Fork of OxiEML (cool-japan/oxieml) extended with the anti-holomorphic companion operator eml★(x, y) = exp(x) - ln(conj(y)) from Monnerot (2026). Extended grammar: S → 1 | eml(S, S) | eml★(S, S) This is the first symbolic regression engine that natively covers both holomorphic and anti-holomorphic function spaces in a unified algebraic framework. Two operators and one constant generate all elementary functions including conj(z), Re(z), Im(z), |z|², and conformal primary wavefunctions. Features: Rust implementation, gradient-based symbolic regression, canonical constructions (conj, real_part, mod_squared), parser support for E* notation, complete EML tree enumeration with EmlStar topologies. Paper: Monnerot, A. (2026). eml★: Minimal Anti-Holomorphic Extension of the EML Sheffer Operator. DOI: 10.5281/zenodo.20091022GitHub: https://github.com/antparis/oxieml-starOriginal OxiEML: https://github.com/cool-japan/oxieml