plotting my own doom ..

Or as it's more commonly known here, writing a syllabus of topics for my Prelim Exam. A good bit of them I've seen before, but it still seems a scary task to remember it all .. or at least enough to keep my committee happy.

If you're curious, here is a tentative breakdown of the subjects:

1. Preliminaries
   
2. Lips¢hitz Functions and C0nvex Functions
   
3. Sob01ev Spaces and Fine Properties
   
4. Highlights from Ge0metric Mea$ure The0ry

Believe me, you can say a lot about each heading. I only hope that I can, too. Wish me luck!

---

As I was browsing through books for syllabus items, I found this theorem in a book of P. Mattila. The result is credited to P. Jones, which generalises a previous result of G. David.

Maybe I am rather dull and ignorant, but it seems a little unbelievable.

Let Q be the unit cube in Rⁿ and let m ≤ n. Given ε > 0, there is a number N_ε where for all 1-Lipschitz functions f mapping Q into R^m, there are balls {B_i : i = 1 to N} with N < N_ε, where Hⁿ_∞(f(Q \ U_i B_i)) < ε and f|B_i is N_ε-biLipschitz!

Essentially, this means that off a small set set of small image, 1-Lipschitz functions from the unit cube to a higher-dimensional Euclidean space are locally bi-Lipschitz (and probably with very large biLipschitz constant). It's not even clear to me that a 1-Lipschitz function should be locally invertible, much less locally biLipschitz!

Edit: perhaps thinking of $ard's Lemm@ gives plenty of motivation. Now that I think about it, I'm not as impressed. Since n ≥ m, then the Hausdorff content given by Hⁿ_∞ doesn't detect the image that well, since in some ways, the image is an "m-dimensional set." I'd be much more impressed if the theorem used H^m_∞ instead.

Anyways, back to work.