Category — Physics

I decided, back in January, that this was the year I’d transition my tcl scripts for re-formatting the APS March Meeting program into Python. I’m putting my python program here, but I don’t know if it will run on your system without tweaking. It needs two text files as input which I’ve traditionally cut-and-pasted from the APS March Meeting website: the Epitome, and the list of invited speakers (but see the discussion below).

march2tex.py

I admit what I do here is sort of quaint, in this age of cloud-based apps and so forth: a script to take two text files and generate three LaTeX files which, when run, make 3 pdf files that you can print. But I do it in part because the APS Meetings app is still garbage: it’s a generic scheduling app that doesn’t take into account the particular structure of the APS March Meeting and isn’t optimized for its main task: helping attendees decide which of more than 60 places they ought to be at any one point. But a rant about the app can come later.

Here are the pdf outputs of my scripts. These use the Epitome and Invited Speaker lists as of a week ago, February 15th.

Mar18_Invitedpy

Mar18_Epitomepy

Mar18_Gridpy

And if you like the idea of these but want to tweak the LaTeX formatting, here are the LaTeX sources for these:

Mar18_Epitomepy.tex

Mar18_Invitedpy.tex

Mar18_Gridpy.tex

Every year, there are changes to the format and structure of the meeting that force tweaks in the code, but this year was one of the most dramatic changes. The titles of invited talks used to use LaTeX commands to show mathematical symbols, Greek letters, sub- and super-scripts, and so forth. This year, there is only one invited talk that shows the $-delimited LaTeX commands in the invited talk list (L35.1), but if you view the abstract, the subscripts that these commands produce are rendered.

Many of the other talks with chemical formulas in their titles just write the titles out without subscripts, which looks ugly. So I found a regular expression that matched chemical formulas (and which needs the python regex package), and a LaTeX package that typeset them (mhchem), to make them look decent. But the regex isn’t good enough to capture something like “Fe_1-x“. And then the regex also can’t tell that PHYS21 isn’t phosphorus-hydrogen-yttrium-henicosa-sulphate. So this required an unfortunate manual tweaking of the input file.

All of the Greek letters, and all the accent marks on speaker names, which used to be displayed with LaTeX commands, are now rendered. This good for readability, of course, but a disaster for a script that’s designed to take plain text and format everything with LaTeX. Parsing all the rendered formatting and unicode would add a whole extra level of programming difficulty.

I’m glad to have switched to python, though. My tcl script was merged together from what had been separate scripts to generate the different files, and I found a number of redundant arrays and unused variables, so the new program is much cleaner. Perhaps for next year, it will be able to scrape epitome and invited speaker list directly from the web. Or even better, perhaps APS will release a machine-readable scientific program. But then I might have to try to develop my own app.

I’ve been back from the 2009 APS March Meeting for two weeks now and so the window of relevance for writing about it is rapidly closing. It was held in Pittsburgh this year, following the same format as last year. The meeting seems to be getting bigger each year: when I first attended, in 2003, there were about 5600 attendees; this year’s meeting drew 7000.

Sessions
For a number of years now I’ve taken the online Epitome and Invited Speaker List and run them through tcl scripts to make TeX files that give me speaker and session information in a format I think is more useful. This also allows me to look at overall meeting statistics: There were more sessions this year, 558, than in previous years; last year there were 517 sessions. What seems to be growing most sharply are invited talks: there were 825 this year, compared to about 730 in each of the previous three years. Not surprisingly, this corresponded to an increase in the number of sessions with 5 invited talks: there were 95 such sessions this year, about 75 in each of the previous 3 years, and only 15 back in 2005.

Reunion
I only stayed through Wednesday of the meeting this year, taking an evening flight back home. I was rather irritated to find that the Cornell alumni reunion was held on Wednesday night, instead of Tuesday, like it always had been. I don’t know if this had even been published before I made my travel reservations, although I don’t know that I would have stayed an extra night just for that.

Projection
The disappearance of viewgraphs now appears complete. I was one of a handful of holdouts who was still using viewgraphs as late as 2006. Last year, I only saw one talk given using viewgraphs and this year I saw zero. There are still overhead projectors in the rooms, but they are kept on the floor beside the table upon which the computer projector sits. It’s amusing to read the note in the 2002 newsletter of the Division of Condensed Matter Physics:

More and more scientists want computer projection for their talks. This past year, computer projectors were available in invited session rooms only. Projectors are very expensive (~$400/ day/session) and would raise the registration fee at the conference significantly if placed in all rooms. Also, set-up time between talks makes staying on a 12 minute schedule for contributed talks very problematic. APS will continue to increase the availability of computer projection, but will not commit totally to them until price and technical interfacing problems become more tractable.

