Graduate level course in computational physics

Question

I want to start the foundations for a new totally open source graduate level course.  I have noticed that a lot of researches don't have solid enough background in elementary computer science skills nor SW development paradigms.  Another issue I want to tackle is the lack of knowledge of statistical, data analysis and measure theory concepts.  So, I would like to write a full semester course to address the above.  I also want the course to be both theoretical with hard hands-on assignments. 

Please let me know which subjects should be included.

I thought to include those:

- DSP and Information theory
- Graphs
- P, NP and NP-completeness
- Intro to Tensorflow, neural networks and deep learning
 

Comments

How can you ever create a good graduate course if you are not even knowledgeable enough to know what to include?

---

@ArnoldNeumaier it might be the case that he is knowledgeable enough about these topics as a researcher but lacks experience in teaching and concepting courses about them ...

Answer 1

I think it would be pretty difficult to have a "general" graduate level computational physics course since research often demand pretty specific computational methods for the problem you are treating. For example, for my graduate level research, I used finite element methods extensively - but there's no way one course could cover FEM in depth by itself - forget all the other numerical methods out there for computational physics.... 

Answer 2

Well, we don't seem to be biting.

Perhaps instead of a ad hoc topics list, you can offer a rough outline showing the development towards areas of study in graduate physics. Or a relevance matrix showing course objective topics and their applicability against the areas of active physics investigation. Such an outline or matrix would help us offer physics problems useful for your course. 

Surely, a new graduate physicist has had fundamental parametric and non-paramentrics and has had  some experience with commercial tools for collecting and analyzing data and displaying observational data.  But, if you are finding this is not the case,  then any overview course outline for statistics applied on a selected (widely used by physicists) analytical tool set will work.  Then your question really is:  are there any active  research or experimentation programs among our forum members that could be offered as case studies supporting these programs or at least as validation of work already done?