The challenge of providing Ph.D. students in Physics with an overview of statistical methods and concepts useful for data analysis in just three hours of lectures is definitely a serious one, so I decided to take it as I got invited to the "Indian Summer School" in the pleasant lakeside town of Traunkirchen, Austria. 
This is the first school organized here for Austrian Ph.D. students. It features lectures in topics ranging from searches for new physics at colliders, to cosmology and dark matter searches, to QCD and quark-gluon plasma, to statistics for physicists. I was happy to see that there is lot of time allocated by the organizers for "hands-on" sessions -in the afternoon- where students can interact with the lecturers and solve exercises offered to them. Supervised solution of exercises is the most effective way for practical learning, especially for the topic I have to teach.



Above, the town. the building in front, with the docks, is my hotel; the academy is just left of it.

The place is also well suited. There is an "international academy" in a characteristic old-fashioned building facing the lake, with everything needed for a straightforward organization. And there are two very nice hotels at two minutes walking distance, also facing the lake. Furthermore, here the tourist season is basically over, so there is only us (students and teachers, plus organizers) and the locals around. A quiet place for meditation and study!

As I gave my first lecture, which dealt with point estimates, estimators, and their properties, I was confronted with a dilemma I have had to face a few other times in the past. Should I insist that the level of lectures be suited to the knowledge base I assume 2nd year graduate students should have on the matter, or try to gauge the appropriate level from my audience -with the almost certain outcome of having to lower it significantly ? I am not implying that the students are below average, at all! The opposite is true - it is my fault: I have unrealistic expectations.

Should a PhD student automatically know what is an expectation value ? Or how to write down a likelihood function ? Or what is a covariance matrix ? The answer to the above is no, they are not necessarily supposed to. It depends on what is their background, and what they do in their day-to-day work. So all those things, and the relative introductory material, have to be taught with patience to start with. But then, where do we get in three hours ? Not very far. 

That is the dilemma: if I tune the level of my lectures such that students can follow them without struggling, we do not cover a lot of grounds, and my presence at the school is basically unjustified, as a reader of any textbook would do a better job than me and my confusing slides. I am tempted to somehow think that by exposing students to material that is at least in part above their head, I have a chance of stimulating their interest and curiosity to learn enough to make sense, one day, of my slides in full. 

In the end, today's lesson was tough but not overwhelmingly so, at least according to some reactions I collected during dinner tonight. We'll see how it goes when we start covering the more complex material of the next two lessons... I will post some sample material here so maybe you will be able to give me your own opinion.