This morning, as a sociological adventure and at a friend’s suggestion, I attended the Harvard Conservative Minyan’s Rosh Hashanah morning service. It was certainly very different. The Reform services I’m used to are done largely in English, dominated by responsive readings in which the Rabbi or speaker says one line, and the congregants respond with another. The parts of the service that are conducted in Hebrew are generally accompanied by a translation, also spoken aloud. The Hebrew itself is almost always sung or chanted, at a pace chosen to ensure that everyone can sing along to the music, even if they don’t speak the language. The cantor, who leads the singing, often has a guitar, and the net effect is not so far from a folk music festival. Prayers are generally presented in the shortest available form, since the audience doesn’t understand the text anyway.

In contrast, the Conservative service, at least as performed by these Harvard students, was dominated by Hebrew. There were a few words in English, and of course the sermon itself, but the great majority of the service was rapid-fire Hebrew readings. The readings were Torah, Haftarah, extended, repeated prayers, and everything else. They were a mix of spoken, chanted, and sung, often switching modes for a tiny fraction of a reading, without any delineation in the text, and then switching back. The Hebrew, especially toward the end, was chanted so fast that the pitches could only be hinted at, and read so fast that I could barely track the speaker in the book. These readings were performed solely by the leader, since no group of people could possibly read aloud coherently at that speed. Some of the songs were melodies I knew, and some were not. It was a tremendously confusing experience. Also, the service was over 4 hours long, but the etiquette of Conservative services permits worshipers to come and go at any time.

I have difficulty summarizing the difference between the services. The closest I can come is: Reform optimizes for the common case being a congregation that cannot understand spoken Hebrew, and attempts to satisfy those who do by retaining the key prayers and readings. Conservative optimizes for the common case of a congregation that understands Hebrew, or at least wants to listen to it spoken, and attempts to satisfy those who are not skilled in Hebrew by minimizing any requirement for audience participation in Hebrew. It’s much more complex than that, though, as there is are also differences in the participant/observer balance, the amount of repetition, and a number of other aspects.

Anyway, after the service we wandered around Harvard for a bit, bought some unexpectedly fantastic traditional Rosh Hashanah apples at an unexpected farmers’ market, and then went to the Tashlich service. Tashlich is a simple ceremony in which Jews throw the lint from their pockets into the river, to symbolize throwing away all of the sins and bad memories that have accrued over the previous year. Meta-symbolically, we sometimes throw bread crumbs in lieu of lint, and this year the Harvard Hillel offered rose petals as a more ecologically aware alternative to bread, though I don’t think rose petals are a good symbol of dirty things we’d like to be rid of. This Tashlich was incredibly anarchic, and not very much like the inter-congregational ceremony run by all of the synagogues in Westport.

In the evening I went out to dinner with my grandparents and cousins at Locke-Ober. Locke-Ober is the second-oldest restaurant in Boston, unless that’s Jacob Wirth. It’s a dark, beautiful place that seems a perfect setting for a meeting of corrupt politicians and gilded age titans of industry, and indeed that’s pretty much what it was (the main room was open to men only until the 1970s or later).

Rosh Hashana

It’s Rosh Hashanah (ראש השנה , for all you Unicode fans). That means it’s the Jewish New Year, although confusing, it’s actually the start of the seventh month in the Hebrew calendar, so it’s the same numerical Hebrew year (5769) as it was yesterday. Anyway, that’s how the Bible says to do it, so that’s how we do it.

I went to the Harvard Reform services, which were ok. The cantor’s microphone was flaky, and I didn’t particularly like the sermon. I generally don’t like sermons that turn into popularized neuropsych lectures. Maybe they work better for people less closely connected to this stuff, but when a Rabbi starts talking about right- and left-brain behaviors I lose a lot of interest.

I was at services with some friends, one of whom shrewdly observed that since we were all dressed up fancy, we should go out to eat somewhere nicer than usual. We settled on Dali, a tapas bar near Harvard (it’s right next to Evoo). It was actually much less expensive than I expected, and definitely a good investment in suit-time. Then we went back to his apartment for ice cream (it’s traditional to eat something sweet) and wound up discussing Fourier optics with his landlord, who had come over to debug a running toilet.

It was nice.

As the traditional benedictions say, may you be inscribed for blessing in the Book of Life this year, and may we forgive all of our transgressions against each other.

Test party

Last night my roommates and I had a test party. We wanted to see (1) if people would actually come and (2) whether we had enough space in our current layout to make it work.

