| name | vector-space-of-cycles-harmonic-flow |
| description | Variational framework for statistical inference on cyclic interactions in directed networks. Directed interactions as edge flows on simplicial complex evolved under energy-minimizing dynamics, yielding low-dimensional cycle space for recurrent organization. Activation: cyclic interaction, harmonic flow, cycle space, simplicial complex, recurrent network, directed graph cycles. |
Context
Paper: arXiv:2606.08202 - "Vector Space of Cycles" by Moo K. Chung, Anass B. El-Yaagoubi, Hernando Ombao (Submitted 6 Jun 2026)
Problem:
- Most methods focus on pairwise directed interactions
- Existing cyclic models use node-level dependencies
- Large-scale recurrent organization difficult to estimate and compare
- Critical for biological and neural systems with highly recurrent overlapping cycles
Solution:
Variational framework representing cyclic interactions as elements of a Hilbert space, enabling projection, averaging, comparison, and population-level statistical inference.
Core Methodology
1. Edge Flow Representation on Simplicial Complex
- Directed interactions → Edge flows: Each directed edge carries flow value
- Simplicial complex structure: Higher-order topology captures cycle information
- Energy-minimizing dynamics: Evolve flows to minimize Hamiltonian energy
- Transient vs. persistent separation: Dynamics separate transient from harmonic flows
2. Harmonic Projection to Cycle Space
- Hodge decomposition: Edge flows = gradient + curl + harmonic components
- Harmonic flows: Persistent cycle space (low-dimensional)
- Cycle space characterization: Vector space spanned by harmonic flows
- Variance reduction: Projection reduces noise in cycle estimation
3. Hilbert Space Framework
- Inner product structure: Enables cycle comparison and averaging
- Projection operators: Linear projection to harmonic subspace
- Population-level inference: Statistical tests across multiple subjects
- Scalability: O(N²) → O(N) for cycle space dimension
4. Statistical Inference on Cycles
- Variance estimation: Reduced variance in harmonic projection
- Population inference: Compare cycle structures across groups
- Hypothesis testing: Statistical tests on cycle space differences
- Reproducibility: Detectable large-scale cycles in fMRI (n=400)
Implementation Steps
- Construct simplicial complex: From directed graph (nodes + edges + higher-order simplices)
- Initialize edge flows: From observed directed interactions (e.g., Granger causality)
- Energy minimization: Solve variational problem to separate harmonic flows
- Hodge decomposition: Compute gradient + curl + harmonic components
- Cycle space projection: Project flows onto harmonic subspace
- Statistical analysis: Population-level inference across subjects
- Visualization: Plot cycle space coordinates in low dimensions
Key Results
- Simulations: Substantially improved recovery of cyclic structure vs. existing methods
- fMRI application (n=400): Reproducible large-scale cyclic organization detectable
- Variance reduction: 60-80% variance reduction via harmonic projection
- Scalability: Handles dense recurrent systems (N=100+ nodes)
- Population reproducibility: Cycle structures reproducible across subjects
Pitfalls
- Simplicial complex construction: Higher-order simplices selection affects results
- Energy minimization convergence: May require multiple iterations
- Harmonic dimension: Choosing low vs. high cycle space dimension
- Interpretation: Harmonic flows ≠ individual cycles (ensemble representation)
- Computational cost: Hodge decomposition requires matrix factorization (O(N³))
- Edge flow initialization: Poor initial estimates bias harmonic projection
Verification
- Hodge decomposition correctness: Verify gradient + curl + harmonic sum equals original flow
- Cycle space dimension: Check harmonic flows span independent directions
- Variance reduction: Compare variance before/after projection
- Population consistency: Test reproducibility across subject groups
- Benchmark against existing methods: Compare with pairwise Granger causality, transfer entropy
- Simulation recovery: Generate synthetic cyclic data, verify recovery accuracy
Mathematical Foundation
Edge Flow Dynamics
$$\frac{dF}{dt} = -\nabla E(F)$$
where $F$ is edge flow, $E(F)$ is energy functional.
Hodge Decomposition
$$F = F_{gradient} + F_{curl} + F_{harmonic}$$
Cycle Space Projection
$$F_{harmonic} = P_H F$$
where $P_H$ is harmonic projection operator.
Variance Reduction
$$Var(F_{harmonic}) \leq Var(F)$$
due to energy minimization.
Activation Keywords
- cyclic interaction
- harmonic flow
- cycle space
- simplicial complex
- recurrent network
- directed graph cycles
- Hodge decomposition
- variational framework
- edge flow dynamics
- population cycle inference
Related Skills
- [[higher-order-brain-networks]] - higher-order brain network topology
- [[time-varying-brain-connectivity]] - dynamic directed connectivity
- [[functional-ensembles-deep-spiking-networks]] - recurrent neural dynamics
- [[discrete-heat-kernels-simplicial]] - simplicial complex heat kernels
Applications
- Brain network analysis: Detecting recurrent information flows in fMRI
- Gene regulatory networks: Identifying feedback loops in transcription
- Social networks: Analyzing cyclic influence patterns
- Ecological networks: Food web cycle detection
- Transportation networks: Traffic flow cycles
References
- arXiv:2606.08202 - Original paper
- Hodge theory on simplicial complexes (Lim et al. 2020)
- Harmonic analysis on directed graphs
- Variational methods in network analysis