I completed a ERAMUS+ exchange at the University of Bergen during my third semester of the Master degree programme Scientific Computing at Heidelberg University in the autumn semester of 2022. The purpose of this report is to summarise my stay and will hopefully be useful to future students considering a semester abroad. In general, I think a semester abroad is very well suited to the Scientific Computing programme, and, despite a rather small Mathematics faculty, the University of Bergen is a reasonable and well organised partner university. I will first touch on my planning of the exchange, then go into some detail of the actual studying - choosing modules, accreditation, impressions of attended courses and such - and finally talk about all other non-university related important things, such as getting there, living, and some noteworthy peculiarities.

Planning the Stay

Scientific Computing students can use the ERASMUS quota of the maths faculty in Heidelberg, which is what I ended up going for while planning the stay. Apparently, it is possible to also use quotas from the computer science faculty, but only if places are left after allocating them to computer science students. Another way of getting abroad that a friend went for is using the 4EU+ partner university network which is, as far as I understand it, basically a simplified ERASMUS stay.

A priority of mine has been to be able to take at least a few ECTS back home, so I wouldn’t have to study for longer than the usual Master’s duration of 4 semesters. This is where the dilemmas of applying for a semester abroad starts, because you need to know which of the partner universities have courses that you can have accredited back home a year before when preparing the ERASMUS application. The quick answer is, you don’t know and have to resort to an educated guess, because the document that specifies how accreditation works for your case, the learning agreement, is only done after having chosen, applied for, and having been admitted to, the stay at the partner university. This is one of the reasons I went for the University of Bergen in the end, because it is a very organised university where it is planned a few semesters in advance which courses happen at what time.

Otherwise, I found the application to be straightforward - multiple applications are done in the process, but the important one seems to be the one to the local ERASMUS office in the beginning, all other ones, including one to the partner university, seem to be formalities.

One curious thing with the application was that a letter of recommendation was needed from a faculty professor - as the application is done so early, I was in my first semester in Heidelberg, and trying to get Professors that don’t know you to produce a letter of recommendation for you is not much fun. Anyhow, after much convincing, I got one of the Professors to do it for me, but still to me this felt like a nonsensical bureaucratic hurdle to me.

After the initial admittance into the program (around February 2022), not much happened until the formal application at the partner university that included the course selection (around April 2022), that has been approved by the Heidelberg Universities’ relevant stakeholders (ERASMUS advisor & examination office) by means of signing the first part of the learning agreement before. The process sounds confusing, because it is (it doesn’t help that the learning agreement has multiple parts that are signed by different people at different times), but it was rather smooth in the end and there wasn’t too much trouble.

Top 3 things I would do differently would I have to plan the same stay again:

1. (If going abroad in the winter semester), plan the summer semester in Heidelberg well. Because Heidelberg semester end and Norway semester start are very close together, and some courses in Heidelberg, particularly computer science ones, think it’s a good idea to grade using projects during semester breaks, I found myself having to do a final project for a Heidelberg course in the first few weeks abroad - while others enjoyed the amazing weather and remaining summertime.

2. Don’t book a return ticket (if flying) - the exact exam dates are a lottery as they are published sometime mid-semester, and I happened to win the lottery, with most final exams already in the beginning of December - quite a few weeks ahead of the official semester end date mid-December.

3. Unfortunately, this info is hard to come by, but put more research into other’s experience with courses before committing them to the learning agreement - as will be detailed in the next section, course quality differs significantly, and obviously such things are hard to notice just from the official course websites.

Study Programme

As I felt the existing semester abroad reports about the University of Bergen in the Math/CS field were lacking in this regard, I will go into some detail regarding the courses I attended in this section, after giving a brief overview of what I attended.

A central part of a semester abroad organised through ERASMUS+ is the learning agreement, which essentially declares which courses are taken at the guest university, and, importantly, how they are credited at the home university. Accreditation essentially works by finding courses that are similar enough to existing courses in the home university to be deemed equivalent by the examination office. Then, attending the similar course at the guest university counts as having attended the equivalent course in the home university for the purposes of the transcript of records.

While this procedure makes it very simple to take ECTS back home, obviously it somewhat restricts the usefulness of a semester abroad - taking courses that are not part of the curriculum of the home university is not really a thing. In my particular case for example I was denied taking a course in Information Theory at the University of Bergen and having that accredited towards my computer science modules, because Information Theory is not taught in Heidelberg - despite being a reasonable field of study for the Scientific Computing master content-wise, at least in my opinion. That being said, it should be noted that effort is taken on the Heidelberg side to get a reasonable learning agreement together, despite some modules not having an exact 1:1 correspondence. The following table that summarises my particular learning agreement exemplifies this.

