This third way would also exploit its basis in programming to promote self-driven exploration, drawing inspiration from how computing has been integrated into science education through the long-running CSE initiative. I will provide examples ranging from how to explore the central limit theorem to performing Bayesian marginalization.
In the natural sciences we apply the scientific method in order to uncover the basic rules and principles that govern nature: we make observations, formulate hypotheses, test these hypotheses with experiments and develop theories. Rinse and repeat. But what if you are unable to conduct experiments? Or if the underlying rules suddenly change? These are the difficulties an economists are faced with. In my talk I hope to shed some light on general economic thought and how one goes about modelling “the economic machine” that emerges through human interaction. If the time allow, I will also provide an example of economic cost/benefit analysis done by Menon Economics for the Norwegian Coastal Administration.
Pizza after the seminar.
Industrial research and development are to a great extent the art of extending existing tools and methods with new techniques and data with the objective to optimise and improve operations. Expert Analytics is a Oslo based consultancy working in the fields of data and computational sciences, and in this presentation we will briefly tell you who we are and walk you through a few examples of what we have done for some of our clients. We will also present an internal product development project, where we make sense out of sensors, and where we're planning to open for summer internships next year for help with data exploration.
Stian may also say something about possible master of science thesis projects. He is PhD student.
In this talk, I will briefly describe four different perspectives from the educational research literature on what it means to learn scientific computing: 1) modeling, 2) practices, 3) computational thinking, and 4) computational literacy.
I will also touch on the epistemologies inherent to each theory - what they have to say about scientific knowledge, cite key references, and suggest instructional approaches tailored to each approach.
Title: Computational Thinking and Computational Fluency: Past, Present, and Future
Intergrating computing in the diciplines change both the subject and the way we teach. How do we meet these changes?
Constructive alignment of learning aims, examination and learning activities is a convincing principle for course design. However, to make a meaningful alignment, we first need to precisely define and understand these aspects in the context of a given course. For instance, if learning aims are to truly guide/define examination and learning activities, we need a rich conception of learning aims that goes well beyond the short, general and typically vague formulations provided on course web pages. As always, devil is in the details. To initiate a discussion, I will show examples of how we have tried to face the challenge of being sufficiently concrete in our approach to constructive alignment in the course IN1000 (a large introductory programming course at UiO).
After having taught computing in a mathematical setting for many years, I have collected a number of observations and hypotheses about how students develop their understanding. In particular I have seen many examples of how computing triggers understanding of mathematical concepts. Some of these are quite standard, eg. numerical integration as a tool to extend the understanding of the definition of the integral. In this talk I will discuss some fairly obvious examples of this kind, but I will focus on some less obvious examples. Hopefully, this will illustrate that computing can be integrated fruitfully with mathematics in areas far beyond what we do today.
In the new CompSci MSCA PhD program we will educate 32 PHD-students across the faculty. They will all have a 20 ECTS intensive introduction course in computing and data science. This will open new opportunities to develop PhD-level courses in the spirit of the CS program. I will introduce a discussion of what the contents and level of such courses should be with a particular focus on how Fys4150 and Fys-stk4155 can be adapted to students with diverse computational skills.
It would be a mistake to assume that students have learned the thing you just presented to them. Formative assessment is thus concerned with informing both the teacher and the student about how much students understand about a topic, and discover any misunderstandings.
The seminar will be in Norwegian: "Vi presenterer en modell for kompetanseheving i realfaglig programmering for lærere. Modellen tar for seg opplæring i programmering på fagenes premisser, og vi ser på hvordan en slik modell kan brukes for lærere i høyere utdanning."
Computational thinking are by some defined as the capability to resolve problems algorithmically and logically, including skills related to representing, organizing and identifying patterns in data. This may be seen as leaning in a direction of discrete and observable processes. The Norwegian translation to "algoritmisk tenkning" can be read even clearer in the direction of defining explicit, deterministic instructions to achieve a well understood outcome. At the same time, computational thinking is not only promoted as a means to allow the development of concrete code/algorithms, but also as a way of thinking constructively about phenomena in a variety of fields. And it is clearly not the case that all phenomena in nature and society only involve discrete, directly observed entities - to the contrary, many relations and processes we may be interested in are continuous and probabilistic in their nature, where we have to constructively relate to risks, uncertainties and underlying patterns. An interesting question is whether the ability to devise algorithms to solve well defined problems and the ability to relate constructively to questions in an uncertain world should be seen as two aspects of the same skillset, or as separate skills that are cultivated through separate learning experiences.
What do we mean by "learning programming", why do many find it challenging to learn, what are really the main challenges, which aspects do people in different settings need to learn and how do they best learn it? Geir Kjetil Sandve will give a brief introduction to a discussion where we look forward to hearing experiences and viewpoints inspired by the myriad settings where people in our environment have been involved in programming.
Tor Ole Odden:
"The field of educational research has a massive literature base, with many journals that have been publishing articles for almost a century (or longer). How do we sort through and make sense of literature at this scale? We have begun using an unsupervised machine learning technique from the field of natural language processing, known as latent Dirichlet allocation, to analyze articles from the fields of physics education research and science education research. This technique allows us to extract latent themes, or topics, from the literature and quantify the rise and fall of those topics over time.
In this talk, I will present the basics of the technique, describe some of its underlying theory and applications, and showcase some of the trends that it reveals in how science education theory and practice has evolved over the last 20-100 years."
Doctoral candidate John Mark Aiken at the Department of Physics, Faculty of Mathematics and Natural Sciences, is defending the thesis "Understanding University Student Pathways Towards Graduation with Machine Learning and Institutional Data" for the degree of Philosophiae Doctor.
Computer science was originally invented as a tool to support learning in other disciplines, including engineering and economics. Today, most of computer science education is aimed at preparing future software developers. How do we broaden the appeal of and access to computer science education, to something closer to what the inventors of the field had in mind?