It was definitely a success. I invited a few friends who weren’t already wiped out from the Biophysics retreat, and made a batch of chocolate covered peanut-butter balls. We had three confirmed guests, but eventually a total of 16 people showed up, proving that my roommates do, in fact, have friends.

One of the people I invited informed of another party, being held by another friend at his house, about 300 feet away, so halfway through our party, I stepped out for a few minutes to run through the rainstorm, knock on the wrong door, find the right house, drop off the snacks, say hello, and walk back to my own party… but not before my friend invited me, and all of my guests, over to join up.

Moving 16 people through a storm is not really possible, but once our party was winding down, I led the last few over to my friend’s much nicer apartment (he’s a real working professional, so he has every right to a much nicer apartment). It was a good mix of people, so we stayed pretty late.

This morning I woke up around 11:55 and rinsed out all the beer bottles for recycling before I realized that I was missing Dim Sum with the same crowd. I got there towards the end of brunch, which is just as well, since I’d forgotten my wallet at home. I still got a chance to see some friends from high school with whom I haven’t spoken in years. I met them just after they started dating; they are now married with a wonderful baby boy. It was a good time.


I’m back from the retreat. It was pretty much the same as the last two years, with lots of interesting talks by students and faculty, a poster session, lots of food, and a crazy party.

The annual party is the creation of Michele, the department administrator, whose creative impulse knows no bounds. This year, we covered every surface in the presentation hall with fabric and props from a theater supply company. Half of the room was “Olympus”, decorated with ivy-laced Doric columns, cardboard “statues”, with the walls and ceilings covered by gossamer cloud-pattern cloth. The other half was Tartarus or the Labyrinth, with a maze drawn on the floor in tape, the walls covered in a dark stone pattern, and the entrance guarded by a Minotaur (played by an excellently costumed professor of Neuroscience).

The bulk of the party was a team competition, with events such as “pin the fig leaf on the Adonis” (a cardboard “statue”, not a volunteer), pass the olive, and an Ancient Greece trivia quiz. I wrote a bunch of questions for the quiz about Greek science, medicine, engineering, astronomy, and mathematics, though due to a miscommunication many were much too hard.

Michele had asked me last week to sing “Venus (if you will)” by Frankie Avalon, and this somehow also involved me wearing a rather strange costume of a Spartan Warrior. I practiced a bit, and flubbed one stanza in the performance, but it was alright. Michele also refused to take back the costume, since she has nowhere to store it.

I guess I know what I’m wearing for Halloween.

Retreat, retreat!

The next two days are the Biophysics department retreat, which is about 75% scientific conference and 25% party. I don’t know if I’ll have an internet connection, so I may be hard to reach.

Today I built something. It’s a motor, mounted to pull sinusoidally on a string that runs through a tube. As a device, it’s not very interesting, but some people, and my mother in particular, will be shocked that I actually built something out of wood and screws and such.


I am searching for a lubricant. Specifically, I need a lubricant that will work well with vinyl.

As you might imagine, Google returns mostly the wrong sort of answers for when I search for lubricants and vinyl.

Anyway, if you know of a standard lubricant that works well between nylon and soft vinyl, and won’t dissolve either, please let me know.

EDIT: Oh man, Google is creepy. Now this page is coming up in my searches for lubricants.

Lunch and $

I was working with another student on a project today. We started work around 11 AM. By 3 PM I was really hungry, since we still hadn’t broken for lunch, but my coworker seemed undeterred. I told him that I was too hungry to continue, and asked if he wanted to join me. He explained that he wasn’t eating lunch, because it is Ramadan.

In other news… I don’t understand this finance thing at all. Specifically, the current problem appears to be essentially that banks have loaned out a lot of money that they are not going to get back, because the people they loaned it to (homeowners) don’t have enough money to pay their mortgages. They did not expect so many people to be unable to pay. The solutions being proposed seem to take the form of a government-funded entity that will buy all these mortgages in pure liquid US$, at a sufficiently high price that the banks are able to continue to function. What will then happen is that when homeowners fail to pay their mortgages, and are evicted from their houses, the loss will pass to the government. The government may hope that this will happen sufficiently infrequently that they will ultimately make more money from the mortgage payments than they lose in foreclosures, but this is by no means certain, and most analysts seem to think that is unlikely. Therefore the plan will cost many billions of dollars.

I don’t understand why anyone thinks this is a good plan. To me, the obvious solution is to provide subsidies to homeowners who are deemed sufficiently close to the edge of foreclosure. This determination would look very similar to FAFSA, the formula used by the federal government to determine how much families can afford to pay for college tuition. Just as FAFSA is used to determine a student’s eligibility for educational financial aid, this system would be used to determine a homeowner’s eligibility for homeownership financial aid. The aid could also take a similar form, such as ultra-low-interest loans backed by the Fed.

