Response to "Comment on 'Manifolds of quasi-constant SOAP and ACSF fingerprints and the resulting failure to machine learn four-body interactions'" [J. Chem. Phys. 156, 034302 (2022)].

It is uncontested that a machine learning scheme cannot correctly reproduce physical properties that vary on a manifold in configuration space if the fingerprint, used as an input for the machine learning scheme, is constant on this manifold. In our original paper, we found several manifolds whose fingerprint variation is sufficiently small to prevent machine learning based on a standard training scheme even if some 40 configurations on the manifold are included in the training set. Response to "Comment on 'Manifolds of quasi-constant SOAP and ACSF fingerprints and the resulting failure to machine learn four-body interactions'" [J. Chem. Phys. [Extracted from the article]