Skip to content

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

35 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Graph Preconditioning for High-Dimensional Fixed Effects Regression

This repository contains the paper by Alexander Fischer and Kristof Schröder, along with the code used to produce its tables, figures, and benchmark results.

The paper explains why the method of alternating projections can be slow in fixed-effects regressions and introduces a graph-based preconditioner for the LSMR solver. The benchmarks compare this approach with PyFixest's MAP solver, R's fixest, and Julia's FixedEffectModels.jl and GLFixedEffectModels.jl.

Compile the paper

Pixi manages the Python and Typst environment. Install the locked environment and compile the paper with:

pixi install --locked
pixi run compile

The compiled PDF is written to graph_preconditioner_hdfe.pdf. Compiling the paper uses the tables and figures already in the repository; it does not rerun the benchmarks.

Reproduce the benchmark results

The full benchmark suite also uses R and Julia. Those runtimes and their packages must be installed separately. See REPRODUCING.md for the required versions, thread settings, and commands.

The complete reproduction can take a long time and includes benchmarks with up to ten million observations. The measurement rules are recorded in PROTOCOL.md, while benchmarks/README.md describes how the benchmark code is organized.

Repository guide

  • graph_preconditioner_hdfe.typ is the paper source.
  • benchmarks/ contains run-local data generators, native runners, explicit experiment configs, and the small coordinator.
  • scripts/ turns recorded measurements into paper tables and figures.
  • generated/ contains the table fragments and values included by the paper.
  • within-docs/ contains technical notes about the solver.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages