allaun/Research-Stack
1
0
Fork 0
mirror of https://github.com/allaunthefox/Research-Stack.git synced 2026-07-31 03:05:21 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
42 commits 18 branches 0 tags 425 MiB
Commit graph

16 commits

Author SHA1 Message Date
Brandon Schneider
1a9e95385b integrate: erans field effect spectrum into master synthesis 
Evolve erans from flat histogram coding to spectral decomposition
of residual field (field effect spectrum).

Changes to master synthesis:
- Added erans_field_effect_spectrum to theoretical_foundations
- Updated stage_13: compute residual correlation matrix C,
  eigen-decompose C = UΛU^T, code spectral coefficients with erans
- Updated source to include erans-field-effect-spectrum
- Added 3 new compression gain sources:
  * erans_spectral_compaction (10-20% gain from energy compaction)
  * spectral_pattern_separation (2-3% gain from spectral overlap)
  * famm_spectral_pruning (3-5% gain from residual spectral energy)
- Updated estimated aggregate gain: 20-35% reduction (was 18-28%)
- Added 6 spectral keeper phrases
- Updated core_synthesis to mention spectral decomposition
- Added 5 new tags: erans-field-effect, spectral-encoding,
  residual-field-spectrum, field-effect

Field effect spectrum: residual correlation matrix C captures how
residuals propagate through manifold. Spectral energy compaction
(90% energy in 10% coefficients) provides 10-20% gain over flat
histogram coding. Spectral overlap measure improves OAC gate precision.
FAMM delays use residual spectral energy for context efficiency.
 2026-05-08 14:50:02 -05:00
Brandon Schneider
4626f23be5 ingest: Master Synthesis — complete compression architecture 
Combines ALL theories from first portion to now:
- Density field encoding (semantic manifolds, Morse-Smale complex)
- GCCL-GEC (glyph packets, chirality, typebook, eigenbook)
- OAC (observer-admissible cavities, S3C shells, spherion shaping)
- Hypercube-rhomboid (shear matrix, Gram matrix, geometric compression)
- Radius-ratio motif compression (local admissibility quantization)
- Maximum math density (custom logographic notation, full Unicode)
- Hippocampus tabula plena (full slate initialization, FAMM pruning)
- Engram consolidation (neuron dropout, pattern separation)
- FAMM delay lines (preshaped delays, Q16.16 fixed-point)
- S3C shells (multi-scale coordinate encoding)
- PIST n-D bundle (perturbation encoding)
- erans (enumerative rANS entropy coding)

Core synthesis: Start tabula plena (full Unicode 1,114,112 codepoints +
custom glyphs + omniversal chirality) → represent as semantic density
field → extract Morse-Smale topological skeleton → apply shear matrix
(orthogonal hypercube → correlated rhomboid) → FAMM consolidation
(uniform → sparse structured delays based on eigenvalue spectra) → S3C
shell coordinates → radius-ratio quantization → logographic glyph
selection → GCCL packet construction → OAC speculative manifestation →
gain test filtering → math notation eigenvector encoding → repeat position
encoding → PIST perturbation bundle → erans residual entropy coding →
sparse structured archive.

14 encoding stages, 19 decode stages. Archive format MCA1 with 17 sections.

13 compression gain sources: tabula plena pruning (90-99% of Unicode),
FAMM delay pruning (10-20% context efficiency), OAC gate pruning (2-5%
bloat avoidance), radius-ratio quantization, gain test pruning, shear matrix
pruning (15-30% structured regions), topological skeleton (50-150MB vs
1GB), math notation density, repeat encoding, S3C shell efficiency, PIST
bundle efficiency, erans entropy, hippocampus pattern separation, composite
promotion.

Estimated 18-28% reduction vs current Hutter best + navigable capability.
Biological fidelity: follows hippocampus engram consolidation dynamics
(neuron dropout, pattern separation, discrimination thresholds, inhibitory
plasticity, composite promotion).

18 keeper phrases. Core: The density field is the manifold; the glyph
packets are the navigators; the shear matrix is the map; FAMM is the
temporal wiring.
 2026-05-08 14:50:02 -05:00
Brandon Schneider
0fd42c53ca ingest: Hippocampus Tabula Plena combined approach 
Combines maximum math density + unified compression architecture +
hippocampus engram consolidation (Tomé 2024) + tabula plena insight
(Live Science 2024: hippocampus starts full slate, prunes to sparse).

Key insight: hippocampus starts tabula plena (densely wired, hyperconnected)
and prunes to sparse structured during maturation. Compression does the
same: start with maximum math density (full Unicode 1,114,112 codepoints +
custom glyphs + omniversal chirality) and prune via FAMM delays,
OAC gates, radius-ratio quantization, gain tests to minimal representation.

