The Research section is organized around reproducible experiments (eval suites), targeted investigations (capacity and decoding behavior), and research-only specifications that are not yet implemented.

Experiments

Benchmarks

Performance & accuracy across strategies, geometries, and reasoning priorities.

Evaluation

Saturation Evaluation

Capacity study: superposition stress tests and holographic membership queries.

Capacity

Research Snapshot (Dec 2025)

Archived snapshot of metrics, tables, and diagrams previously shown on the index page.

Snapshot

Directions

HDC Strategy Directions

Dense, SPHDC, Metric-Affine, EMA, and EXACT research directions.

Directions

URC Implementation Vision

Step-by-step implementation plan through DS49/DS51/DS50 and the backend roadmap.

URC

Dynamic Size Representations

Elastic geometries (EMA/EXACT) and why “growth” is a key capacity research theme.

Research Theme

Universal Theory Engine (UTE)

Roadmap for compositional theories, evidence, causality, uncertainty, numeric models, revision, and experiment planning.

Research Theme

Agentic Code Generation

Using LLMs + eval data to synthesize patches, simulations, and DSL theory fragments in a reproducible loop.

Agents

Holographic Reasoning Directions

Priorities, decoding workflows, and when holographic steps help vs. need symbolic validation.

Directions

Roadmap

Near-term and long-term plan for scaling experiments and DSL growth.

Planning

Research Specs

These pages summarize DS documents that are explicitly marked as research / exploratory / proposed (not fully implemented), and connect them to related experiments.

Session Holo API (Proposed)

High-level Session functions for holographic query and batch Q/A over loaded theories.

DS27

Persistence & KB Serialization

Deterministic storage concerns (EXACT dictionary, KB format), and why persistence is a separate milestone.

DS03b · DS25 · DS26

Pure Symbolic Baseline (Exploratory)

Why a fully separate term-engine backend is kept as research reference rather than planned work.

DS24

Trustworthy AI Patterns

Planning, compliance, bias/consistency, and scientific theory validation patterns (research-only).

DS08 · DS28–DS31