In the University of Bergen, standard 4+2 SWS courses seem to be worth 10 ECTS, while in Heidelberg it is typically 8 ECTS - as such, I could only get 8 ECTS accredited. I would say this is a fair reduction as I did not really notice a big difference in workload between these 10 ECTS modules and the typical 8 ECTS ones in Heidelberg.

Content-wise, the courses were a mixed bag. I thoroughly enjoyed Computer Graphics and Numerical Linear Algebra, and very much despised Advanced Algorithmic Techniques and Stochastic Processes.

Computer Graphics was my favourite course - Prof. Stefan Bruckner was a highly motivated lecturer with great didactic skills. Content-wise it was essentially a tour de force of the field - after basic data structures and representation followed an introduction into the 2D and 3D viewing pipelines, perspective transformations, and illumination techniques. After these basic topics, more advanced aspects such as ray-tracing and radiosity were discussed. I would describe the lecture as very applied and sometimes wished for a bit deeper dives into theoretical aspects, but the concept worked well overall. Besides the lecture there were two chunks of homework, first, weekly exercises, and second, a project that ran parallel to the project with hand-ins on a bi-weekly schedule. The exercises were very much on the easy side and could be completed rather quickly - the project on the other hand was quite complex. Basically, a codebase containing an existing C++/OpenGL 3D renderer had to be incrementally extended with various aspects discussed in the lecture, such as lighting, texturing, and animation. The project was graded and influenced the final grade by 50%, along with a standard written exam that was rather easy - it was a mix of computer graphics related maths problems and knowledge questions. The maths problems in particular did not go beyond very basic linear algebra. The weekly exercises had no influence on the final grade, but 50% had to be completed in order to get admitted to the final exam.

Numerical Linear Algebra was a classic blackboard lecture that was rather well executed by Prof. Malychev. Content-wise the lecture heavily relied on the classical textbook Numerical Linear Algebra by Trefethen & Bau, which is well-written with students in mind and thus makes self-study easy. Occasionally though, Prof. Malychev would deviate from the textbook and talk about some other interesting aspects of Numerical Linear Algebra not covered in the textbook such as Weyl’s Theorem on Eigenvalues. The weekly exercises were cherry-picked ones from the textbook. The exercises were not graded or checked, as such their only purpose was to reinforce contents of the previous week’s lectures. The exercises would be discussed and solved by Prof. Malychev himself in the weekly exercise sessions - I would have preferred a more classic exercise session style with a teaching assistant where student’s solutions were discussed to get more interaction and insights, but this worked as well. Exam admittance was handled through two obligatory assignments throughout the semester. Both had to be passed to get exam admittance. The first obligatory was math-heavy, the second was programming-heavy (MATLAB). No clear grading criteria was given for these obligatories (such as: you need to solve x% of the exercises to pass) - instead we have been told by the Prof. that he would look at all submitted assignments and then decide who passed and failed afterwards - a terrible practice in my opinion, however everyone who made a reasonable effort passed the obligatories, so at least grading was fair. Grading of the course was handled through an oral examination - only about 8 students were in the course - which lasted 45 minutes. A relatively lengthy list of topics (covering basically the entire Trefethen & Bau textbook except some less important algorithms and theorems, plus areas where Prof. Malychev deviated from it) was published right at the start of the semester, and during the oral exam, questions and exercises relating to these topics were asked (e.g.: Proof that this and that algorithm is backward stable, what does this and that Theorem say). Grading was fair and reflected the knowledge of the topic accurately for all the students that I spoke to in my opinion.

