I started my university education in 1996 at the University of Oslo, Norway, and completed my Master’s degree in astronomy in 2000. The topic of my Master’s thesis was N-point correlation functions and non-Gaussianity. During the course of this work, my supervisor, professor Per Lilje, brought me into contact with Drs. Kris Gorski (ESO at the time) and Tony Banday (MPA). The result was an early paper in which we analyzed the old COBE-DMR data using real-space four-point correlation functions for the first time.

“Until WMAP, no strong violations of [the cosmological] principle had ever been observed, but now we, and others, were starting to find some surprises here and there.”

This paper was published in 2002, half a year before the first WMAP release in February 2003. I had by then started my graduate studies (also at the University of Oslo) and was therefore ready to work on the WMAP data right away, using similar tools and techniques.

I have always been interested in questions regarding the origin of the universe, and was quite sure I wanted to work in cosmology since my second year in high school. Then, when I had to choose a topic for my Master’s studies, there were only two realistic options, namely theoretical general relativity in the Physics department or CMB analysis in the Astronomy department. It was more chance than anything else that I ended up with CMB analysis—the GR professor wasn't around the week I had to decide—but in retrospect I think it was a good choice!

I'm not a member of the WMAP team. However, the WMAP data are publicly available (big thanks to NASA!), and anybody can download them and do whatever they want with them. Also, since I am a member of Planck, the next generation CMB satellite, it was only natural to start working with WMAP data as soon as they became available. And it turned out to be a lot of fun as well!

After publishing the first paper on the old DMR data together with Banday, Gorski, and Lilje, we wanted to repeat the analysis with the new WMAP data. However, since N-point correlation functions are computationally quite expensive, we decided to try splitting the data into two halves, analyzing one hemisphere at a time. And then we found a most surprising thing: the functions we found were very different in one hemisphere than in the other! Specifically, while one hemisphere was populated by lots of bumps and wiggles, the opposite was virtually flat. This can perhaps be compared to being on a sailboat at sea one day—in the south, we see a violent storm raging, while in the north, it is sunny and quiet.

The problem is that a fundamental assumption of modern cosmology is the so-called "cosmological principle," namely that the universe should be isotropic and homogeneous. It should look the same everywhere and in all directions. Until WMAP, no strong violations of this principle had ever been observed, but now we, and others, were starting to find some surprises here and there. However, the statistical significance of these detections are only marginal, typically at the 99% level, and it is therefore difficult to really tell if these things happened by chance, or whether something is really amiss.

After observing this first asymmetry, we contacted Dr. Frode Hansen, previously a student of Kris Gorski and a post-doc in Rome at the time, who had developed a complementary technique for measuring the same things. We asked him to split the sky into two, but we did not say what to expect. When he came back with similar results after a couple of days, we went ahead and drafted the paper.

If the findings are more than "bad luck" at the 1% level, it would have dramatic consequences for modern cosmology, since the cosmological principle would have to be abandoned. On a practical level, it will be impossible to simplify Einstein's equations for the universe the way we do it today, and all calculations will be infinitely more complicated. On a fundamental level, it would imply that inflation, the quantum-mechanical process we today believe created all structures in the universe, must be revised.

It will certainly be interesting to get access to high-sensitivity polarization maps from Planck, with which this issue can be studied further. Until then, it is quite likely that this feature will remain a peculiarity with uncertain interpretation.

Actually, I finished a new paper on this very same issue today, using the three-year WMAP data that were released in March 2006. With a quite different and, in my opinion, much more elegant technique, we address the question of statistical significance once more. This has been an issue of great debate since the first publication, and deserves some more attention. Using this Bayesian approach, we find that the evidence is indeed substantial, although not decisive. Perhaps somewhat unsatisfactory—but highly intriguing.

Beyond that, I spend most of my time on preparations for Planck, mostly developing algorithms.