Thursday, September 22, 2011

Superluminal neutrinos - a case study in how good science is done

As many people have now heard, the OPERA collaboration is reporting a very surprising observation.  The OPERA experiment is part of CERN, and is an experiment meant to study neutrino flavor oscillations.  The idea is, the proton beam at CERN creates a beam of neutrinos.  Since neutrinos hardly interact with normal matter, they move in a straight line right through the earth, and pass through the experimental station in Gran Sasso, Italy, where some small fraction of them are then detected.  There are (according to the Standard Model) three flavors of neutrinos, the electron neutrino, muon neutrino, and tau neutrino.  It has been determined experimentally that those flavors are not exact "mass eigenstates".  That means that if you start off with a tau neutrino of particular energy, for example, and let it propagate for a while, it will change into a muon neutrino with some probability that oscillates in time.  Anyway, OPERA wanted to study this phenomenon, and in doing so, they measured the time it takes neutrinos to go from their production point at CERN to the detector in Gran Sasso, using precisely synchronized special clocks.  They also used differential GPS to measure the distance between the production point and the detector to within 20 cm.  Dividing the distance by the time, they found much to their surprise that the neutrinos appear to traverse the distance about 60 ns faster than would be expected if they traveled at the speed of light in vacuum.

So, what could be going on here?  There are a few possibilities.  First, they could have the distance measurement wrong.  This seems unlikely, given the use of differential GPS and the sensitivity (they could clearly see the change in the distance due to a 2009 earthquake, as shown in Fig. 7 of the paper).  Second, they could have a problem in their synchronization of the clocks.  That seems more likely to me, given that the procedure is comparatively complicated.  Third, there is some other weird systematic at work that they haven't found.  Fourth, neutrinos are actually tachyons.  That would be all kinds of awesome, but given how challenging it would be to reconcile that with special relativity and causality, I'm not holding my breath.

Why is this an example of good science?  The collaboration spent three years looking hard at their data, analyzing it many different ways, checking and cross-checking.  They are keenly aware that a claim of FTL neutrinos would be the very definition of "extraordinary" in the scientific sense, and would therefore require extraordinary evidence.  Unable to find the (highly likely) flaw in their analysis and data, they are showing everything publicly, and asking for more investigation.  I want to point out, this is the diametric opposite of what happens in what I will term bad science (ahem.  Italian ecat guys, I'm looking at you.).   This is how real experimental science works - they're asking for independent reproduction or complementary investigation.  I hope science journalists emphasize this aspect of the story, rather than massively sensationalizing it or portraying the scientists as fools if and when a flaw is found.

Thursday, September 15, 2011

State of Texas threatens physics departments at smaller public universities

This article is both sad and frustrating.  The coordinating body of the Texas state government that runs the public universities in this state has recommended that a number of places shut down their physics departments.   In particular, this affects two schools near Rice that are historically African American serving, Prairie View A and M and Texas Southern.   (Unfortunately, the article doesn't have a link to the actual Texas Higher Education Coordinating Board recommendations, so I don't have any further information, like which other universities here may be affected.)   

Depressingly updated:  see NY Times story here.

I understand that financial times are tight for the state.  (Look at the "Texas Miracle" in action as we slash the state's education budget.)  The bit that really galls me is the rationale:  enrollment in the upper division courses is small, so we should eliminate the whole department.  This idea that somehow the only valuable and cost effective courses are those with large enrollment is ridiculous, and it seems to have infected the public university system in this state, driven by misguided, bean-counting thinktank types.  If you follow this reasoning all the way, we should only have large service courses, and never have upper division, specialized courses in anything, and of course all of these should be taught by non-tenure-track, non-research-active instructors.  That would surely cut costs.  It would also be a disaster in the long run. As is stated in this article, if you used the same criteria in terms of size of upper division courses across the country, you'd end up shutting down 2/3 of the physics departments in the US, to say nothing of other disciplines.  I can't imagine the situation is any better in, e.g., math, or chemical engineering, or any technical discipline.  I'd also love to see numbers about how much collegiate athletics is net costing the state in public funds, vs. how much it costs to keep these programs going.  Hint: most universities lose money on athletics.

I'd love to try to fix this, but given the politics here (hint:  Rick Perry likes these policies, and his political party controls both houses of the state legislature), it's hard to see a workable path forward.  It's not like this is going to be an honest debate about how to structure the state's higher education system (which we can and should have) - it's an ideological full-court press.   

Think I'm exaggerating?  The superintendent of the THECB, Raymund Paredes, is a close buddy of both Rick Perry and his pal Rick O'Connell, the guy who thinks that a bachelor's degree even in a technical field should be obtainable for $10000 total, period.  You could do that, of course, but it would involve converting our colleges and universities essentially into community colleges or correspondence schools.  I've yet to see any evidence that these guys have an appreciation for science or engineering at all.  They want UT and TAMU to play good football, and they espouse populist rhetoric about wanting to cut costs, but they don't seem to want academic excellence at universities.

