Skip to content

Decision ADR 0013 Moduli Infrastructure Ownership

Chris Dare edited this page Nov 15, 2026 · 1 revision

ADR-0013 — Own reusable moduli infrastructure in the monorepo

  • Status: partially superseded by ADR-0014. Its paths (the Decision list) and the SF8/SF9 ownership lines under Consequences are superseded; the monorepo choice and the final trust clause remain in force.
  • Date: 2026-08-17 (America/New_York)
  • Decider: Chris Dare
  • Decision: implement relative-perfect, Quot, stack, and good-moduli infrastructure under the repository's Mathlib-style subject hierarchy.

Context

The stability-in-families roadmap previously parked Quot schemes, algebraic stacks, and good moduli spaces behind an ownership decision. The repository is now a single DerivedAlgGeo library and needs these subjects for both the current stability program and later derived-algebraic-geometry developments. A pin-based external-owner boundary no longer matches that direction.

Decision

The monorepo owns the reusable layers, with generic geometry separated from Bridgeland-specific adapters:

  • relative-perfect objects and their moduli problem live under DerivedAlgGeo/AlgebraicGeometry/Moduli/PerfectComplex;
  • Quot and bounded parameter-space constructions live under DerivedAlgGeo/AlgebraicGeometry/Quot;
  • stacks in groupoids and descent live under DerivedAlgGeo/AlgebraicGeometry/Stacks;
  • good-moduli constructions live under DerivedAlgGeo/AlgebraicGeometry/Moduli/GoodModuli;
  • stability-specific use of those layers remains under DerivedAlgGeo/CategoryTheory/Triangulated/StabilityCondition/Families.

Dependencies point from generic categorical and site foundations to stacks, from coherent and derived geometry to Quot and perfect-complex moduli, and from stacks plus parameter spaces to good moduli. The families layer may consume those subjects. None of the generic geometry subjects imports a Bridgeland-specific theorem merely to state its core definitions.

The current relative-perfect object file temporarily imports the honest Dqc construction from the families layer because that construction predates this decision. Issue #528 owns moving the general scheme-derived seam to a neutral algebraic-geometry module; this temporary edge must not spread to the stack or Quot substrates.

Consequences

  • Issue #214's parked ownership decision is resolved in favor of the monorepo.
  • SF8 owns relative-perfect objects, their groupoid-valued moduli problem, boundedness witnesses, and Quot/parameter spaces.
  • SF9 owns the reusable stack substrate, algebraicity applications, semistable reduction, and good moduli spaces.
  • No structure field, typeclass instance, or roadmap status may stand in for an unproved algebraicity, boundedness, properness, or descent theorem.

Exported from .claude/decisions/ADR-0013-moduli-infrastructure-ownership.md at the commit that removed it from the tree.

Home

Architecture

CI and runners

Decision records

Roadmaps

Clone this wiki locally