Optimal Nonlinear Generator Selection
Background: The Generalized Covariance (GCov) test utilizes a sequence of nonlinear transformations (generators) to probe for dependence; however, the choice of these generators significantly impacts the test’s power and asymptotic efficiency.
Question / Future Work: Determining a principled, automated selection procedure for the sequence of nonlinear transformations used in GCov tests remains an open challenge, particularly for models with heavy-tailed or non-standard error distributions.
Why It Matters: The efficiency and sensitivity of GCov-based tests are highly dependent on the choice of generator functions, making this a critical optimization problem for reliable inference.
Metadata & Links
- created_at
- 2026-04-17T05:45:33Z
- source_papers