2006-04-01

code bugs, LRG clustering, cosmic rays

The NYU Computer Science Colloquium was by Xie (Stanford) on automatic detection of bugs in code, through direct boolean modeling of the source code. He noted some remarkable statistics, eg: that it is believed that most commercial code has of order 10 bugs per thousand lines of code; that bugs cost society on the order of 0.6 percent of the GDP; that GM vehicles are expected to have 100 million lines of code by 2010 (they already have a million). Xie showed that security-critical bugs (eg, for buffer overflow) can be found automatically with few false positives, and he is responsible for finding hundreds of bugs in open-source software, such as openSSH, mySQL, and the Linux kernel.

Kazin and I discussed the possibility of explaining Majedi's correlation function by looking at the separation distribution of the two most massive subhalos inside halos massive enough to contain two LRGs. Wechsler's results (discussed here earlier) suggest that this project will work well.

Sigl (Paris 7) gave a nice overview of the current and future of very high-energy cosmic rays, which is bright, given the start of the Auger Observatory in Argentina. Auger is almost certain to see anisotropies at very high energies, because the GZK cut-off means that they come from nearby, and there just aren't that many possible sources, even if they come from gamma-ray bursts!

No comments:

Post a Comment