Wednesday, September 14, 2011

Lab habits + data management

The reason I had been looking for that Sydney Harris cartoon is that I was putting together a guest lecture for our university's "Responsible Conduct of Research" course. I was speaking today about data management and retention, a topic I've come to know well over the last year through some university service work working on policies in that area. After speaking, it occurred to me that it's not a bad idea to summarize important points on this for the benefit of student readers of this blog.  In brief:
  • Everything is data.  Not just raw numbers or images, but also the final analyzed graphs, the software used to do the analysis, the descriptions of the instrument settings used to acquire the raw numbers - everything.
  • The data are the science.  The data are the foundation for all the analysis, model-building, papers, arguments, further refinements, patents, etc.  Protect the data!
  • If you didn't document it, you didn't do it.
  • Write down everything.  Fill up notebooks.  Annotate liberally, including false starts, what you were thinking when you set up the little sub-experiments or trials that go into any major research endeavor.  I guarantee, you will never, ever in your life look back and say, "I regret that I was so thorough, and I wish I had written down less."  After years of observation, I am convinced that good notebook skills genuinely reduce mean time to thesis completion in many cases.  If you actually keep track of what you've been doing, and really write down your logic, you are less likely to go down blind alleys or have to repeat mistakes.
  • You may think that you own your data.  You don't, technically.  In an academic setting, the university has legal title to the data (that gives them the legal authority that they need to adjudicate disputes about access to data, including those that arise in the rare but unfortunate cases of research misconduct), while investigators are shepherds or custodians of the data.  Both have their own responsibilities and rights.  Some of those responsibilities are inherent in good science and engineering (e.g., the duty to do your best to make sure that the published results are accurate and correct, as much as possible), and others are imposed externally (e.g., federal funding agencies require preservation of data for some number of years beyond the end of an award).
  • Back everything up.  In multiple ways.  With the advent of scanners, digital cameras, cheap external hard drives, laptops, thumbdrives, "the cloud" (as long as it's better than this), etc., there is absolutely no excuse for not properly backing up data.  To repeat, back everything up.  No, seriously.  Have a backup copy at an off-site location, as a sensible precaution against disaster (fire, hurricane, earthquake, zombie apocalypse).
  • Good habits are habits, and must be habituated.  It took me more than 25 years to get in the habit of really flossing.  Do yourself a favor, and get in the habit of properly caring for your data.  Please.

Monday, September 12, 2011

Help finding a Syndey Harris cartoon

I am trying to find a particular Syndey Harris physics cartoon, and google has let me down. The one I'm picturing has an obvious experimentalist at a workbench strewn with lab equipment. There's an angel on one shoulder, and a devil on the other. Anyone who has this cartoon, I'd be very grateful for a link to a scanned version! Thanks.

Wednesday, September 07, 2011

Single-molecule electric motor

As a nano person, I feel like I'm practically obligated to comment on this paper, which has gotten a good deal of media attention. In this experiment, the authors have anchored a single small molecule down to a single-crystal copper surface, in such a way that the molecule can pivot about the single anchoring atom, rotating in the plane of the copper surface. Because of the surface atom arrangement and its interactions with the molecule, the molecule has six energetically equivalent ways that it can be oriented on the metal surface. It's experimentally impressive that the authors came up with a way to track the rotation of the molecule one discrete hop between orientations at a time. This is only do-able when the temperature is sufficiently low that thermally driven orientational diffusion is suppressed. When a current of electrons is properly directed at the molecule, the electrons can dump enough energy into the molecule (inelastically) to kick the molecule around rotationally. In that sense, this is an electric motor. (Of course, while the rotor is a single small molecule, the metal substrate and scanning tunneling microscope tip are macroscopic in size.) The requirements for this particular scheme to work include cryogenic temperatures, ultrahigh vacuum, and ultraclean surfaces. In that sense, talk in the press release about how this will be useful for pushing things around and so forth in, e.g., medical devices is a bit ridiculous. Still a nice experiment, though.  I continue to find the whole problem of nanoscale systems driven out of thermal equilibrium (e.g., by the flow of "hot" electrons) to be fascinating - how is a steady state established, where does the energy go, where does irreversibility come into play, etc.

Friday, September 02, 2011

Playing with interfaces for optical fun and profit

A team at Harvard has published in Science a fun and interesting result.  When light passes from one medium to another, there are boundary conditions that have to be obeyed by the electromagnetic field (that is, light still has to obey Maxwell's equations, even when there's a discontinuity in the dielectric function somewhere).  Because of those boundary conditions, we end up with the familiar rules of reflection and refraction.  Going up a level in sophistication and worrying about multiple interfaces, we are used to having to keep track of the phase of the electromagnetic waves and how those phases are affected by the interfaces.  In fact, we have gotten good at manipulating those phases, to produce gadgets like antireflection coatings and dielectric mirrors (and on a more sophisticated level, photonic band gap materials).  What the Harvard team does is use plasmonic metal structures to pattern phase effects at a single interface.  The result is that they can engineer some bizarre reflection and refraction properties when they properly stack the deck in terms of phases.  Very cute.  I must confess, though, that since Federico Capasso was once my boss's boss at Bell Labs, I'm more than a little disturbed by the photo accompanying the physorg article.