Tuesday, September 24, 2024
Context: In my role as division director of Information and Intelligent Systems (IIS) at NSF, I’m sending out a short message to the IIS mailing list on the Second Tuesday Every Month (STEM). This is the installment for September 2024, closer to being on time than last month’s.
As an organization, NSF is an interesting mix of churn and stability. Since roughly half of the program staff are “rotators” (IPAs, like me---see STEM Update #1) with a four-year time limit, at least an eighth of the staff is replaced every year. (Indeed, the division is not at full strength at the moment, so make sure to keep an eye out for job postings!) But permanent staff, once they join, often stick around. After all, it’s a great place to work and NSF doesn’t really have a lot of head-to-head competitors in the science-funding space. So, when a permanent staffer finishes at NSF, it’s kind of a big deal. Today, I wanted to acknowledge Ephraim Glinert (eh-FRY-eem GLIN-ert).
Ephraim was born in Gotham City Hospital, which I’m pretty sure is also where Batman got his start. He began his academic career in math, but, when computers came along, they turned his head. He ended up getting a PhD in computer science and his research focused on assistive technologies. He was hired by NSF originally to work on a precursor to AI called Knowledge and Cognitive Systems.
At NSF, he focused on programming languages, assistive technology, and computer graphics. (In his words, “Graphics proposals are math in disguise”.) He was hired by Juris Hartmanis, Turing Award winner and head of CISE from 1996 to 1998, at least in part because Hartmanis liked Ephraim’s Jerry Garcia tie. Ephraim first came as a rotator, then came back as a permanent staff member shortly after. (I feel like it’s worth pointing out that although most permanent staff were originally rotators, most rotators do not become permanent. Being an IPA is a stepping stone, not a one-way door, in case you were concerned.)
In his time at NSF, Ephraim saw a lot of changes. Computer graphics went through a phase where it seemed that Hollywood ought to be the place funding it. AI went through a phase where it seemed that no one wanted to fund it. He shared with me that he believes the most significant advance in assistive technology has been the smartphone, which puts powerful sensing and computation into everyone’s pocket.
Ephraim commuted to NSF from upstate New York, but he’s been remote, for health reasons, the two years I’ve been at NSF. I did get to meet him once in person because I served on a review panel that he ran. I remember being super impressed by his mastery of the material being discussed and especially the procedural elements of the review process. He tells me that the hardest part of being an NSF program officer is having to choose projects to fund when several can be funded and there simply isn’t the money to support them all. But he’s always tried to be as fair as possible and was constantly on the lookout for opportunities to support those whose strong proposals didn’t make the cut for some reason.
He served as the program officer on 1820 (!) projects, with start dates ranging from 1988 to 2024. His peak year was 2009, which is the start year for 122 projects he oversaw. The projects he was responsible for amounted to $652,034,865. (To put that into perspective for readers of a certain age, Ephraim funded more than 100 Steve Austins.) The projects cover 47 states (including DC and PR as states), only missing WV, VT, SD, ND, and AK. By my count, he has funded 68 ACM Fellows (including at least one Turing Award winner), but also 210 CAREER awards that helped kickstart researchers' professional trajectories. Indeed, he funded 4 CAREER projects of people who went on to be named as ACM Fellows! Another of the 1000+ PIs he supported was none other than Sethuraman Panchanathan, currently NSF’s director.
Ephraim expressed that it’s the right time to retire from NSF for him personally, but also because he is confident that the areas he’s been nurturing for decades are in good hands---the division has made sure to hire top people to handle the topics in Ephraim’s portfolio. His engagement in the HCC cluster will be sorely missed by his colleagues at NSF and by the many many researchers he has helped support over the years. We are very grateful and proud of all he accomplished.
Last month, I shared a puzzle inspired by real-life events. (I can even show you the picture of the skewed water bottle, if you are interested!) The puzzle went like this:
I have an empty, sealed, cubical one-gallon water jug from Wegman’s (the closest supermarket to NSF HQ). Its sides are rigid, but the top and bottom are quite flexible, allowing the jug to skew. The container (filled with air) was at room temperature when I put it in the fridge, which is 37 degrees (Fahrenheit). After the container chilled completely, I took it out and observed that it collapsed a bit and became a rhomboid prism (rhombus on the top and bottom, square sides). Amazingly, one of the internal angles of the rhombus, in degrees, exactly matched the room temperature, in degrees. How warm is NSF (to within a tenth of a degree)?
Many thanks to the folks who sent their solutions. Putting together some of the best elements of what I’ve seen, here’s my version.
There’s a few facts from a few different “classes” you can use. From chemistry class, PV=nRT (aka, the ideal gas law) lets you relate volume and temperature. We can rewrite the equation as V/T = nR/P, and we have two scenarios (room and fridge) with different temperatures and volumes but constant n, R, P. So, we can set the ratios equal in the warm (w) room and cold (c) fridge cases:
Vw/Tw = Vc/Tc .
We can assume that the cube has all unit sides, so Vw = 1. We also know the temperature in the fridge, which is Tc = 37F = 496.67 Rankine. (We need to use Rankine or Kelvin, something absolute-zero based, to apply the ideal gas law. That’s an easy place to slip up.)
If x is the Fahrenheit temperature and angle in degrees we are looking for, then Tw = x + 459.67.
How about Vc? Now another formula from another class---geometry---comes in handy. The volume of a rhomboidal prism with all unit sides is the same as the area of the rhombus at the top and bottom. The area of a rhombus can be written in a few ways, but we want to relate it to x, so we can use: side^2 × sin(x). The side is length 1, so Vc = sin(x). Caveat: A lot of trigonometric functions expect radians, so one might have to use Vc = sin(pi x/180) to make sure x is interpreted as degrees. Using degrees when you should have used radians makes you guilty of original “sin”.
Putting these pieces together, we have:
1/(x + 459.67) = sin(pi * x / 180)/496.67 .
Ok, yuck. But a little binary search gets us that the two sides of the equation are almost equal at x = 69.74341115858, so 69.7 is the answer I was looking for. Kind of remarkably (I didn’t know it would turn out that way when I started developing the puzzle), that is pretty close to the temperature in the building!
Until next time.
-Michael