While working through the close to 250 applications for PhD positions at the International Max Planck Research School (IMPRS) for Astrophysics¹ for the current round of applications, I noticed an odd pattern: Reference letter writers who saw a candidates strengths more in theory than in experiment seemed to rank these students higher compared to those whose strengths are more in experiment.

I have many reasons to like Python: rapid prototyping, powerful libraries, great interactive tools like the IPython shell and notebook, and the ability to write beautiful code. I will admit, the last one definitely is in the eye of the beholder. Brandon Rhodes gave a good talk at PyCon Canada a couple of year ago in which he explained what makes Python beautiful to him. One of the points he made, and one that I wholeheartedly agree with, is that you can make Python code look just like the math you are coding up. We know how to write math; we know how to typeset equations so they look beautiful and are easy to read, and thus easier to understand. We can use that knowledge to write more readable Python code.

But I’m not a mathematician, I’m an astrophysicist. Can I take this further and make my code look like physics, not like math?

Or why “The error bars overlap” is a meaningless statement.¶

In my work as an astrophysicist I have encountered quite a bit of confusion on how to quantify the level of disagreement between two or more measurements with error bars, or even more complex multi-dimensional confidence volumes. This is even more true this term when I am teaching an undergraduate lab course. Thinking a little about this myself, I realized that this is not something I have ever seen written up. At least not in any concise form. So I decided to create this little document, not only to educate others, but also to organize my own thoughts. I put an emphasis on developing the concepts and I make no attempt at rigour.

Suppose we have three measurements of the gravitational acceleration on the Earth’s surface, all with their own error bars (or measurement uncertainties, if you prefer). These are shown in the plot below and the error bars indicate their $1\sigma$ standard deviation.