Skip to content

Latest commit

 

History

History
150 lines (123 loc) · 3.96 KB

outline.md

File metadata and controls

150 lines (123 loc) · 3.96 KB
Raw

Top-level

  • The goal for the first session is to
    1. familiarize people with Julia's basics
    2. demonstrate common Julia tools (VSCode, terminal, basic computer stuff)
    3. showcase a typical Julia workflow
  • Roughly follow the Julia 1.6 docs

Pitch

I will be teaching a series of instructional Julia sessions once a month. Each month's session will be four hours of lectures, practical examples, and guided projects tailored towards economics research computing using Julia. The first course takes place on Wednesday, January 12th, from 9am to 1pm.

Each month covers a different topic and can be attended in isolation, though the first session covers the basics of Julia and may be useful for more advanced sessions if you are not currently familiar with Julia.

Topic map:

  • Julia Basics (January 12th)
  • Parallelization for Economists (February)
  • Optimization and Automatic Differentiation (March)
  • Performant Programming and Best Practices (April)
  • Bayesian Inference and Probabilistic Programming (May)

The first session is currently being held virtually via Zoom. Later sessions may be in person, circumstances permitting.

No prior Julia experience is required, though it will be useful to have experience programming in any language.

To register to attend, please visit

https://stanford.zoom.us/meeting/register/tJcufuiupj4iHd1pzgQqdsV5jIr8DJY5pF5n

Sketch

Getting started

  • Julia spiel, say some words about what features it has
    • Fast
    • Expressive
    • Easy
    • Pretty good package ecosystem
    • Portable
    • JIT compiled
    • Hackable
  • What Julia is not (yet)
    • A smash-and-grab analysis tool (i.e. going to Julia for less than a minute or two can be counterproductive)
    • A complete substitute for old/robust tools like SAS or Stata, both of which excel at a lot of things (i.e. the GMM libraries in stata are quite good, SAS is highly optimized for streaming large data, etc.)
  • Quick demo of cool things
  • How to install Julia (raw binaries vs. jill?)
  • How to edit Julia files (VSCode, vim, emacs, etc.)
    • Hello world!
    • Hello world for econometricians -- OLS!

Language basics

  • Variable basics
  • Primitive types (numerics/strings/characters)
  • Basic operators (math/bool/etc.)
    • PRACTICE: Basic numeric operations
    • PRACTICE: String concatenation
  • Functions
    • returning singletons, tuples, etc.
    • Arguments
    • Keyword arguments
    • Composition/pipes
  • Array types
    • Indexing
    • Length, size, etc.
    • PRACTICE: map, broadcasting,
    • PRACTICE: reduce
    • PRACTICE: push!, append!
    • LinearAlgebra basics
  • Broadcasting!
  • Control flow
    • begin
    • if
    • short-circuiting
    • while/for
    • try/catch/finally
  • Types
    • Abstract (briefly)
    • Composite
    • Mutable vs. immutable
    • Parametric
    • Constructors (inner vs. outer)
  • Methods and multiple dispatch
  • Brief section on Julia vs. R, Matlab, Python

Package tour

  • CSV and DataFrames
    • Loading data
    • How to sling data like a badass (groupby, select, filter, etc.)
    • Queryverse
  • Statistics (mostly OLS, fixed effects, clustering, etc.)
    • FixedEffectModels (faster than R's felm, and Stata's reghdfe, though comparable to R feols)
    • GLM (basics mode)
    • Tables with RegressionTables.jl
  • Visualization (basic plots)
    • Plots.jl
    • Makie, AlbebraOfGraphics
    • UnicodePlots.jl
  • Maximum likelihood
    • Distributions.jl
    • Optim.jl

Cool stuff

  • SciML and the whole differential equations universe
    • ModelingToolkit
  • Optimization
    • JuMP
    • Convex
    • Optim
  • Graphs
  • OnlineStats
  • Math crap
    • ApproxFun
    • TaylorSeries
    • Polynomials
    • QuadGK

Derivatives

  • ChainRules
  • ForwardDiff
  • ReverseDiff
  • FiniteDiff/FiniteDifferences
  • Experimental
    • Zygote
    • Enzyme
    • Diffractor

Bayes

  • Turing
  • Soss
  • Stan

Performance

  • ThreadsX
  • Transducers
  • BangBang

ML stuff

  • MLJ
  • Flux
  • Clustering
  • DecisionTree
  • Lasso