Specification of a computational architecture derived from the relational derivation. The machine’s primitives (cells, traces, profiles) are operational reflections of the mathematical structures produced by the derivation.

The full specification is in the architecture document.

Status

This specification was originally drafted as a monolithic document. It needs revision to:

  • Align terminology with the current derivation (the 18-step version in relationality/texts/).
  • Split into modular sections following the ASR directory organization.
  • Remove references to the obsolete “quasicrystalline hypertensor topos machine” naming.
  • Add concrete implementation examples.

Ideas from triage (not yet integrated)

The triage/ directory contains substantial supporting material for this specification that has not yet been incorporated:

Information engine pipeline (triage/specifications/information-engine.md) — a five-stage canonical pipeline: intake/coding, measurement/normalization, towers and flow, sheaf/hypertensor state, domain views. Key idea: domain specifications (physics, knowledge graph, policy, finance) are slices of the shared hypertensor, not separate systems. No separate “data vs knowledge” — all state is recognitions and observables acted on by earned acts.

Relational Information Physics Computer (triage/computer-architecture/informational-physics-computer.md) — top-level architecture document describing the computer as: core engine (information engine pipeline) + surrounding services (runtime, measurement fabric, analytics, knowledge/traceability graph, storage, network, policy). Invariant: all persistent state is recognitions and sections of the hypertensor sheaf.

Parametric runtime (triage/computer-architecture/parametric-runtime.md) — the external I/O surface. Programs behave identically across contexts (abstraction theorem property). Context manager maintains indexed fibers per tenant; parametric interpreter respects effect/coeffect rules. The runtime is explicitly thin: FastAPI routes to runbooks via RewriterEngine.

Runtime architecture (triage/computer-architecture/runtime.md) — evaluator generates tasks, scheduler enforces dependency order, concurrency layer executes, policy gate checks before dispatch, topology constraints guide placement, measurement emitters stream telemetry.

Simulation specifications (triage/specifications/) — 30+ specs for individual subsystems including: Heyting algebra simulation, measurement fabric, Markov dynamics, modal logic, sheaf topos, spectral theorem, knowledge graph, directed homotopy, infinity site, modality tower. Each is a self-contained specification for simulating one algebraic structure or subsystem of the relational computer.

Theorem files (triage/theorem/) — 70+ theorem files providing the mathematical derivations that the specifications operationalize. These form the justification layer: acts produce theorems, theorems produce specifications, specifications produce architecture.

Architecture modules (triage/computer-architecture/) — 25+ architecture module descriptions including: storage service, network controller, job orchestrator, monitoring/audit, identity/access, deployment, data catalog, sheaf assembly, syntax service, testing harness.

These materials form a complete theorems-to-specifications-to-architecture stack. When the monolithic architecture document is split into modular sections, these triage files provide the content for each module.