Package: didgpu 0.1.2

didgpu: GPU-Accelerated, Checkpointed Difference-in-Differences Estimator

A clean-room reimplementation of the heterogeneity-robust dynamic difference-in-differences estimator of de Chaisemartin and D'Haultfoeuille (2024) <doi:10.1162/rest_a_01414>, designed for long-running econometric work: per-cell checkpointing to disk, resumable runs after crash or out-of-memory, parallel cluster bootstrap, configurable backends (pure R, optional 'Rcpp' and 'CUDA'), and bit-for-bit numerical equivalence with the reference 'DIDmultiplegtDYN' package across binary, multivalued, and continuous treatments.

Authors:Joshua Ammons [aut, cre, cph]

didgpu_0.1.2.tar.gz
didgpu_0.1.2.zip(r-4.7-x86_64)didgpu_0.1.2.zip(r-4.7-arm64)didgpu_0.1.2.zip(r-4.6-x86_64)didgpu_0.1.2.zip(r-4.6-arm64)didgpu_0.1.2.zip(r-4.5-x86_64)
didgpu_0.1.2.tgz(r-4.6-x86_64)didgpu_0.1.2.tgz(r-4.6-arm64)didgpu_0.1.2.tgz(r-4.5-x86_64)didgpu_0.1.2.tgz(r-4.5-arm64)
didgpu_0.1.2.tar.gz(r-4.7-arm64)didgpu_0.1.2.tar.gz(r-4.7-x86_64)didgpu_0.1.2.tar.gz(r-4.6-arm64)didgpu_0.1.2.tar.gz(r-4.6-x86_64)
didgpu_0.1.2.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
didgpu/json (API)

# Install 'didgpu' in R:
install.packages('didgpu', repos = c('https://joshuaammons.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/joshuaammons/didgpu/issues

Uses libs:
  • c++– GNU Standard C++ Library v3

On CRAN:

Conda:

causal-inferencecudadifference-in-differenceseconometricsevent-studygpu-computingpanel-datacpp

3.02 score 3 stars 40 exports 8 dependencies

Last updated from:adf4955f45. Checks:15 OK. Indexed: yes.

TargetResultTimeFilesSyslog
linux-devel-arm64OK1665
linux-devel-x86_64OK1727
source / vignettesOK380
linux-release-arm64OK1557
linux-release-x86_64OK1402
macos-release-arm64OK1094
macos-release-x86_64OK2450
macos-oldrel-arm64OK1247
macos-oldrel-x86_64OK2889
windows-devel-arm64OK2032
windows-devel-x86_64OK1945
windows-release-arm64OK2067
windows-release-x86_64OK1914
windows-oldrel-x86_64OK1839
wasm-releaseOK209

Exports:didgpudidgpu_aggregate_cellsdidgpu_backend_infodidgpu_bacondidgpu_bootstrap_moredidgpu_bydidgpu_by_pathdidgpu_comparedidgpu_compute_pathsdidgpu_cpu_hellodidgpu_csdidgpu_cs_aggregatedidgpu_cs_continuousdidgpu_did_continuousdidgpu_did_staticdidgpu_equivalencedidgpu_estimate_runtimedidgpu_event_study_datadidgpu_fectdidgpu_fect_equivalencedidgpu_fect_placebodidgpu_freyaldenhovendidgpu_glancedidgpu_has_cuda_supportdidgpu_honest_diddidgpu_init_checkpointdidgpu_joint_placebodidgpu_lb_frac_affecteddidgpu_load_checkpointdidgpu_loodidgpu_resumedidgpu_run_saxpydidgpu_simulate_paneldidgpu_simulate_panel_bidirdidgpu_summarize_paneldidgpu_test_sharp_nulldidgpu_tidydidgpu_twfeglance.didgpu_resulttidy.didgpu_result

