The relative activity on this vs the ATM thread would seem to support the hypothesis that math graduate school is less interesting than lesbian sex.
Give it time, AL. In a hundred years, maybe this thread will be important to the burgeoning field of artificial goldfish habitats.
Sometimes I like to imagine what it's like to have math graduate school.
What is it like to be a bisected angle?
Honestly, that sounds really fun. Or maybe I'm just really bored.
You don't really read about vacuum cleaners, Mentioner.
Conceptual space abhors a vacuum.
I need to invent a new mathematical formula to indicate how many ways that entire post is banned.
The denouement of that is depressingly accurate. Also, the whole vacuum analogy reminds me of http://xkcd.com/1133/.
What's interesting is that this does nicely illustrate the differences between two fields where I can actually read technical papers. In biology papers the process is pretty well laid out to the extent that I think even less trained people can understand what's going on. In chemistry papers you really need to already know the underlying reactions of what's happening- they throw around so-called "named reactions" quite a lot with the assumption that you know that a Sonogashira coupling (kinky!) means that this plus that became this other thing through this mechanism. If you don't know that, trying to follow the illustrations doesn't make sense. In biology authors spell out much more explicitly what's going on.
I just started reading a paper whose title begins "A gentle tutorial" and whose first sentence is "Recall the definition of the maximum-likelihood estimation problem."
Ok! I can totally recall that.
I working with MLE processes right now.
Great! Can I give my present problem to you?
You probably know way more about it than I do.
That's a pretty good description of graduate school in theoretical physics, and I think my students would say it's also good for the theoretical reaches of statistics and computer science.
Hey, Cosma, you could probably do this thing I'm supposed to do in your sleep. Just sayin'!
15: Does it involve regression?
It's true that I discount my consulting rates if the client will accept solutions that come to me in dreams.
(Oh, just e-mail me, it's got to be better than writing tomorrow's problem set.)
That's a pretty good description of graduate school in theoretical physics
Has almost nothing in common with my experience, as far as I can tell.
Let me reiterate, essear, that we're not talking about literal vacuum cleaners.
Because it doesn't mention airports and hotel wifi?
Still. I bet I could find some string theorists who say it sounds familiar.
23: God, my comments must be boring. I'm currently using wifi on a plane. For about five more minutes before my battery dies.
Dealing with an unknown, as yet undescribed object with complex properties is a pretty common experience in science. Usually, clarifying how the thing works and when you can see it requires effort from people working with different tools. Problems compact enough to fit into one person's head are unusual. Reading about Barbara McClintock's work to identify transposons before sequencing existed is mindbending. Her face should be on the twenty.
This seems like an OK place to drop a link to this nice overview of a topic that I find pretty interesting, random matrices.
I have emailed you, Cosma.
Math researchers ( users ?) take more words to describe pictures than CS folk.
Complete with regressions, plots, etc:
The other day I did have a student saying, basically, "My senior seminar paper will be all wrapped up if I can just figure out a formula for Bell Numbers", (except she didn't call them Bell numbers, obviously) . I admired her enthusiasm, but she was really not hearing what I was saying, when I told her that that was not going to happen by Friday.
26.last: I don't know as much about random matrices as I should. Does anyone really understand why they work so well for nuclear physics? Or for the Riemann zeta zeros?
I thought that the connection was: the interactions for a linearized many-particle Hamiltonian can be approximated by a random matrix with suitable mean and variance.
Well, yeah, but: why? Is there some sense in which almost all matrices have spectra that look like random matrix spectra? Is this quantified? (Probably, I suppose.) But there are lots of other problems where the answers don't look like random matrices.
I guess I should read more-- I'm moderately familiar with random matrix ensembles but not with how you would know they apply to a given problem.