FAMM pruning model: uniform delays (young hippocampus) → preshaped delays
based on eigenvalue spectra → sparse structured delays (mature hippocampus).

10 compression gain sources: tabula plena pruning (90-99% of Unicode unused),
FAMM delay pruning (10-20% context efficiency), OAC gate pruning (2-5%
bloat avoidance), radius-ratio quantization, gain test pruning, shear matrix
pruning (15-30% structured regions), topological skeleton (50-150MB vs 1GB),
math notation density, repeat encoding, erans entropy.

Estimated 15-25% reduction vs current Hutter best + navigable capability.

14 encoding/decode stages. Archive format HFC1 with 16 sections.

13 keeper phrases. Core: Don't start blank. Start full, then prune.
 2026-05-08 14:50:02 -05:00
Brandon Schneider
4d3a08ab11 ingest: Unified Compression Architecture synthesis 
10-layer architecture synthesizing all expanded theories:
- Layer 0: Raw UTF-8 → semantic density field
- Layer 1: Density field extraction (Morse-Smale complex)
- Layer 2: Hypercube-rhomboid shear (Gram matrix as dictionary)
- Layer 3: S3C shell coordinate encoding
- Layer 4: GCCL-GEC packet encoding (7-field glyphs)
- Layer 5: OAC speculative manifestation
- Layer 6: Radius-ratio local quantization
- Layer 7: FAMM temporal sequencing
- Layer 8: PIST perturbation encoding
- Layer 9: erans residual entropy coding
- Layer 10: Archive assembly

9 component interdependencies mapped (density→GCCL, GCCL→OAC,
shear→S3C, S3C→FAMM, radius→GCCL, FAMM→PIST, PIST→erans,
OAC→receipts, shear→EigenBook).

10 compression gain sources quantified (geometric shear 15-30%,
topological skeleton 50-150MB vs 1GB, glyph kernels, S3C shells,
OAC speculation 2-5%, FAMM context 10-20%, PIST bundle 5-8%,
erans entropy, radius-ratio quantization, Gram dictionary MB→KB).

7 implementation phases defined (Foundation → Density Field →
GCCL-GEC Core → Shear/Eigen → OAC/Speculation → PIST/erans →
Integration → Benchmark).

14 keeper phrases. Core synthesis: density field = manifold,
glyph packets = navigators, shear matrix = map.
 2026-05-08 14:50:02 -05:00
Brandon Schneider
c145d6acef ingest: GCCL-GEC full compression architecture spec 
Geometric-Cognitive Compression Law / Glyph Eigen Codec.
9 archive components: D (decompressor), 𝔊 (GlyphBook), Χ (ChiralityBook),
Τ (TypeBook), 𝕌 (EigenBook), Γ (packet stream), Θ (params), Ε (residuals), R (audit).

7-field packet: γᵢ (glyph), χᵢ (chirality), κᵢ (coordinate), τᵢ (type),
UᵢΛᵢaᵢ (eigen descriptor), θᵢ (params), εᵢ (residual).

5 model families: Wiki structural, same-referent, arithmetic/date,
fractal/generator, eigenfield.

5 implementation phases: toy codec → arithmetic → same-referent →
eigen descriptors → PUA glyph acceleration.

7 stack integrations mapped (density field, S3C shells, OAC,
hypercube→rhomboid, FAMM, erans, radius-ratio).

10 keeper phrases. Core rule: glyph ≠ symbol; glyph = callable kernel.
 2026-05-08 14:50:01 -05:00
Brandon Schneider
7d6929aaa2 ingest: Density Field Encoding theory — beyond UTF-8 
Text as n-dimensional semantic density field rather than 1D byte sequence.
Topological features encode structure:
- Peaks = named entities/articles
- Ridges = hyperlinks/citations
- Saddles = topic transitions
- Vortices = cyclic refs/templates
- Voids = template structures
- Level sets = semantic granularity

6 stack integrations mapped (PIST perturbation, S3C shells,
OAC lazy manifestation, hypercube→rhomboid shear, FAMM temporal
pathing, erans residual entropy).

8 keeper phrases. Navigable compression paradigm.
 2026-05-08 14:50:01 -05:00
Allaun Silverfox
8215f6055f feat: add dimensional shell eigenvector resonance probe 2026-05-07 16:46:58 -05:00
Brandon Schneider
b7c4a0d367 ingest: erans enumerative rANS reference + AGENTS.md rules 1.10, 1.11 
erans (izabera): streamable single-pass rANS, enumerative coding bound.
NO LICENSE — algorithmic ideas captured as reference only, zero code copied.
5 key ideas: single-pass adaptive, enumerative bound, shrub DS,
streaming renorm, histogram rice coding.