Foreclosure is tremendously toxic. Forcing poor people to move out of houses they cannot afford leaves them with even greater expenses of moving out, finding new housing, commuting longer to work, and being divided from the social fabric that provides them with valuable but inexpensive services like babysitting by the neighbors or sharing meals with relatives. The remaining empty houses are a haven for crime, and lend the street the depressing air of a ghost town, which lowers the value of other properties in the neighborhood. This has been known to cause the neighbors, unable to sell their houses, to fall into foreclosure as well, creating a vicious positive feedback loop that evicts families from their homes and leaves lenders with huge losses.

By stabilizing homeowners directly, the Schwartz plan (yes! vote for Schwartz!) also saves the banks. The banks knew these subprime mortgages would result in many defaults, and factored this into their calculations from the beginning. Unfortunately, they seriously underestimated the true rate of default. By assisting homeowners at risk for default, the Federal government would also be reducing dramatically the risk of default on those loans. Banks would find the expected value of their mortgages rise back to the levels that they had originally anticipated, and likely higher, as defaults become rare. This effect would naturally propagate through the rest of the derivatives markets that are based on these loans.

The Paulson plan does nothing to stop homeowners from falling into default and bankruptcy. Since each foreclosure increases the risk of foreclosure in other properties, the Paulson approach will result in wave after wave of failed mortgages, and all of these losses will pile up on the government. By contrast, the Schwartz plan would stem the tide of foreclosures. In fact, by stopping the feedback loop, the Schwartz plan would also result in higher government tax revenues and lower expenses related to Welfare, Medicare, and other expensive programs onto which foreclosed families often fall. In other words, it’s less expensive too.

I can hardly imagine a more obvious proposal, especially for the Democrats, who love to hand out Your Tax Dollars to poor people in need. They can even attack the Paulson plan for handing Billions of your Hard Earned Money to Foreign banks, which is true. Instead, the only plan I’ve seen from the Democrats is Christopher Dodd’s proposal. That proposal has some mention of this concept, under the acronym of HOPE, but still appears to spend most of the $700 billion buying securities. It also contains a passage that would require companies to issue new shares to the government along with troubled assets, resulting in (1) crashing the stock market and (2) a fascist financial structure in which the government owns the corporations. It’s probably still better than Paulson’s plan, but it doesn’t have anything close to the feel-good socialism of the Schwartz plan.

Anyway, there’s probably some good reason why the smart people who know about these things are saying what they’re saying, but I totally don’t get it.

LHC Quench

My timing is impeccable. It appears that yesterday, scant hours after I wrote a post about explosive quenches of superconducting electromagnets, one of the ring magnets at the LHC quenched.

Interestingly, this magnet is of the same field strength, and therefore probably of similar construction, as the MRI machine I mentioned at the Martinos Center. However, the magnets at the LHC are much longer and narrower than an MRI machine, as the former needs to contain a miles-long beam of protons, while the latter needs to fit human patient comfortably.


Dear Blue State Obama Fans,

Many of us who live in states like Massachusetts whose electoral outcome is not in question feel that our vote does not count. We often do not bother to go to the polls on election day, because we are already certain of the outcome.

This election day, please vote anyway.

According to the statistical analysis here, there is a 3% chance of an Electoral College tie. The analysis is admittedly approximate, but this figure is probably near the true probability. This chance is much higher than the probability that Massachusetts will vote for John McCain.

In the event of an Electoral College tie, the election is decided by the House of Representatives. The only neutral criterion they have for deciding a tied election is the popular vote. Everybody’s vote counts equally for the popular vote, regardless of where you live.


I mentioned that my first proposed cause for the still-unexplained steam was an uncontrolled quench. An uncontrolled quench is what happens when an MRI machine fails. The sequence is as follows:

First, the liquid helium coolant begins to leak. In fact, liquid helium is very difficult to contain, and reliquefying helium that has evaporated into gas takes very specialized equipment and is generally not done on site, so the level of liquid helium is always decreasing slowly, and must occasionally be refilled.

Second, once the liquid helium level falls below a critical amount, a patch of the main electromagnet may cease to be superconducting. An MRI machine uses a strong magnetic field, generated by a superconducting electromagnet in which currents can run in circles for many years without requiring any power input. If a patch of the electromagnet warms to more than a few degrees above absolute zero, the current density it is capable of superconducting may fall below the amount of current running through it. At this point it begins to act like a normal metal wire, dissipating some small fraction of the energy running through it and generating heat.

