GitHub

RxInfer and GraphPPL integration

CausalDynamics.jl handles identification (backdoor adjustment sets, prepare_for_rxinfer). GraphPPL.jl specifies the observational generative head; RxInfer.jl runs variational message-passing inference.

Load the extension:

using CausalDynamics
using RxInfer   # activates CausalDynamicsRxInfer

Workflow

  1. Build a DAG and attach tabular data (DataFrame or NamedTuple).
  2. prepare_for_rxinfer(g, X, Y; node_names=...) — confounders + identifiability flag.
  3. infer_backdoor_effect(g, data, X, Y; ...) — Frisch–Waugh partialing in Julia, then a conjugate GraphPPL head and VI for τ.

Confounders in the identified adjustment set are partialled out of outcome and treatment (residualise_backdoor) so the RxInfer head is a single conjugate slope τ on residualised (y, x). A full multi-coefficient GraphPPL regression with latent β vectors is reserved for a later release (non-conjugate without custom initialisation).

using CausalDynamics, RxInfer, Graphs, DataFrames, Random

g = DiGraph(3)
add_edge!(g, 1, 2)
add_edge!(g, 1, 3)
add_edge!(g, 2, 3)
names = Dict(1 => :Z, 2 => :X, 3 => :Y)
data = DataFrame(Z=randn(100), X=randn(100), Y=randn(100))

result = infer_backdoor_effect(g, data, 2, 3; node_names=names)
result.τ_mean   # posterior mean; do not use Statistics.mean(result.τ_posterior)

See examples/rxinfer_backdoor.jl in the package root.

API

FunctionRole
has_rxinfer()Extension loaded?
prepare_for_rxinferIdentification bridge (always available)
residualise_backdoorFrisch–Waugh partialing of (y, x) on confounders
backdoor_graphppl_modelGraphPPL @model generator (backdoor_gaussian_ate)
ppl_data_from_specTabular → residualised RxInfer data tuple
infer_backdoor_effectFull identify + partial + infer pipeline
posterior_mean_τScalar mean from τ_posterior marginals

Process vs Pearl naming

Package APIs stay in Pearl / SciML vocabulary. For a process-metaphysics gloss used in the CDCS book, see the book’s Concept Reference (Table 8), not this package manual.

Dependencies

Optional weak dependencies: GraphPPL, RxInfer. They are not required for core CausalDynamics or the CDCS book Project.toml.

If Pkg.resolve() fails after adding RxInfer, cap Graphs at 1.13 in your application environment (RxInfer 5.x vs DataStructures 0.19). See AgeSCM docs/RXINFER_DEPS.md for a resolved stack example.

Further reading