plurigrid/asi / ies/music-topos/.codex/skills/oriented-simplicial-networks
oriented-simplicial-networks
VerifiedOriented Simplicial Networks
Category: Phase 3 Core - Geometric Deep Learning
Status: Skeleton Implementation
Dependencies: categorical-composition, persistent-homology
Overview
Implements directional simplicial neural networks (Dir-SNNs) with asymmetric message passing operators, E(n)-equivariance constraints, and persistent homology tracking for topological feature learning.
Capabilities
- Directional Message Passing: Asymmetric operators respecting simplex orientation
- E(n)-Equivariance: Rotation/translation invariant representations
- Persistent Homology: Track topological features during training
- Simplicial Attention: Higher-order attention mechanisms on simplicial complexes
Core Components
-
Simplicial Complex Builder (
simplicial_complex.jl)- Construct oriented simplicial complexes from data
- Boundary operator computation
- Coboundary and Laplacian matrices
-
Dir-SNN Layers (
dirsnn_layers.jl)- Asymmetric message passing on simplices
- E(n)-equivariant convolutions
- Higher-order pooling operators
-
Persistent Homology Tracker (
persistent_homology.jl)- Compute persistence diagrams during forward pass
- Track birth/death of topological features
- Bottleneck/Wasserstein distance metrics
-
Training Loop (
train_dirsnn.jl)- Integration with Flux.jl
- Topologically-aware loss functions
- Gradient flow on simplicial manifolds
Integration Points
- Input from:
sheaf-theoretic-coordination(sheaf structures on simplicial complexes) - Output to:
categorical-composition(functorial network composition) - Coordinates with:
formal-verification-ai(verify topological invariants)
Usage
using OrientedSimplicialNetworks
# Build simplicial complex from point cloud
complex = SimplicialComplex(points, max_dimension=2)
# Create Dir-SNN model
model = DirSNN([
SimplicialConv(in_features=3, out_features=16, dimension=0),
SimplicialConv(in_features=16, out_features=32, dimension=1),
SimplicialPooling(dimension=1)
])
# Train with persistent homology tracking
train!(model, complex, labels; track_topology=true)
References
- Bodnar et al. "Weisfeiler and Lehman Go Cellular" (NeurIPS 2021)
- Hajij et al. "Topological Deep Learning" (Nature 2023)
- Carlsson "Topology and Data" (AMS 2009)
Implementation Status
- Core data structures
- Basic message passing
- Full E(n)-equivariance
- Persistent homology integration
- Benchmark suite
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 193k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 193k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 193k | Updated | Claude CodeCodex | Verified | Install |