Advanced Algorithmic Techniques was a course focused on Parameterized Complexity, the main research area of Dr. Petr Golovach, the lecturer. Some other algorithmic techniques such as randomized algorithms and approximate solvers were also discussed, but to a much lesser extent. The course was heavy on graph theory and for all attendees, including the teaching assistant, a PhD student on the topic, extremely difficult. The lectures themselves were densely packed Latex-Beamer presentations. Weekly exercises would either be vague graph-theoretic formulations of problems, requesting for example proof that the problem was in a certain complexity class, or taken from Parameterized Complexity, an advanced textbook on the topic. Very similarly to Numerical Linear Algebra, exercises would have no effect on grading and the like, but two obligatory assignments had to be passed for exam admittance, which also had an influence on the final grade. Final grading would happen through an oral exam. The obligatory assignment was unreasonably difficult - one exercise contained a problem we later found out was only solved in a 2015 paper, another exercise the teaching assistant failed to even understand so asked Dr. Golovach for help, and we have been told they spent over an hour together trying to figure out the exercise and did not reach a complete solution in the end. The first incident I understand - maybe the Professor thought this was just an easy problem to solve given what was said in the lecture, the second I don’t - why would a Professor ask a question he did not know how to solve himself. Originally excited about the lecture and topic, I ditched this course mid-semester, some time after the first obligatory assignment because I got too annoyed by these sorts of things and didn’t really need the credits. That said, the exercise sessions with the teaching assistant were great - as everyone was in the same situation of basically not understanding anything, figuring out the exercises as a team was fun and rewarding. To summarise, except with a deep interest in parameterized complexity I can’t really recommend taking this course.

Stochastic Processes, teached by Prof. Hans Karlssen, was not unfortunately not a joy to attend, either. The course textbook by Ross is a good, definitive work, but this course only covers a tiny subset of chapters, and I thus find it inadequate - working with it is very difficult as there are backreferences to chapters that are not part of the curriculum, etc. The cross-references from the Lecture to the textbook were sometimes inadequate to find the relevant part of the textbook, especially the part related to Poisson Processes. The lecture is wholly unstructured, we frequently had to ask for clarification as to what part of the curriculum Prof. Karlsen was now talking about. There were frequent jumps in topics, for example during the part about Poisson Processes Prof. Karlsen would sometimes jump back to the chapter on Markov Chains (the previous chapter) to "finish something", i.e., present some theorem or result he missed talking about at earlier times but wanted to. These were usually unrelated or only marginally related to the "main topic" of the lecture. The lecture was sometimes on blackboard, sometimes on slides. The slides were not provided in a timely manner to study them before or after the lecture again, sometimes incomplete and updated only at a much later time. Still, the slides had the clearest structure of all content of the lecture. For the blackboard parts, they were largely unstructured scribbles, and I failed to properly produce lecture notes from them, and eventually stopped taking notes because it just made no sense. The entire situation was not helped by the fact Prof. Karlssen speaks very broken English, sometimes even switching back to Norwegian without noticing it. The exercises and exercise sessions were frankly a disaster, as the teaching assistant was not knowledgeable about the lecture contents in any way beyond some basic R programming (about half of the exercises was R programming, the other half were mathematical problems). A tutorial session would look like this: the teaching assistant would arrive 15 minutes late, then arrive, sit there in silence for the remaining time while the students would attempt to solve the exercises, and then leave 15 minutes early. By the end of the semester, only 1-2 students would bother coming to the lecture (around 15 initially), because it was, to be frank, a waste of time. To summarise, it was not a joy to attend and I hope it didn’t ruin the subject for me, because the topic in itself is actually quite interesting - I can only recommend staying away from this lecturer though.

Living in Norway

At the time of writing, with the University of Bergen, international students are guaranteed a place to stay during a semester abroad. Typically, they are allocated into Fantoft, a suburban neighbourhood of Bergen sporting a huge student housing unit with capacity of probably around 2000 students (judging from the amount of postboxes). Consequently, most students at Fantoft are internationals. Norwegians are around as well, but they are a clear minority. The majorities of the international student population are from Germany and Spain. Being a German student myself, the fact that a majority of international students are German as well quickly became mildly annoying, because both the Germans and the Spanish then tended to clump together into big groups where only German and Spanish is spoken - thereby basically excluding students of other nationalities. While trying to avoid being a part of this effect as much as possible, sometimes a day would go by without having had to speak English once!

These interesting social phenomena aside, I found Fantoft a quite enjoyable place to live - don’t believe the horror stories, but don’t get your expectations high either - obviously the experience will most heavily depend on the type of people you end up living with, which is hard to control. This is amplified by the fact that 1-semester stays are practically always allocated into a room that is shared with another student (though I know of a few exceptions). Because everyone involved is uncomfortable with this situation though, typically the impression I got was that people are respectful and try to make the situation work - at least this was the case for my roommate (another German, of course). One friend just couldn’t get the relationship with her roommate to work out though, but by talking to Sammen, the student housing company managing the place, she managed to move rooms, so this is possible, should things ever not work out. I was a bit surprised that the shared room situation didn’t end up bothering me very much - in fact I got used to it very quickly. Other than that, Fantoft has all amenities you would expect, cheap laundry rooms, study halls, common rooms, and even a little student-organised bar hosting the occasional social event. Being a student village, sometimes it gets a bit loud at night, but the windows are rather well sound isolated.