To be sure, there are problematic computers and I did see talks where roughly half of the time was taken up with computer fiddling. 

Context
On the topic of presentations, one thing that lots of speakers do, which really bugs me, is to show a graph of some raw data, usually a spectrum of something taken with a well-established experimental technique, but without giving any explanation. If I don’t use a technique myself, even if I know in principle how it works, I don’t know if it’s considered good or unexpected or interesting or disappointing if your graph has wiggles, or is flat, or has a bump in a particular place, or a big spike, or a big dip, or if it shifts a little as you twiddle some parameter, or shifts a lot. Context, my fellow physicists! Tell us what your measurement technique does, what shows up in your graph, what ordinary data would look like, and why your particular measurement is interesting.

Books
I also ended my one-year physics-book-buying drought. I buy interesting physics books knowing that I’m not also buying the time it takes to work through them. I have one book purchase from two years ago that I’ve made a concerted effort to actually work through, but am perhaps only 20% done with it. And it’s not even a very challenging book. But I went ahead this year anyway, and took advantage of Cambridge University Press’s Wednesday afternoon buy-2-get-50%-off sale to pick up an otherwise ridiculously overpriced Elasticity with Mathematica and Geometric Algebra for Physicists, and also bought Group Theory: Applications to the Physics of Condensed Matter.

On to Portland
I’m looking forward to visiting Portland for next year’s March Meeting. I consider Portland one of my favorite cities but in reality all I’ve only spend several hours there at a time, waiting to change trains. But with a streetcar and Powell’s, who couldn’t love Portland? I had been sure that, a couple of years ago, I also saw Seattle on the list of upcoming March Meeting locations, but it seems to be gone now.

By trade, I am a physicist; specifically a condensed-matter experimentalist. More so than other scientists, physicists are rather sharply split into two categories: theorists and experimentalists. Einstein and Feynman were theorists, working with equations and making calculations. Experimentalists work in labs and conduct experiments and analyze data.

A day spent taking data, then, is usually a good day: data is, after all, the lifeblood of experimental physics, and if you’re taking data then that means your apparatus is working, which is always a good thing.

I spent today taking data. Even though the process is heavily automated and computer controlled, the apparatus needs constant supervision. As is the case with most of the experiments I run, there is a very short duty cycle for attention: I’ll need to make an adjustment, then let the experiment run for a few minutes, then make another adjustment, and wait a few minutes, and so forth. And although the taking of data is an inherently good and fulfilling process, the time scale that my experiments work on can be frustrating.

The intervals between which my attention is needed are long enough so that, after a while, it’s sort of boring just to sit there and wait. On the other hand, the intervals are too short to get anything meaningful done. I often think that I should be able to read journal articles in these time spans, and I usually have an article or two with me that I intend to read, but this always fails: reading journal articles must be done in large, uninterrupted time blocks. If I do try to read a journal article, and get in depth reading it, then I’d be likely to let the experiment languish, and in the end that’d be a terribly inefficient way to take data.

As it turns out, surfing the web is the sort of activity that can be broken into minute-long chunks, and since the data taking computer is usually also hooked up to the internet, it actually works out quite well to browse the web during the gaps when the computer is doing all the other work.

Most of the data I take measure some sort of quantity as a function of temperature, and typically the experiment cools to it’s lowest temperature, and then the temperature is raised in steps, with a datum taken at each temperature. A large part of the waiting around I do is waiting for the temperature to stabilize. In many experiments, the speed with which the temperature can be changed and stabilized sets the rate at which I can take data. Inasmuch as I’m browsing the web while I’m waiting for the temperature to stabilize, being lazy and being careful can sometimes look very much like one another. To be careful, you really want to make sure the temperature is stable, that it isn’t drifting or oscillating, and generally, this means waiting longer. But waiting longer also means more time browsing the web and less time doing real physics, so if you’re waiting longer than it takes for the temperature to become as stable as your experiment needs, then you’re being lazy.

This year, I only went to the first three days of the March Meeting; extended trips are less feasible now with a baby at home. The trade show and posters are only Monday–Wednesday, and a fair number of attendees–particularly college faculty with classes to teach–stay for only a few days. So Thursday and Friday the conference is sort of a ghost town, but one thing I’ve heard from quite a few people this year is that it looks like several interesting sessions are scheduled for Thursday and Friday.

