Relax regularity assumptions underlying the equivalence

Relax the regularity assumptions, including the local product structure that requires the existence of regular (quadratic-mean differentiable) coordinate submodels perturbing the target parameter and nuisance parameter independently, under which the equivalence between Neyman orthogonality and pathwise differentiability is established for semiparametric models.

Background

The paper proves that Neyman orthogonality and pathwise differentiability are equivalent under a set of structural and smoothness conditions. A central requirement is a local product structure (Assumption 1), which guarantees the existence of regular submodels that independently perturb the target parameter and nuisance parameter to first order. Additional differentiability and boundedness conditions regulate behavior along submodels and enable differentiation of expectations.

The authors note that, especially in complex semiparametric settings (e.g., constrained nuisance spaces or functionals defined implicitly), verifying such assumptions can be challenging. They explicitly flag relaxing these conditions as an open direction, motivating a search for weaker, more broadly applicable assumptions that still ensure the equivalence.

References

Several directions remain open. Foremost, the regularity conditions we impose, notably the existence of coordinate submodels witnessing local product structure, can be nontrivial to verify in complex semiparametric problems, such as those involving constrained nuisance spaces or functionals defined through implicit equations. Relaxing these conditions, extending the equivalence to settings with non-smooth functionals, and developing systematic tools for constructing coordinate submodels in applied problems would be natural next steps.

On the Equivalence between Neyman Orthogonality and Pathwise Differentiability  (2603.15817 - Chen et al., 16 Mar 2026) in Section 4 (Discussion)