How gabrielecerioli/inverse_inference_weighted_spatio_functional_networks works
This is a research implementation for generating synthetic spatio-functional networks and estimating their physical/social mixing parameter, physical interaction scale, and latent coordinates. Users can generate benchmark networks, run exact inference (including optional profile-likelihood uncertainty), or estimate large-network parameters by fitting independent induced subnetworks and aggregating their results. The graph reflects README-described capabilities and sampled source; generator metric evaluation is documented, but some implementation details were not sampled.
Main entry points: exact_inference.py Exact inference · subsampled_inference.py Subsampled inference · spatiosocial_network_generator.py Network generator
Components 21
Network generation
- Network generator
spatiosocial_network_generator.py - Social GMM
spatiosocial_network_generator.py - Physical coordinates
spatiosocial_network_generator.py - Dual kernels
spatiosocial_network_generator.py - Balanced propensity
spatiosocial_network_generator.py - Link sampling
spatiosocial_network_generator.py - Synthetic network
spatiosocial_network_generator.py
Exact inference
- Exact inference
exact_inference.py - Distance and r₀
exact_inference.py - Spectral initialization
exact_inference.py - Poisson objective
exact_inference.py - Hybrid optimization
exact_inference.py - Profile confidence interval
exact_inference.py - Parameter estimates
exact_inference.py
Subsampled inference
- Subsampled inference
subsampled_inference.py - Subgraph sampling
subsampled_inference.py - Ensemble aggregation
subsampled_inference.py - Ensemble estimates
subsampled_inference.py
Latent evaluation
- Procrustes alignment
spatiosocial_network_generator.py - Metric deficit
spatiosocial_network_generator.py
Other components
- Research user
Connections 23
- Research user → Network generator: requests generation
- Network generator → Social GMM: samples latent positions
- Network generator → Physical coordinates: samples locations
- Network generator → Dual kernels: combines coordinates
- Dual kernels → Balanced propensity: forms raw weights
- Balanced propensity → Link sampling: sets expected intensity
- Link sampling → Synthetic network: samples links
- Research user → Exact inference: requests inference
- Exact inference → Distance and r₀: computes distances and r₀
- Exact inference → Spectral initialization: initializes latent positions
- Exact inference → Poisson objective: defines likelihood
- Exact inference → Hybrid optimization: optimizes parameters
- Hybrid optimization → Poisson objective: evaluates objective
- Exact inference → Profile confidence interval: optionally profiles lambda
- Exact inference → Parameter estimates: returns estimates
- Research user → Subsampled inference: requests ensemble fit
- Subsampled inference → Subgraph sampling: selects induced subgraphs
- Subgraph sampling → Subsampled inference: provides subgraph inputs
- Subsampled inference → Exact inference: fits each subgraph
- Subsampled inference → Ensemble aggregation: aggregates replica fits
- Ensemble aggregation → Ensemble estimates: returns means and errors
- Research user → Procrustes alignment: requests coordinate alignment
- Research user → Metric deficit: requests metric evaluation