To read up on some of the physics presented at the March Meeting, a physics blogger’s account starts here. Physics journalists are blogging from the meeting here. Another attendee’s take is here. (I’ll update periodically if I find new blogged accounts.)
Photographing Talks
One phenomenon that’s becoming more common, and which is at least slightly disturbing, is for audience members to use digital cameras to photograph all the slides that a speaker presents. It’s obnoxious when the camera makes faux-shutter sounds, and it’s really obnoxious when the flash fires. When the flash is used, it’s also a sign that the photographer is an idiot, because the flash will make the image of the projected slide come out worse: physicists should be able to figure this out. If the photographs are unobtrusive, I haven’t quite figured out what the ethics of the situation are. I’d think if you really wanted someone’s viewgraphs, or data, you should just send them an email and ask: I’d share my data and graphs if someone asked for them.

Viewgraphs
I saw 39 talks this March Meeting, and only one used viewgraphs, and that was after the speaker attempted to use the computer but had some difficulty getting the computer to cooperate with the projector. The APS recommends that speakers have viewgraphs as backups; most don’t, but this speaker did, and the talk otherwise went off without a hitch. One out of 39 is 2.6%, which is a smaller fraction than last year (but not significantly so).

Microphones
One piece of technology that number of speakers decided to do without was the (supplied) wireless microphone, which is a real shame. Perhaps the folks up front can hear the speaker fine, but there’s always a background rustling of papers and backpacks and schedule books, and people are talking outside the room, and if the doors are open this really filters in, and if the doors are closed then they keep opening and closing as people straggle in. It’s really hard to hear an un-miked speaker in the back of the room.

Seating
Because of the 39 parallel sessions, in which you wish you could be two or three places at once, people do lots of session-to-session shuffling. It’s always awkward to squeeze in to the open seats mid-session. The solution to this: more aisle seats! Instead of having two columns of seats, with an aisle down the middle and sometimes aisles down the sides, the APS should have the convention center arrange the seats to be at most 4 across, in 3–5 columns, so that everyone who wants an aisle seat–which is everyone who’s shuffling between sessions–could get one. It would cut down on the capacity a little, but although there are some overflowing sessions, there are plenty that are sparsely attended. To this end, the APS should try a little harder to predict the attendance at the sessions, and put the popular ones in the large rooms. I know this is hard, but isn’t this the sort of problem that physicists should be able to tackle?

New Orleans Convention Center
The March Meeting isn’t like a trade show, where you need lots of floor space, preferably contiguous, for all the vendors to set up their booths. Rather, we need lots and lots of meeting rooms. Convention centers, as a rule, have both, but some do the meeting rooms better than others. The shorter the distances between rooms, the better, because at the March Meeting it is common to try to shuffle between different rooms during a session. In New Orleans, there were two giant rooms on the first floor, and the rest were on the second floor. Most of these were along one long linear corridor, but a handful of rooms were in a parallel corridor on the other side of the center. To get between these corridors you went through an elevated walkway that overlooked the trade show floor. The convention center is along the Mississippi river. Between the center and the river are active railroad tracks. In the rooms on the river side, one could clearly hear the horns of the trains that used this track. I suppose this isn’t a problem inside a noisy trade show floor, but when you’re trying to listen to a talk, it makes you wonder why the architect didn’t specify more robust soundproofing on the walls that face the river.

The trade show:

And the posters:

The PAR 124 lives!
Perhaps the most exciting development, from my perspective, was the unveiling of the Signal Recovery 7124 Lock-In amplifier, so new it doesn’t appear on their website. Lock-in amplifiers are one of the most useful pieces of equipment in experimental physics research: They perform frequency selective signal detection and amplification, showing the amplitude and phase that appears on a signal in relation to a reference signal. They are tremendously valuable for processing noisy signals.

Years ago, a company called Princeton Applied Research (PAR) produced a truly amazing lock-in, the PAR 124. Built with all analog electronics, and mostly from discrete components, it was very sensitive, very quiet, and had been a staple of low-temperature research labs for decades. Contemporary lock-ins are, as a rule, digital, making use of Digital Signal Processing (DSP) to process the input signals, and do more with them than ever possible with analog electronics. But the digitization process itself is noisy, sending a sort of switching noise back up the input line to the experiment. This added noise and energy is very much unwanted in low-temperature experiments, so low-temperature labs kept PAR 124s around and sought them out from second-hand equipment dealers.

