Thursday, August 20, 2009

Random Recoiling

I recently came across several papers/articles or blog entries that make me recoil in some fashion or another, here is a batch:
  • NSF Workshop: Electronic Design Automation-Past, Present, and Future. Discussions include some blog entries including mine.
  • I wonder if this Panalogy Principle is a good way of thinking about manifold signal processing.
  • It looks like the gaussian nature of the brownian motion is not verified after all.
    "...In both sets of experiments, there were many features in full agreement with Einstein and the bell-shaped curve; but there were also features in significant disagreement. In those cases, the beads moved much farther than the common curve could predict. In those extreme displacements, diffusion behavior was not Gaussian, the researchers report. The behavior was exponential.These large displacements happen less often, but when they do occur, they are much bigger than we previously thought possible," Granick said..."
  • Here is a document written by some danish mathematicians on how much Mathematics there is in space related projects. If they had waited a few more months, they would have put compressive sensing as an encoding mechanism, but then again they needed to talk to the folks at ESA not JPL. I was a little underwhelmed by the problems they found that math was solving. The whole process of integration and verification (I&V) is a mind buggling process for instance as one needs to make sure that the models developed for each parts behave somewhat nicely when connected to each other. It is not as interesting as developing trajectories but it's a more difficult problem in the end. I would not be surprised to find out that one of the reason most space projects go over budget has to do, in part, to this mismatch.
  • In the world of nuclear weaponry, here is an interesting inverse problem revolving around the following question: How do you establish trust by letting an adverserial party look at your nuclear weapon without opening it up ? The process should allow the querying party to be using an instrumentation that allows for some sort of inverse problem (is the warheard there?) while at the same time, the inverse problem should not be good enough to give too much information about how the weapon is assembled. This is all the more important as the querying party include countries that do not have nuclear weapons. The issue is that you do not want to be proliferating by allowing inspection. I recall a story that was told to us at the Nuclear engineering seminar back at Texas A&M about a visit by the U.S in the (then) U.S.S.R on this verification business. Back then, before 1990, the US team was asked to verify if a series of missiles had nuclear warheads in them. Actually, the issue was whether the missiles under inspection had more than one warhead as the same missile could have one or more warheads in them. Of course, the U.S team could not come within a pre-specific number of yards from the missiles. The soviet military officer in charge asked the U.S team to pick one of the missile at random and do their measurement at once. Recall we were in the middle of the Cold war, so the whole thing was pretty serious and ceremonious, as this exchange was probably the first one since the beginning of the Cold war. But as soon as the US team decided on the missile to be remotely inspect, one of the red guards jumped and smile to the surprise of the visiting U.S team. After the test had been performed, the person in charge of the U.S delegation asked why the Red Guards had this unexpected behavior. The Head of the soviet team responded that all the guards had a betting pool and that this one soldier had won. Sometimes, you think the game theory is about one thing when in fact it really is about another one.

No comments: