Hooray, my coursework is over! I still need to defend my thesis, but no pressure because I also got a great job offer! The offer is with LSI (lsi.com) in Boulder as a firmware engineer writing embedded C++ that performs various types of RAID on large hard disk arrays. I'm still working out the details of the contract, but I'm really excited about the position. Also, Marissa got a job offer the same day! Now we just need to find a place to live between East Denver and North Boulder... I've accepted the longer commute under the stipulation of unlimited allowance for books on tape.
For my research, I'm currently trying to use the notion of trust to compute a version of the pearson correlation coefficient that incorporates transitive correlation. It is well known that correlations are not transitive - if i is correlated with j and k, j is not necessarily correlated with k. I believe that if i and k are strongly correlated, then k's correlation with j can be an estimator for i's correlation with j. This relies on sufficiently large sample sizes (and intersection sizes), so we'll see how it pans out. This computation is a nasty cubic problem because we have to find a k for all pairs of i and j, which works out to about 17770*(17770*17769/2) or 2,805,484,330,050 iterations. Using four school machines I was able to do the computation overnight.