TODO.md

TODO

2021.01.14

Primary

• Pure Rust implementation of Linear Algebra
  • LU (Completely Pivoting)
  • LU (Partial Pivoting)
  • QR
  • SVD
• Add more IO options for DataFrame
  • CSV (csv feature)
  • NetCDF (nc feature)
  • JSON
  • Arrow IPC
  • Parquet

Subs

• Implement WithJSON for DataFrame
  • to_json_value
  • from_json_value
• Implement various pdf
  • Bernoulli
  • Beta
  • Binomial
  • Dirichlet
  • Gamma
  • Student's t
  • Uniform
  • Wishart
• Implement special polynomial
  • Legendre
  • Bessel
  • Hermite
• Implement convenient structure of Neural Network
• Documentized
  • Vector
  • Matrix
  • Linear Algebra
  • Functional Programming
  • Statistics
  • Interpolation & Spline
  • ODE
  • Macros
  • Machine Learning
  • Optimize
  • Automatic Differentiation
  • DataFrame
• Add Statistical regression
  • Gaussian Kernel
  • Logistic Kernel
• Make or Use pure Rust plot library
• Implement more Eigenvalue algorithms
• Implement more spline algorithms
• Complex matrix

Complete

• Can choose API - MATLAB, Python, R
• Implement Plot
• Re-write numerical module
• Optimize
  • Linear Regression
  • Non-linear Regression
    • Gauss-Newton (But not yet merged)
    • Gradient Descent
    • Levenberg-Marquardt
• Implement DataFrame
• Implement higher order automatic derivatives
• Generic trait for Automatic differentiation (Create AD trait)
• Separate DataFrame from dataframe feature. (And rename dataframe feature to some awesome name)
• Reduce compile time
  • Replace proc_macro for AD with ordinary macro or Enum
• Make csv optional
• Remove dual, hyperdual and modify Real, Number (How to bind f64 & AD effectively?)