Tri-Engine Reasoning Architecture
Reference: This page is a source-based technical synthesis of the LaTeX chapter cited below. For exact syntax, availability or ABI signatures, verify the versioned source, manifest and executable tests.
Scope and source boundary
The numerical, semantic and action-authority paths have different roles and should not be collapsed into a single “reasoning engine.”
The neuro-symbolic cycle distinguishes observations, candidate proposals, numerical scores, logical validation and governed materialization. It is useful to rank likely actions or relations before paying the full symbolic evaluation cost, but a candidate only becomes an authoritative fact after the appropriate semantic contract is satisfied.
Engineering rules
- Track the transition from observation to proposal to accepted fact.
- Use ranking as an optimization, not a substitute for proof.
- Retain provenance and uncertainty until validation completes.
- Apply bounds, fallback strategies and explicit rejection on incomplete results.
Chapter outline (original LaTeX headings)
- Effect We Want
- Complete Runtime Cycle
- Neural Inputs
- Safe and Approximate Modes
- Materialization Policy
- Learning from the Dictionary and Lattice
LaTeX provenance
Primary chapter: logicells-neuro-symbolic-architecture-guide/chapters/operational-neuro-symbolic-cycle.tex.