Quotient Descent and the EML Operator: Why Continuous Composites Are Not Primitives

START HERE For reviewers and technical readers: Download `QuotientDescentFULLPACKAGE. zip`. For quick inspection: Open `QuotientDescent. pdf`. For the runnable solver: Download `pcᵥdmₗifteddescentₛolver. zip`. For the immediate evidence: Open `x2ᵣesult. json`, `sinᵣesult. json`, and the benchmark CSV/JSON outputs inside the full package. This bundle places the EML operator inside the Phase Calculus lifted-state framework and includes a sidecar head-to-head benchmark against a shallow EML symbolic-regression comparator. Odrzywołek’s EML result is an important independent discovery: the standard scientific-calculator repertoire admits a striking single-gate compression, eml (x, y) = (x) - (y), together with the constant \ (1\). This bundle does not present EML as an adversarial target. It treats EML as independent evidence for the same deeper structural pressure motivating Phase Calculus: ordinary elementary mathematics has a smaller generative substrate than its visible calculator grammar suggests. The layer distinction is the key result. EML is primitive relative to the ordinary calculator basis. Phase Calculus is primitive relative to the retained lifted-state layer. In Phase Calculus, the EML node is a Red-filter continuous-shadow composite: eml₏₂ (x, y): =Sub₏₂ (Exp₏₂ (x), Log₏₂ (y) ) =eˣ- y. It is therefore not primitive at the lifted-state layer. It is a quotient descendant of the native primitive evolution PCop (): =ₖ, () (), on the full lifted object = (A, q, , , c), evolved by the operator core \Q, B, L\, under the exact quotient law E = G. EML trees embed recursively into Phase Calculus continuous-shadow syntax. The Red image preserves real-analyticity on natural domains and Liouvillian exp-log closure, so functions such as \ (|x|\) near \ (0\) and generic irreducible quintic roots remain outside finite Red scalar discharge. A state-complete Red simulation of one non-trivial lifted step requires three independent retained carriers, so a single scalar EML node is not state-complete relative to the Phase Calculus lifted object. The primitive Phase Calculus operator is already operational. The bundle includes execution witnesses for: native \ (\) prefix streaming: \ (1, 000, 000\) certified decimal digits emitted in \ (1. 65086778\) s, with safe certificate \ (Dₒ₀₅₄=1, 366, 163\) ; certified bank-indexed \ (\) -digit access: 1024-digit block queries at \ (5. 3941900 10^-7\) s/query, equivalent to \ (1. 8983388 10⁹\) certified digits/s inside the native bank; Bring-quintic certification: the strongest non-toy witness in the bundle. All five roots of \ (x⁵-x+1\) returned at depth \ (21\), with half-width below \ (6. 2 10^-9\) and projected residuals below \ (4 10^-15\) by retained lifted-state branch transport, carrying sheet identity and branch history internally until terminal projection. This demonstrates that Phase Calculus is not merely a scalar identity-compression trick; it operates in a state-complete arena where scalar radical discharge is structurally insufficient. Phase Calculus VDM lifted descent solver: a metriplectic, self-terminating dynamical solver derived from the same VDM runtime lineage that produced the cognitive-runtime evidence stack. The solver keeps symbolic recovery inside the extended lifted object, carries VDM field coordinates \ ( (, ) \), debt, \ (kT\), walker/bond state, and Phase Calculus coordinates together, alternates primitive lifted transport with metriplectic relaxation, and opens scalar projection only once at terminal readout. sidecar comparison against a depth-limited EML symbolic-regression comparator. The \ (\) figures deliberately separate streaming generation from certified bank-indexed access: the latter is not a scalar-collapse nth-digit oracle, but a retained-state query over a certified native bank. The head-to-head sidecar is included for provenance and layer separation, not as a takedown. In the shared local benchmark, the EML comparator recovers its native shallow \ ( (x) \) and \ ( (x) \) forms, while the Phase Calculus VDM lifted descent solver recovers \ ( (x) \), \ ( (x) \), \ (x²\), and \ ( (x) \) under terminal projection. This demonstrates that retained lifted-state descent and scalar EML tree search occupy different computational layers. The three included papers give the closed proof surface: `QuotientDescentₐndEMLOperatorᵥ1₅. tex` — main theorem paper with claims C0–C6, state-complete lower bound, projector-preserved invariants, worked \ ( (x) \) descent, and execution artifacts. `PrimitiveOperationDescentᵥ1₅. tex` — step-by-step descent spine from survivor mark to EML. `OriginₐndDependencyChainᵥ1₅. tex` — compact dependency and priority map. The release also includes Lean formalization, SymPy validation, reviewer notebooks, \ (\) -spigot ledger, Bring certificates, benchmark CSV/JSON outputs, closure certificate, and SHA256 hashes.