AGENTS.md additions:
- 1.10: Never assume any instruction set (SIMD opportunistic, not structural)
- 1.11: Never incorporate unlicensed code (reference notes only, write from scratch)
 2026-05-07 02:18:29 -05:00
Brandon Schneider
ef841e7617 ingest: Hypercube → Hyper-Rhomboid Hutter Prize implications 
7 concrete changes to enwik compression:
- Shear pre-transform: 15-30% on structured regions (40% of enwik)
- S3C shell position encoding: 5-10% positional overhead reduction
- OAC speculative motifs: avoids 2-5% bloat, enables aggressive testing
- FAMM preshaped context: 10-20% context efficiency gain
- PIST n-D token encoding: 5-8% with cross-position probability sharing
- Gram matrix dictionary: MB → KB overhead
- Metric entropy coding: 10-15% entropy reduction in structured regions

The Big Fold: 4 separate Hutter components collapse into 1 shear matrix.
Estimated 12-22% overall compressed size reduction.
 2026-05-07 02:08:46 -05:00
Brandon Schneider
0f2c2b57f4 ingest: Hypercube → Hyper-Rhomboid composition theory 
Orthogonal tensor (hypercube) assumes independent axes.
Shear into parallelotope (hyper-rhomboid) models entangled dimensions.
The shear angle encodes correlation strength; the Gram matrix
of the shear IS the compression dictionary.

6 stack mappings:
- PIST n-D: Cartesian → Bundle → Radial = hypercube → rhomboid → collapsed
- Topological state machine: transition = shear on state tensor
- N-D Gene Hypothesis: gene = n-D rhomboid, 3D structure = projection shadow
- FAMM: preshaped delay = sheared time-domain rhomboid
- OAC: latent cavity in sheared rhomboid space
- Waveprobe: curvature = local shear angle of coordinate basis

3 compression interpretations + information gravity metric tensor
 2026-05-07 02:04:03 -05:00
Brandon Schneider
930cafaf8f ingest: Observer-Admissible Cavities theory — radius-ratio → Pidgen-hole → S3C/Spherion → OAC 
6 key concepts formalized:
- Radius-ratio rule → admissible motif classifier (CN3-8 thresholds)
- Pidgen-hole theory → typed hole + residual codec
- S3C shell coordinates → n=k²+a with throat/mirror/mass
- Spherion shaping → pyramid protrusions/voids as compression teeth
- S_n(n^n) → Matryoshka shell with latent combinatorial interior
- Observer-Admissible Cavities → touch-manifesting lazy holes

7 cross-references to existing modules, 4 new primitives identified.
7 keeper phrases preserved.
 2026-05-07 00:47:10 -05:00
Brandon Schneider
679945c1a0 ingest: dair-ai Agentic Engineering Wiki (51 tips, 7 categories) 
Cross-referenced against our prover orchestration layers:
- Plan-Execute-Verify-Replan ↔ L0-L3 pipeline
- Agents as specialists ↔ 11-agent swarm
- Guardrails ↔ ProverWatchdog
- Sandbox testing ↔ Virtual FPGA tests
- Trajectory-aware eval ↔ BFS audit trail

5 gaps identified, 4 strengths confirmed
 2026-05-07 00:27:02 -05:00
Brandon Schneider
70f9b10127 ingest: MS myelin glucose signaling article (2026-05-04) 
Brain glucose levels regulate OPC fate: high glucose → proliferation,
low glucose → maturation. Acetyl-CoA from glucose drives histone
acetylation for OPC division; ketone bodies substitute for myelin
synthesis. Ketogenic diet rescues myelin in ACLY-deficient mice.

Connects to: N-Dimensional Gene Hypothesis, PIST polymorphic shifter,
topological state machine, FAMM delay lines, waveprobe manifolds
 2026-05-06 23:53:20 -05:00
Brandon Schneider
453a366949 collapse: prover orchestration layers, FAMM verilator harness, swarm topological prober, spec sheets, virtual FPGA system tests, merge conflict resolution 
- Prover-Integrated Orchestration Layers (L0-L3): Goedel-Prover-V2 watchdog, BFS-Prover-V2 swarm consensus, bf4prover topology adaptation
- FAMM Verilator benchmark: uniform vs preshaped delay comparison (4.4x speedup)
- Swarm topological device prober: 11 agents probing traces, caps, delays, errors, vias, PDN
- Spec sheet puller: 10 components with key params and topological relevance
- Virtual FPGA system tests: 6/6 passed, 134K ops/s throughput
- Fixed merge conflicts in AI-Newton test_experiment.ipynb
 2026-05-06 23:42:01 -05:00
Brandon Schneider
31f953bada Consolidate research stack updates 2026-05-05 21:09:48 -05:00
Brandon Schneider
4eee4a07f6 initial: sovereign research stack (consolidated, weightless, and lfs-optimized) 2026-05-04 18:11:36 -05:00