from the Lang.NEXT conference this week. I heard most of this stuff not from the talks, but from the awesome people who were attending the conference.
- Alloy – A language for relational models. You enter a specification of your problem in a first-order relational language (quite powerful) and then it tries to find counterexamples within some small space. If no counterexamples are found, you can have a fairly high degree of confidence in your model (as they say, “most assertions are wrong; most flaws have small counterexamples”).
- Combinatorial Sketching. “Here’s what I want my algorithm to do, and I know that it probably involves a for loop and addition and multiplication. Figure it out for me, kthxbye.” (Thanks to Peter Alvaro for the last two links; check out his research on Bloom, too: a very nice way to simplify and analyze distributed systems)
- Cyc, a queryable database of “common sense”.
- Labanotation – a standardized system for recording human motion. Traditionally used to write down choreography.
- Structure and Interpretation of Classical Mechanics, in which Sussman (creator of Scheme) writes about classical mechanics in a functional style. I have read the first chapter, and I have to say, functional calculus notation makes a lot more sense to me than the traditional notation. Go figure.
- Algorithmic information theory
- Smooth interpretation — a way of smoothing out digital programs so that their parameter spaces can be searched by gradient descent.
- Lytro, the coolest piece of hardware I have seen in years: it’s a camera which records the directions of incoming rays, so that you can focus the picture after the fact. For example.