From the Phase Calculus side, I wrote a short note showing that EML is not primitive in a dependency-ordered sense. It is a clean continuous-shadow composite:

eml_PC(x,y) = Sub_PC(Exp_PC(x), Log_PC(y)) = e^x - ln y

Phase Calculus starts from a lower discrete lifted state

Xi-hat = (A, q, theta, kappa, c)

and recovers arithmetic, phase, native π, calculus, and finally EML itself via quotient descent.

Why this matters in practice: The same framework delivers a native, certified Pi-spigot with strong real-world performance:

1,000,000 decimal digits streamed natively in ~1.65 s Certified safe prefix of 1,366,163 digits Bank-indexed random access up to 1.90 billion digits/second (1024-digit blocks)

Phase Calculus derives EML naturally, but EML cannot derive or reconstruct the underlying lifted-state machinery.

I’m looking for technical attacks on the dependency chain, especially the quotient-descent step and the continuous-shadow composite theorem. Full paper, Lean formalization, SymPy validation (PASS result), Jupyter notebook, and all benchmark data are here:

Paper: https://www.academia.edu/165859479/Quotient_Descent_and_the_EML_Operator_Why_Continuous_Composites_Are_Not_Primitives

Validation bundle: https://zenodo.org/records/19688963