# Iterative Identification Closure: Amplifying Causal Identifiability in Linear SEMs

Source: [Iterative Identification Closure: Amplifying Causal Identifiability in Linear SEMs](<https://www.sota2.com/research/paper/69dc78757d0172140ec78a3c>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-14T05:00:11.146Z

Authors: Ziyi Ding, Xiao-Ping Zhang

Published: 2026-04-10T13:20:22.000Z

## Abstract

The Half-Trek Criterion (HTC) is the primary graphical tool for determining generic identifiability of causal effect coefficients in linear structural equation models (SEMs) with latent confounders. However, HTC is inherently node-wise: it simultaneously resolves all incoming edges of a node, leaving a gap of "inconclusive" causal effects (15-23% in moderate graphs). We introduce Iterative Identification Closure (IIC), a general framework that decouples causal identification into two phases: (1) a seed function S\_0 that identifies an initial set of edges from any external source of information (instrumental variables, interventions, non-Gaussianity, prior knowledge, etc.); and (2) Reduced HTC propagation that iteratively substitutes known coefficients to reduce system dimension, enabling identification of edges that standard HTC cannot resolve. The core novelty is iterative identification propagation: newly identified edges feed back to unlock further identification -- a mechanism absent from all existing graphical criteria, which treat each edge (or node) in isolation. This propagation is non-trivial: coefficient substitution alters the covariance structure, and soundness requires proving that the modified Jacobian retains generic full rank -- a new theoretical result (Reduced HTC Theorem). We prove that IIC is sound, monotone, converges in O(\|E\|) iterations (empirically \<=2), and strictly subsumes both HTC and ancestor decomposition. Exhaustive verification on all graphs with n\<=5 (134,144 edges) confirms 100% precision (zero false positives); with combined seeds, IIC reduces the HTC gap by over 80%. The propagation gain is gamma~4x (2 seeds identifying ~3% of edges to 97.5% total identification), far exceeding gamma\<=1.2x of prior methods that incorporate side information without iterative feedback.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Causal Effect Estimation on 6-node graph (n=5000, 100 trials)](<https://www.sota2.com/research/sota/causal-effect-estimation-on-6-node-graph-n-5000-100-trials.md>): Causal effect estimation; 6-node graph n=5000 samples, averaged over 100 trials; Absolute Bias: 0.

- [Causal Identification on General random graphs n=6 (exhaustive)](<https://www.sota2.com/research/sota/causal-identification-on-general-random-graphs-n-6-exhaustive.md>): Causal Identification; General random graphs n=6 (exhaustive); Identification Rate: 97.5.

- [Causal Identification on Random Graphs (n=4) (exhaustive enumeration)](<https://www.sota2.com/research/sota/causal-identification-on-random-graphs-n-4-exhaustive-enumeration.md>): Causal Identification; Random Graphs (n=4) (exhaustive enumeration); Both Identified (%): 288.

- [Causal Identification on Random Graphs n=5 (exhaustive enumeration)](<https://www.sota2.com/research/sota/causal-identification-on-random-graphs-n-5-exhaustive-enumeration.md>): Causal Identification; Random Graphs n=5 (exhaustive enumeration); Count (Both Identified): 108368.

- [Causal Identification on Large random graphs n=10](<https://www.sota2.com/research/sota/causal-identification-on-large-random-graphs-n-10.md>): Causal Identification; Large random graphs n=10; Identification Rate: 98.6.

- [Causal Identification on Large random graphs n=20](<https://www.sota2.com/research/sota/causal-identification-on-large-random-graphs-n-20.md>): Causal Identification; Large random graphs n=20; Identification Rate: 99.2.

- [Causal Identification on Large random graphs n=50](<https://www.sota2.com/research/sota/causal-identification-on-large-random-graphs-n-50.md>): Causal Identification; Large random graphs n=50; Identification Rate: 99.3.

- [Causal Identification on Large random graphs n=100](<https://www.sota2.com/research/sota/causal-identification-on-large-random-graphs-n-100.md>): Causal Identification; Large random graphs n=100; Identification Rate: 99.7.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.09309>)

[PDF](<https://arxiv.org/pdf/2604.09309.pdf>)