A final word to lose about the housing at Fantoft - it is basically made out of two parts, the old building, and the new one. The old building consists mainly of “apartment-style” units - around 20m^2, small kitchen, desk, rest room, fridge, bunk bed - that are shared by two people. The new one consists mainly of “flat share”-like units with a capacity of 16, each individual room being shared by 2 people, but with a huge communal kitchen and living space in the centre. If looking for social contacts, try to get into the new building (it is possible to issue preferences when applying)! They are much, much nicer, and beyond the own roommate, social interaction in the apartment-style old complex is very difficult.

Still in Fantoft is a huge Gym that also houses staff from Sammen, the organisation managing the place, that handles all tenancy related things and that can be contacted in case of any problems. I highly recommend getting the semester membership to the Gym - it’s insanely cheap at around 100€ for the whole semester, and gives access to 2 climbing walls, a big strength section, a Sauna, and lots of endurance equipment such as exercise bikes and treadmills as well.

Getting Groceries is also possible without having to travel far, with a store from chain “Meny” being located right next to the student housing. It’s a fine shop for everyday needs, but I found the chain “Kiwi” is the cheapest, along with the international/asian markets like Global Foods and Mango in the city centre.

Fantoft is very conveniently placed directly midway along a light rail line called Bybanen connecting Bergen airport with Bergen city centre - if arriving and departing from Bergen airport, this makes things very comfortable, but be aware the light rail doesn’t run during night hours, so a flight during very early or late hours might be a problem. Going to the city centre - where the natural science (“Realfagbygget”, Rail stop: “Nygard”) and computer science buildings (“Høyteknologisenteret”, Rail stop: “Florida”, the stop before Nygard) are located takes around 20 minutes. Public transport is cheap for students, with a monthly ticket (allowing all public transport along a relatively large radius around Bergen) around 50€.

That Norway is an expensive country goes without saying - checking prices of common items can be done before arriving after all. What surprised me though is that despite the high prices, the quality is not even good. The cucumber you buy in Norway for 25kr (2.5€ at time of writing) is the same cheapo produced in Spain non-organic one you get in Germany for 0.7€.

Continuing with Money, I used around 1000€/month all-in, including rent, food, and the occasional beer and fun thing - on what I would consider a relatively frugal life-style. For example, I did not eat at a restaurant once during the entire stay - I just couldn’t bring myself to pay 14€ for a Döner. Needless to say, the ERASMUS grant money does not even begin to cover these expenses fully, be prepared to apply for secondary funding sources like Auslandsbafög, or pay out of your own pocket.

I will lose a few words about the weather, because it seems to be a huge concern for a lot of people coming to Norway in the Winter - it was totally fine. The first few months were very sunny and the weather was great for hiking and enjoying nature, and obviously, the further it got into Winter, the worse the weather became, with rain on many days. But by no means did it rain every day. Maybe I just picked a lucky year though - I heard the Summer 2022 in Bergen was very rainy in comparison.

Most semester abroad people I met came to Bergen to enjoy the incredible nature that lies practically directly on the doorstep, and, having never been to Scandinavia before, this was a big reason for me as well. The majority of Norwegians and international students coming to Bergen are outdoorsy people - finding people to plan hikes, surfing or skiing trips with is not difficult at all. A great volunteer organisation exists, BSI Friluft, where it is possible to become a member and where it is possible to rent equipment for cheap prices, and go on trips that basically happen every weekend. To get an impression of the kind of trips that are possible and that are offered, it is worth checking out the archive of the mailing list [1], where the trips are announced.

Speaking of volunteer organisations, one other great one is BUA, where students can rent equipment such as tents, sleeping bags, skis, and other outdoor stuff for free or very cheap - great for trying out a new sport without committing too much money.

In Conclusion

While being let down a little bit by some of the courses, I still learned a lot and had a great time living in Norway. My experience with all relevant people on the Heidelberg side I encountered on the bureaucratic journey of making the semester abroad happen was very positive, and I’m thankful for it. People really are motivated to send you abroad and making some reasonable studying and accreditation possible. All I can say in conclusion is that a semester abroad really is worth the bureaucratic hassle and planning, I’m happy to have done it, and can recommend it to anyone to do this at some point during their studies!

