Crate antecedent 

Unified static/temporal CausalAnalysis facade (identify → estimate → refute).

Antecedent is an identification-first causal inference engine: identification is evaluated before estimation, refuters run against the estimate, and uncertainty about causal structure (equivalence classes, graph posteriors) is retained rather than silently resolved.

Quick start

use antecedent::prelude::*;
use antecedent::RefuteSuite;

let schema = CausalSchemaBuilder::new()
    .continuous("t")
    .treatment()
    .continuous("y")
    .outcome()
    .continuous("z")
    .context()
    .build()
    .unwrap();
let data = TabularData::from_f64_columns([
    ("t", &[0.0_f64, 1.0, 0.0, 1.0, 0.0, 1.0, 0.0, 1.0][..]),
    ("y", &[1.0_f64, 3.1, 1.2, 2.9, 0.9, 3.0, 1.1, 3.2][..]),
    ("z", &[0.0_f64, 1.0, 0.2, 0.8, 0.1, 0.9, 0.3, 0.7][..]),
])
.unwrap();
let dag = Dag::from_named_edges(&schema, &[("z", "t"), ("z", "y"), ("t", "y")]).unwrap();
let q = AverageEffectQuery::binary_ate(schema.id_of("t").unwrap(), schema.id_of("y").unwrap());
let result = CausalAnalysis::builder()
    .data(data)
    .graph(dag)
    .query(q)
    .refute(RefuteSuite::None)
    .bootstrap_replicates(0)
    .build()
    .unwrap()
    .run(&ExecutionContext::for_tests(1))
    .unwrap();
assert!(result.effect().is_finite());

Prefer prelude for day-1 imports. Component crates remain available for stage-specific work.

SPDX-License-Identifier: MIT OR Apache-2.0

Re-exports

pub use analysis::AnalysisStageEvent;

pub use analysis::BatchAnalysis;

pub use analysis::CausalAnalysis;

pub use analysis::CausalAnalysisBuilder;

pub use analysis::ComputeBudget;

pub use analysis::LatencyMode;

pub use analysis::PreparedAnalysis;

pub use analysis::RdConfig;

pub use analysis::RefuteSuite;

pub use analysis::StageResultSink;

pub use error::AnalysisError;Deprecated

pub use error::CausalError;

pub use estimate::EstimatorId;

pub use estimate::IdentifierId;

pub use inference::BayesianConfig;

pub use inference::InferenceMode;

pub use options::DiscoveryAccept;

pub use options::FdrControl;

pub use planner::CompiledAnalysis;

pub use planner::GraphInput;

pub use result::CausalAnalysisResult;

pub use query::*;

Modules

analysis

Unified CausalAnalysis facade.

callback_plan

Physical-plan marking for Python / slow-path callbacks.

design

experiment / measurement design facade helpers.

discovery

Coarse discovery stage APIs for the facade.

discovery_defaults

Shared discovery constraint defaults and CI resolution.

error

Facade errors.

estimate

Estimation stage types and strategy identifiers.

gcm

GCM workflow helpers (fit → sample → CF → anomaly).

graph

Graph types and construction helpers.

identify

Identification stage types.

inference

Bayesian inference configuration for the facade.

io

Graph interchange and durable artifacts.

options

Shared facade option enums (API hygiene).

planner

Logical / physical analysis planning.

prelude

Day-1 imports for the antecedent facade.

query

Typed causal queries (re-exported from antecedent-core).

result

Analysis result artifact.

review

Graph review gate for discovery outputs.

state

incremental antecedent-state facade helpers.

strategy_table

Identifier / estimator strategy tables for plan compilation and static execution . Incremental extraction from the analysis workflow — does not replace crate::CausalAnalysis / plans / crate::CausalAnalysisResult.

validate

Validation / refutation types.

Structs

CausalPosterior

Causal posterior over an identified functional.

Dag

Static directed acyclic graph over variables.

DenseNodeId

Compact dense node index used in algorithmic paths.

EffectEstimate

Point estimate with uncertainty.

TemporalDag

Directed acyclic graph over lagged (VariableId, Lag) nodes.