I just got an e-mail from the APS about the March Meeting that included the following line: "All March Meeting registrants will receive a CDRom version of the Bulletin, plus a pocket epitome."
I had forgotten about that meaning of the word. Under normal circumstances, people compare the sizes of their pocket epitomes...
1 day ago
4 comments:
On a completely unrelated note, I just wanted to say:
There... are... four... bumps!
Hmmm... maybe best not to get my started on that seminar. I'm sure the work is fine (I'm not qualified to judge it), but the overselling seriously got on my nerves, and Oscar was dead-on correct with his point (and the speaker's attempt to totally bullshit an answer seemed freaking obvious to me). There was some stuff that wasn't quite right about how quantum chaos was mentioned (I did my thesis on that topic...), but that wasn't the main point (or even close to the main point) of the talk, so I let it slide publically.
I had trouble quelling my snickering with the comment about Mexican cars being parked farther apart...
I noticed. I too was highly amused, right from the point when Dan Meiron made the comment about a possible "correlation between ethnicity and beta?"
I also agree about Oscar's comment. I've seen too many talks at this point that present some complex, extended research all built around computing some parameter they've made up, but they have no idea what the parameter really means or could be used for.
Yeah, the _math_ is probably fine, but she would have been much better served to talk about some other application. Also, the \beta \neq 2 from that application was totally hoaky (sp?). It was like 1.96 or something, which if you're fitting to noisy empirical data is essentially indistinguishable from 2. The simplest theories (deriving some value of \beta even if it's from a toy ["spherical cow"] model) are far more likely to give something like beta = 2 than something that one can distinguish statistically from that value. And one would think that somebody trained predominantly in statistics would know this!
Now, the mathematics she's doing is probably good, but she is clearly interested in that and not the applications (as far as I can tell). I think she showed a pretty fundamental lack of understanding in the latter respect.
By the way, I am almost positive I have seen a beta not from {1, 2, 4} show up in physics (from some theory, not from just fitting empirical data). I saw a fitting with this two weeks ago from a person at University of Maryland who is working on certain things. Also, now that I think of it, I know of various theories that predict distributions that interpolate between the distributions that arise from different beta values. (These aren't theorems in the mathematical sense. They are conjectures, but they come from some analysis and I know a class of example equations and boundary conditions to solve to get data to test them [there are some serious open issues here, actually]. In fact, one of my SURF students is going to be doing something very much along these lines this summer, assuming his project gets accepted.)
The only example the speaker needed was the Helmholtz equation with appropriate boundary conditions (which can also be done in the lab). Then take really high-energu solutions (high quantum numbers to go into the "deep semiclassical regime"; the random matrix theory results in quantum chaos are obtained in the semiclassical limit [the 'large' part of the large chaotic systems is dead wrong, by the way, for this application], so one gets the best set of data by looking at the most excited state that your experiments/numerics allow.
OK, I got more technical than I wanted. I wouldn't hire this person (based on the talk---I know there's a metric fuck ton of information I don't have, so I'm conditioning on the given information).
Post a Comment