I will conclude this report with some unstructured notes and anecdotes about the stay that didn’t really fit anywhere else.

Differences of University of Bergen to Heidelberg University

In no particular order:

• Nobody studying in Heidelberg is going to believe this, but there is one (!) e-learning solution, and EVERYONE is using it! Great to see that other universities don’t need 2 different Moodle instances, Mampf, Muesli, Discord, and whatever else to manage their courses, and it really is only Heidelberg with this insanity.

• IT in general is executed very well - one central login gives access to everything, there are video cameras in all major lecture halls that know the course schedule and automatically upload course recordings to the e-learning platform, a digitally signed transcript of records can be ordered automatically at the end of studies, and so on.

• That said, only one course I attended was in a lecture hall with such camera equipment and also made use of it, the others were so small we were in seminar rooms most of the time.

• Working IT is nice to see as a CS/Math student, but based on the limited selection of lectures I heard in Bergen, I draw the conclusion that course quality in Heidelberg is to a much higher standard. While some courses are better structured than others in Heidelberg, I never felt the need to abort attending a course mid-semester because of bad structure and quality.

• I can say a few more things about other lectures I didn’t attend but some friends attended, but take this with a grain of salt: two otherwise competent math students I knew dropped the Partial Differential Equations course, saying it was too work intensive and difficult. The Introductory Machine Learning class was generally praised though people said it was work intensive, but basic concepts like decision trees were mixed with advanced “Sexy” stuff such as Neural Networks. Having heard that I’m happy to have done the FML/AML cycle in Heidelberg instead, because I think separating classic machine learning and deep learning into two separate lectures makes a lot of sense. Finally, I heard some good things about the Fluid Flow in Porous Media class, but do not know details.

Differences in (Student) Life

Also, no particular order:

• Starting a Master in Heidelberg, maybe even new in the city? A nice speech (or email) by the rector, a pub-crawl and a campus tour is all you can expect. What Bergen does for their introduction week completely blows away everything I ever saw or heard from a German university. The week prior to where lectures start, Fadderuken (Mentor’s week) can only be described as a week-long festival consisting of pub-crawls, partying and even less alcohol intensive activities such as hiking. I highly recommend arriving for this week, and doing some activities - it is one of very few opportunities to connect with the Norwegian students. The connections I made during Fadderuken made the whole stay much more pleasant because you know at least some people you see every day.

• For small events or informal activities, everyone pays everyone else with Vipps, basically the Norwegian copy of Paypal - too bad you won’t be able to use it, because you need a Norwegian bank account to register.

• Do you like climbing and would you like to use the nice, cheap climbing gym conveniently placed right next to Fantoft? Be prepared for a world of bureaucracy, because you first need someone to test your belaying skills (€€), then you need to register to receive the so-called Brattkort, a card saying you know what you’re doing (€€), and finally have it shipped to you (€). In total the process takes around 3 weeks and 50 euros of basically wasted money. Norwegians take this very seriously, so there is no way around this. Some great climbing gyms are around though, most notably Vestveggen, the biggest climbing wall in Scandinavia, just a 20-minute bus ride from Bergen - so it’s worth it in the end.

• I initially planned to get a bike to get around, but didn’t in the end for two reasons: firstly, it’s actually not very easy to buy a used bike - the equivalent to eBay Kleinanzeigen is not very active, and secondly, public transport is just very good and also quite cheap.

• Hiking trip suggestion: Nærøyfjord is a beautiful Fjord, one of the few world heritage sites in Norway, around 2h bus journey from Bergen. Grindaflethytta is a nice DNT cabin to stay - it takes a around 6h hike to reach.

• Relaxed cabin trip suggestion: Høgabu is a DNT cabin that’s quite close to Bergen, around a 1h30 bus/train ride away, and just takes a 2-3h hike to reach. Nice, old, cabin in beautiful nature that’s a great destination for a relaxing stay.

• Skiing suggestion: the rail link between Bergen and Oslo, Bergensbanen, passes over mountainous terrain, the highest point of the rail link is at 1200m (starting from sea level), making skiing and winter sport activities quite accessible. Get out at Finse and stay at the conveniently placed DNT cabin 10 minutes from the rail station for a more ungroomed, backcountry skiing trip, or get out at Geilo, for more ski-resort, downhill, or groomed cross country skiing.

[1]: BSI Friluft Mailing list, can at the time of writing (Jan 2023) be found here, if the link is dead, check archive.org and related sources: https://mailman.uib.no/public/bsi-friluft/