Another hard push at reducing my reading list. At the current pace I may not write many more of these posts.
This week I have kept the pace of reading, and I have also finished a generative sketch I’m particularly happy: Iris. You can read about how it was done here.
NOTE: Generative code, ZIO, Scala, Python, Spark, Maths. Expect a similar wide range in the future as well. You can check all my weekly readings by checking the tag here. You can also get these as a weekly newsletter by subscribing here.
Although I’m technically an applied mathematician (dynamical systems and functional analysis being my “main”), for my research masters I split my courses between applied maths and algebraic geometry. I read parts of SGA and EGA for one of the courses (and had to give a talk about Grothendieck and Serre’s vanishing theorems). Grothendieck’s life (and mathematical output) are mesmerizing.
(edited title) Very interesting, I got to learn about the cube type in Postgres (ah, Redshift, will you ever move from Postgres 8?). I faced a similar problem (but more like the other side) on extracting the palette of an image It is quite a hard problem (and using RGB distance is not the best approach).
By Cal Newport, which for me always adds a mental “but” because I have the nagging feeling he usually talks from his navel-gazing perspective. I’ve been a shoemaker, so there’s that for manual competence ¯\﹍(ツ)﹍/¯
Interesting to have as an option. We don’t use Terraform (but cloudformation directly with boto), although for the initial Databricks deployment all the IAM roles required were configured manually (it is not worth automating what may never be used or modified again!). It wasn’t that hard to be fair, although I have some experience with IAM.
Written by a true APL expert (he has written a compiler for a subset of APL, to run on GPUs). I really dig APL, and even was close to moving to a job on it a few years ago. I still think it is the answer to quick data exploration, but other considerations (for instance, check my Mapping as a Tool for Thought slides) make it unpractical for me.
I’ve been doing some high intensity (order of millions) web scraping+NLP lately, so I have delved into Spacy (gensim and Stanza as well). For now I have not needed to extend NER, but having that as a possibility is tempting.
I have a hard time wrapping my head around GPU processes, vertexes and shaders. Luckily there are people that grok that, and this is a very thorough and detailed explanation of how to simulate Game of Life using texture renderers and framebuffers on a GPU.