Last week, instead of studying 3 extra hours for my midterms, I watched the following four fascinating videos. Three are Nobel Prize lectures in economics, and the fourth from a lecture series at MIT.

Lloyd Shapley: Assignment Games - The Mathematics of Matching

Lloyd Shapley, back in 1962, wrote the seminal paper: “College Admissions And The Stability of Marriage.” He developed the deferred acceptance algorithm and proved its optimality for creating pairs of people among a large set when all have ranked preferences.1

Alvin Roth: The Theory And Practice Of Market Design

Roth and Shapley were jointly awarded the most recent Nobel Prize in Economics (2012). Roth has worked extensively with Shapley, and has put a lot of Shapley’s work into practice eg. the deferred acceptance algorithm is now used for student placement in the New York City and Boston Public School systems. He has also done a lot of work on optimizing the kidney market using the top trading cycle algorithm. His work is particularly interesting (I think) because as large tech companies become the center of massive networks of people and begin to engineer marketplaces and facilitate transactions on those networks, market design - if it isn’t already - will become an incredibly important research area.

Daniel Kahneman: Maps of Bounded Rationality

Kahneman is a very well known behavioral economist, who, across countless research endeavors, has proven that we aren’t exactly the perfectly rational creatures we so often assume when modeling reality. He won the Nobel Prize in 2002, and in this lecture he walks through several of his experiments and conclusions in a very approachable fashion. His book - Thinking Fast and Slow - is on my bookshelf and I hope to read it soon.2

Michael Sipser: Beyond Computation - The P vs NP Problem

Michael Sipser is a professor of Applied Mathematics at MIT. He is also a good family friend, and has known me since I first came to the US 9 years ago. Instead of studying for my Comp Sci midterm, I decided to learn more about the P vs NP question and serendipitously found this lecture he gave. He explains the problem and its importance in a very approachable way and I highly recommend it.

###Asides:

  1. I apologize that all the Nobel Prize videos automatically play.
  2. In case you are interested in how I embedded the Nobel Prize videos given that their site doesn’t generate an embed code for you, I used the following structure:

<iframe src="link-to-video-page.html#div-containing-video" scrolling="no"><iframe>

The # at the end of the link to the page allows you to select a single CSS element on the page that you want to embed.


  1. If you are interested, you can read the paper here, or google for it. It is only 6 pages and requires no math, and if you read it, you can say you understood a paper that won a Nobel Prize!

  2. If you enjoy his lecture, or are interested in more of his research, you might like to check out some of his more recent research on happiness, experience, and memory that he discusses in his 2010 TED Talk: The Riddle Of Experience vs Memory.

This semester I am taking a course on data analysis in R. I had never used R before, so I thought I should give myself a fun little project that I could use to get familiar with the program. I decided to analyze data on the current Yale undergraduate students (classes ‘13, ‘14, ‘15, ‘16). Below are various descriptive graphs I created of the undergrad population. If you are interested in the code I wrote you can find it at the end of the post (suggestions for improvement encouraged!)

Distribution of Birthdays Among Yale Students

The Distribution of Birthdays Among Yale Students

Distribution of Students Across the Residential Colleges

The Distribution of Yale Students Among the Residential Colleges

Distribution of Students Across Majors

Students by Major

The Most Popular Majors

The Least Popular Majors

Computer Science Majors by College

Comp Sci Majors by College

Econ Majors by College

Econ Majors by College

Engineering Majors by College

Engineering Majors by College

English Majors by College

English Majors by College

Physics Majors by College

Physics Majors by College

The Code is here:

I have developed a fascination with time lapse movies (but who wouldn’t after watching this or this). What I enjoy most about these movies is how they enable us to see what is otherwise hidden to our senses. As humans I think we are well suited to perceive discrete actions during a moment in time, but poor at seeing patterns over time. Time lapse helps us uncover some of these patterns. Whether it is the movements of a city, the turning of the earth, or the movements of boats with the current, time lapses enable us see changes over time that are otherwise invisible to us. In offering us this view into reality at 2-3x speed, we are able to see our environment and movements in a totally new way.

For similar reasons, I am excited about the rapid development of sensors that enable the quantified self movement. If this data collection becomes powerful enough, I think I will be able to uncover patterns in my behavior that I am otherwise largely unable to perceive (eg. if I drink coffee after 4pm I struggle to fall asleep, or if I exercise I have a productivity boost the next day …). Who knows, but I do think all these discrete moments will compile into fascinating patterns that hide insights that are otherwise invisible.

##Making My First Timelapse:

To create my first time lapse I needed a way to send a signal to my camera every x seconds telling it to take another picture. A device that does this is called an intervalometer and can be plugged into the side of the camera (I use a Canon D60).

You can buy such a device commercially from Canon for $100+. This seemed insane, so I decided to first apply some DIY resourcefullness. I found that people had programmed TI calculators to send the signal, and I used this instructable tutorial to do exactly that. But, alas, I couldn’t get it to work with my Canon D60 (I tried with a TI-83, 84 and various versions of the program). I then decided to splurge on this $35 intervalometer rather than the Canon branded one, and it works great.

The tedious part about time lapse photography is that it requires waiting in one place with your camera for hours. If the end film is set to 30 frames per second, then you need 30 pictures for each second of film. So, if you want a 30 sec video segment you need 900 photos. If you set the intervalometer to take a picture every 5 seconds then you get 12 photos per minute. So, to get the 30 second video segment you need to be in the same place for 75 minutes. Given the cold, I didn’t want to sit outside for 75 minutes to shoot my first test time lapse. So, I set up my camera to take photos of the TD courtyard through my window. This meant I could leave my camera unattended for the day and it would be safe.

When taking time lapse, one important thing to remember is to set the camera to aperture priority (AV). This allows you to specify a fixed aperture for the duration of the shoot, and the camera will set the shutter speed to ensure consistent exposure. This is especially important if the light will change during the shoot (e.g. sunset or sunrise). Also, you want to set the camera to manual focus, not automatic. On automatic the camera will refocus each shot and you won’t get a smooth sequence of identically focused images in the final movie. The viewer will pick up on the micro adjustments to focus that the camera made.

To compile the images at 30fps I used a really basic free software called Time Lapse Assembler. It takes all the JPGs in a folder and puts them in sequence.

My First Time Lapse:

The result is my first time-lapse of the TD courtyard. The first 2.35 minutes are of a sunny day, and the rest is of a snowy day.

As an FYI, the camera will often only count up to a certain integer number of photos e.g. 9999. When it reaches the limit the next photo will be recorded as 0001.jpg. When imported to a folder these shots will then be out of sequence and the time lapse assembler will not assemble them correctly (it will start with 0001.jpg but that is actually a shot from the middle of the sequence). You will need to rename the files so they fall after 9999.jpg. I did this by first moving the files to be renamed into their own folder and then running the following command in terminal before moving them back into the same folder:

for FILENAME in *; do mv $FILENAME Unix_$FILENAME; done

My Second Time Lapse:

After the successful first attempt, I took some basic time lapse shots to make a fun behind-the-scenes promo video of the TEDxYale team setting up balloons for a PR display on cross campus.

Next Steps:

My next step is to learn how to do batch editing with photoshop, so that I can make adjustments to one image of a time lapse sequence and have it applied to every other image.

Some of my Favorite Time Lapses:

  1. Earth
  2. Yosemite HD
  3. The City Limits
  4. Manhattan in Motion
  5. The City of Boston