SC3D: Dynamic and Differentiable Causal Discovery for Temporal and Instantaneous Graphs

Sourajit Das, Dibyajyoti Chakraborty, Romit Maulik · 2026

Read SC3D: Dynamic and Differentiable Causal Discovery for Temporal and Instantaneous Graphs by Sourajit Das, Dibyajyoti Chakraborty, Romit Maulik (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-02830.txt.

Description

Sourajit Das and 2 co-authors present SC3D in a 2026 paper on finding causes in time series data. A time series records values that change over time. Causes can act after a delay, or at the same moment, which makes the search very large.

SC3D works in two stages. First it preselects likely links between variables. Then it refines those links while enforcing that instant links form no cycles. Tests on synthetic systems, chaotic benchmarks, changing dynamics, and real datasets follow. The authors report steadier and more accurate recovery of both delayed and instant causal links than earlier methods.

Catalog record

CREATORSourajit Das, Dibyajyoti Chakraborty, Romit Maulik
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-02830

Browse the full library · About the Tower of Babel