Dependencies:data.tablejsonlitelatticeMASSRcppRcppEigensandwichzoo

Readme and manuals

Help Manual

Help pageTopics
didgpu: GPU-Accelerated, Checkpointed Difference-in-Differencesdidgpu-package
Coefficient extractor for didgpu_resultcoef.didgpu_result
Confidence intervals for didgpu_resultconfint.didgpu_result
Run a checkpointed, GPU-capable dynamic DiD estimationdidgpu
Read all committed cells back into memorydidgpu_aggregate_cells
Report installed backends and their availabilitydidgpu_backend_info
Goodman-Bacon decomposition of the TWFE DiD estimatordidgpu_bacon
Add more bootstrap reps to an existing checkpointdidgpu_bootstrap_more
Estimate didgpu separately for each level of a grouping variabledidgpu_by
Run didgpu separately per treatment trajectorydidgpu_by_path
Compare didgpu r-backend to the DIDmultiplegtDYN referencedidgpu_compare
Add a per-group treatment-trajectory column to a paneldidgpu_compute_paths
Smoke target for the CPU build pathdidgpu_cpu_hello
Estimate Callaway-Sant'Anna (2021) staggered-treatment DiDdidgpu_cs
Re-aggregate a fitted didgpu_cs result with a different schemedidgpu_cs_aggregate
Callaway-Goodman-Bacon-Sant'Anna (2024) continuous-treatment DiDdidgpu_cs_continuous
Continuous-treatment DiD with no stayers (de Chaisemartin-D'Haultfoeuille 2024)didgpu_did_continuous
de Chaisemartin-D'Haultfoeuille (2020) DID_M instantaneous estimatordidgpu_did_static
Pre-trends equivalence (TOST) test for a didgpu event studydidgpu_equivalence
Estimate the runtime of a planned didgpu() calldidgpu_estimate_runtime
Extract long-format event-study data from a didgpu_resultdidgpu_event_study_data
Counterfactual-prediction DiD estimators (fect family)didgpu_fect
Equivalence test on fect placebosdidgpu_fect_equivalence
Run placebo (pre-treatment) tests on a fitted fect modeldidgpu_fect_placebo
Freyaldenhoven-Hansen-Shapiro (2019) pre-event proxy event studydidgpu_freyaldenhoven
One-row glance summary of a didgpu_resultdidgpu_glance
Report whether didgpu was compiled with CUDA supportdidgpu_has_cuda_support
Sensitivity analysis for a fitted event-study DiDdidgpu_honest_did
Initialise a checkpoint directorydidgpu_init_checkpoint
Windowed / subset joint test of pre-treatment placebosdidgpu_joint_placebo
Sharp lower bound on the fraction of always-takers affecteddidgpu_lb_frac_affected
Load checkpoint metadata and manifestdidgpu_load_checkpoint
Leave-one-out (LOO) robustness analysisdidgpu_loo
Quick resume helperdidgpu_resume
Run a SAXPY (y = a*x + y) on the GPUdidgpu_run_saxpy
Generate a simulated DiD panel with known event-time profiledidgpu_simulate_panel
Generate a panel with BOTH switcher-in and switcher-out unitsdidgpu_simulate_panel_bidir
Summarize a panel before fittingdidgpu_summarize_panel
Sharp test of full mediationdidgpu_test_sharp_null
Tidy a didgpu_result into a one-row-per-coefficient data.framedidgpu_tidy
Naive two-way fixed-effects (TWFE) dynamic event studydidgpu_twfe
broom::glance method for didgpu_resultglance.didgpu_result
Plot method for didgpu_loo_result (tornado plot)plot.didgpu_loo_result
Event-study plot of a didgpu resultplot.didgpu_result
Print method for didgpu_baconprint.didgpu_bacon
Print method for didgpu_by_resultprint.didgpu_by_result
Print method for didgpu_cs_continuous_resultprint.didgpu_cs_continuous_result
Print method for didgpu_cs_resultprint.didgpu_cs_result
Print method for didgpu_did_continuous_resultprint.didgpu_did_continuous_result
Print method for didgpu_did_static_resultprint.didgpu_did_static_result
Print method for didgpu_equivalenceprint.didgpu_equivalence
Print method for didgpu_fect_placeboprint.didgpu_fect_placebo
Print method for didgpu_freyaldenhoven_resultprint.didgpu_freyaldenhoven_result
Print method for didgpu_honest_did_resultprint.didgpu_honest_did_result
Print method for didgpu_joint_placeboprint.didgpu_joint_placebo
Print method for didgpu_loo_resultprint.didgpu_loo_result
Print method for didgpu_resultprint.didgpu_result
Print method for didgpu_twfe_resultprint.didgpu_twfe_result
Summary method for didgpu_resultsummary.didgpu_result
broom::tidy method for didgpu_resulttidy.didgpu_result
Variance-covariance matrix of estimatesvcov.didgpu_result