(I’ve used them in many of the mechanical oscillator projects I’ve worked on, but for a slightly different reason. Even when you use an external signal for the reference channel of the 124, the internal oscillator locks to your external reference and then this internal oscillator is used for signal processing. Because of the way this locking is accomplished, the 124 can be used to implement a phase-locked loop, by driving the mechanical oscillator with the reference signal output.)

Princeton Applied Research was bought by EG&G, who was then bought by Perkin-Elmer, then became Ametek, and who finally decided to call themselves Signal Recovery. And whenever they showed up at a trade show, such as at the March Meeting, low temperature physicists (including myself) would always ask if the PAR 124 would ever come back. “We’re working on it,” was always the answer.

And so they were. This March Meeting, they had on display the 7124 lock-in. It uses an all-analog front end, to which you connect your experimental signal. It’s connected by a 5-meter fiber optic cable to a digital DSP lock-in, so you can take advantage of all the advanced features of a DSP lock-in without the digitizing noise getting back to your sample. If I had a spare $15000 sitting around, I might buy one.

I am now in New Orleans, for the 2008 American Physical Society March Meeting. This is a huge conference, lasting 5 days and attended by perhaps 7000 physicists. Most attendees give some sort of presentation, the bulk of which are contributed 10-minute talks; this is the first year that I’m attending without giving a presentation. This meeting only covers what is known as condensed matter physics–the folks doing particle, nuclear, and atomic physics have their own meetings.

Each day, there are three large time blocks of talks, beginning at 8am, 11:15am, and 2:30pm (although Friday only has the first two blocks). Blocks are assigned letters: A, B, and D for Monday, and continuing alphabetically until X and Y for Friday. Poster Sessions and special evening sessions also get their own letters; year-to-year, the assignment of the letters varies. Within each block, individual sessions are numbered, with numbers corresponding to the rooms used. If a room is not used one block, the number will be skipped: this year there are sessions A33 and A35 but no A34; there are, however, sessions P34 (on Wednesday) and V34 (Thursday), both in room 226 of the Convention Center.

Invited talks last 30 minutes, with 6 minutes for questions, contributed talks last 10 minutes with 2 minutes for questions. Talks within each session, of course, are grouped around a common topic. Each block runs for a full three hours and, for an entirely contributed session, can have as many as 15 talks. If you stayed for a full session and then went to the beginning of the next session, you’d get all of a 15 minute break between them.

To accommodate several thousand talks during the 14 3-hour blocks, the conference has, this year, as many as 39 parallel sessions running; last year there were as many as 43 parallel sessions. All total, this year, there are 518 sessions of talks. For each block, sessions 1 through 7 have 4 or 5 invited talks and no contributed talks; other sessions will have zero to two invited talks and as many as 15 contributed talks. All together, there are 729 invited talks this year in the regular sessions, and 5193 contributed talks.

Of course, some popular topics have more than 15 talks, and in this case, there can be multiple sessions with the same title, usually in the same room, with roman numerals appended. This year, in what must be some sort of record for such numbered sessions, there are 16 sessions on “Carbon Nanotubes and Related Materials.” Session 29 of every single block is devoted to this, and blocks B and D double up with Session 30. In these 16 sessions, there are 15 invited talks and 195 contributed talks.

The posters are treated as an afterthought. There’s no time set aside to look at posters; poster authors are instructed to be present during the third time block for each of the three poster sessions, which are, as a result, very lightly attended. Nevertheless, there are 1032 posters to be presented. Some physics bloggers have written about the poster-versus-talk question, and although people in other scientific fields can have a useful discussion about the relative merits of each, the culture of the March Meeting, which highly values the talk, seems unlikely to change.

Since the APS lets anyone who wishes to give a talk do so, any prestige associated from giving a talk is purely illusory (unless it’s necessary to convince the bureaucrats at one’s institution to let you go to the meeting in the first place). The prestige comes only from being sorted into a prestigious session, and getting a large audience. But, in the case with really popular topics, such as the aforementioned “Carbon Nanotubes and Related Materials” series, you still can’t see all of it, and even if you maximized your attendance, at 197 talks, you wouldn’t be able to see talks on any other topic, nor visit the trade show. And I’d imagine that a good number of those 197 talks wouldn’t actually be worth watching.

As far as I know, few other scientific organizations with conferences this big let anyone who wants to give a talk: more commonly, the only talks are invited presentations and the contributed sessions are posters, and there is a set time and some incentive (such as free beer) to attend the posters. I wouldn’t mind seeing the APS adopt this system. Certainly, one advantage of giving a talk is that it doesn’t have to be finished before you leave for the airport. But with 5299 talks, and the possibility to see at most 210 contributed or 70 invited talks, means there’s going to be stuff scheduled at the same time that you want to see. I’ve even known a session chair to be scheduled to give a talk in a different session as the one s/he is chairing! In the nanotubes case, it would probably be easier to scan through the 183 contributed papers as posters, looking at the titles and figures, and decide on the few you’d like to read in-depth, than to make it through 197 talks.

Posters help at the other end of the popularity scale as well, those people working in less popular fields. The attendance at talks can be quite varied, especially for those doing less trendy physics. I’ve been in sessions where the audience is about 6 people, usually four people waiting for their turn to talk and the other two labmates of the speaker. If there are only a handful of people interested in one particular topic, a poster session is more conducive to mutual discussion of each presentation than a handful of 10-minute talks in a sparsely attended session that’s only loosely bound together around a broad theme. And you can certainly fit more substance into a poster than into a 10-minute talk. I know there are several people (my graduate advisor, for example) who think that the March Meeting is so large, and that 10 minute talks are so insubstantial, that the March Meeting is not worth bothering with at all.

But for now, talks it is. Even for 10 minute talks, these days the overwhelming majority are given with laptops and Powerpoint. In 2005, it seemed, there was an equal balance between laptops and viewgraphs. In 2006, I definitely felt in the minority as a viewgraph-user, and I think for my session I was the first to use the overhead projector. Last year, I finally broke down and used a laptop (although apparently I had a problem with the display resolution). I kept tally: last year I saw 20 invited talks, 58 contributed talks, and only 3 used viewgraphs. Of course, with such a packed schedule there really can be no tolerance for slipping from the schedule, so those whose laptops refuse to behave are essentially out of luck. In fact, last year I did see a presenter use his entire 12 alloted minutes trying to re-boot his laptop and get the external display to work. (There is an AV ready room, in which presenters are encouraged to make sure their settings are correct.)

The posters are held in the same space as the trade show, where scientific equipment vendors have booths. The vendors (and their wares) tend to be the same from year to year. Perhaps the most interesting vendors are the scientific publishers, who have for sale all the new physics books of the year. I have a bookbuying habit, and usually pick up three or four new books at the March Meeting, but so far I’ve yet to actually read a book I’ve bought. The vendors hand out disappointingly little swag. And there is no conference swag: all you get when you register is a badge and the schedule book.

One soon realizes that the additional activities: receptions, special sessions, and the like, are also concurrently scheduled, and tend to stay at the same time slots from year to year. The only event I regularly attend is the alumni reunion for my graduate program, which is always held on Tuesday evening.

My strategy, for choosing what to see, is first to look at sessions in my research area. Then, I look for talks given by people I know personally, and also talks by people who I know to give interesting talks. Finally, I look through the lists of invited talks to see which have interesting titles. I might look over the abstracts for the talks or sessions that look interesting, but it’s hard to look through many abstracts beforehand.

To accommodate the large number of presentations, contributed abstracts are restricted to 1300 characters, about a paragraph. Until a few years ago, these were printed and bound and distributed to all conference attendees as two volumes, known as the Bulletin of the American Physical Society (BAPS), each of which resembled a phone book. To cut down on the bulk, they tried in recent years to simply give everyone CD-ROMs while having terminals with an electronic version of the BAPS are set up throughout the conference site. Now, we just get a bound volume with abstract titles; it resembles the phone book of a small city.

Of course now one can also download the program contents beforehand, and in order to facilitate my own planning, I wrote a few tcl scripts to parse the list of sessions and the list of invited talks and to make more readable forms of these. (If they didn’t require so much hand-tweaking each year, I’d post them–anyone who wants them, though, I’d be happy to share.) It’s in part from these scripts that I was able to get the numbers I quote here.

So what’s in it for me? Perhaps the most important thing I get out of the March Meeting is a chance to see what’s happening in a broad swath of contemporary condensed matter physics. The talks are as hit and miss as talks anywhere, and even if some of the same information could be gleaned by reading journals, I simply don’t have the time to read many journal articles outside my immediate field of interest. Of course, meetings are also good chances to catch up with grad school friends who are now dispersed at institutions across the country. But no matter what I end up doing, I find that the March Meeting always rejuvenates my enthusiasm for physics, and provides a motivational boost to continue my research back in the lab.