The heat generated by this patch has two immediate effects. The heat spreads to other nearby areas of the electromagnet, and also begins to warm the surrounding liquid helium. As the surrounding areas get warmer, they too may lose superconductivity, and thus start heating the liquid helium further.

Liquid helium boils easily, and as it is heated, it begins to boil and vent out of the dewar containing the electromagnet. The falling level of liquid Helium reduces the effectiveness of the cooling system, precipitating a positive feedback loop in which the warming of the magnet allows further warming, and ever faster boiling of the Helium.

In a typical MRI setup, Helium is designed to be vented out of the roof of the building. However, there is always the possibility that this venting system might fail, and the Helium might boil into the MRI suite. This creates the principle danger of a quench. Although Helium is harmless to breathe, a full-size MRI often has enough Helium in its cryostat to completely displace the air in the room, and the vicious cycle of a quench means that this can happen extremely quickly. Thus, every MRI suite has an Oxygen alarm, which constantly monitors the level of Oxygen in the air and sounds an audible alarm if it begins to fall.

A second danger of a quench is to the magnet itself. A tremendous amount of energy is stored in the magnetic field of an MRI machine. The 7 Tesla machine at the Martinos Center has a stored energy comparable to several sticks of dynamite. During a quench, this energy is released into the Helium, and then into the windings of the magnet. The temperature of the magnet may rise from 4 degrees above absolute zero to several hundred degrees in a few seconds or minutes. The resultant rapid thermal expansion can damage the magnet irreparably, and a quench can potentially raise the temperature high enough to melt some components. (EDIT: Note that this sort of melting is not seen in medical MRI scenarios. It might be possible in research magnets, like those in particle accelerators, though.)

Finally, a third danger involves the amount of Helium leaked. MRI rooms are typically tightly sealed against electromagnetic radiation. The walls, ceiling, and floors are lined with sheetmetal (beneath the wallpaper and tile), and the door is typically metal as well, with a copper contact spring all around the edge to form a perfect Faraday cage when closed. There is usually a window between the control room and the magnet room, with multiple layers of metal screens and glass. The goal is to ensure that radio waves cannot enter or leave the room during a scan, as this would potentially corrupt the images and disrupt any sensitive nearby medical equipment. Because MRI machines are so loud, a secondary goal is to avoid disrupting any nearby humans.

Because the rooms are so tightly sealed, it is possible that in the event of a quench, the room would experience an overpressure, because the air in the room cannot escape faster than the helium is entering. A 1% overpressure (0.14 PSI) would not be noticeable to the patient, but would require over 500 pounds of force when she attempted to open a typical inward-opening door, leaving her trapped in the room, possibly with a diminishing supply of Oxygen. Therefore, when possible, MRI suite doors open outwards, and have hydraulic opening devices to ensure that they can be opened in any eventuality.

I am telling you about quenches mostly so I can show you a few pictures I took at the 7-T magnet at the Harvard/MIT Martinos Center, where I was working most recently. This MRI machine is one of the most advanced in the world, with an unusually strong field. The magnet consumes Helium much more rapidly than most magnets due to its nonstandard construction (it was built as an experiment, not as a commercial product). The first is the remote monitoring system for liquid Helium levels:
LHe Webcam 1 LHe Webcam 2
It consists of a webcam pointed at the liquid helium gauge, duct-taped into place and lit by a USB LED flex-light attached to a nearby computer.

The second is the pressure relief system in case of overpressure. Note that the door to the MRI room opens inwards, and does not have a functioning hydraulic assist.
7T Door and Bat 7T Bat
The text on this Louisville Slugger standard wooden baseball bat reads “IN CASE OF QUENCH USE BAT TO BREAK WINDOW AND RELEASE PRESSURE IN MAGNET ROOM”. Note that because the window has two thick panes, separated by a number of metal screens, there are two such bats, one inside the room, and one outside, in the control room.

I should perhaps emphasize that I do not mean to imply that safety standards are not applied properly at the Martinos Center. In fact, I perceive this setup to be quite safe. The researchers are vigilant about maintaining proper Helium levels, and general safety education. Uncontrolled quenches are extraordinarily rare in medical MRI scanners, so rare that there are no statistics on their frequency, and they are regarded as almost hypothetical. The 7T scanner room is properly ventilated, so there is no reason to expect an overpressure even if a quench were to occur.

I just think that the setup is hilarious.