Uncertain<T>
66 points by cgc373
66 points by cgc373
I love this article, but I love the name of the type even more. If you say it outloud it’s “Uncertainty.”
Someone else mentioned Kalman filters. I gave a talk on them and how they’re related to baes rule and the general case of a low information state system https://www.youtube.com/watch?v=bQSzZrDDV80. It’s not my finest delivery, but the information is still pretty solid imho.
Surprisingly, the paper doesn’t cite the 2007 recipe, Build your own probability monads. Less surprisingly, it also doesn’t cite the general 2016 recipe for sensor fusion, Sheaves are the canonical datastructure for sensor integration, or even the well-known special case of Kálmán filtering, which makes the choice of GPS somewhat confusing as an example.
This is really cool! If think it would be nice to extend this to support Kalman filters, especially for measurements.
Interesting to see Kathryn McKinley as one of the coauthors on the paper, I mostly know her for her work on GC algorithms.
Related reading that PL-minded folks might find interesting:
https://www.cs.cornell.edu/courses/cs4110/2016fa/lectures/lecture33.html