https://github.com/abess-team/slide
[JASA] Reconstruct Ising Model with Global Optimality via SLIDE
https://github.com/abess-team/slide
binary-random-vector coupling ising-model pseudo-likelihood sparse-learning spin subset-selection
Last synced: 8 months ago
JSON representation
[JASA] Reconstruct Ising Model with Global Optimality via SLIDE
- Host: GitHub
- URL: https://github.com/abess-team/slide
- Owner: abess-team
- Created: 2025-06-05T00:09:51.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2025-10-01T10:57:58.000Z (8 months ago)
- Last Synced: 2025-10-01T12:36:30.563Z (8 months ago)
- Topics: binary-random-vector, coupling, ising-model, pseudo-likelihood, sparse-learning, spin, subset-selection
- Language: R
- Homepage:
- Size: 809 KB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Instructions for Reproducible Materials
## Organization
- **Bash script** (`batch.sh`): automate the execution of all numerical simulation studies
- **R scripts** (`.R`): implement baseline methods, evaluation metrics, simulation studies, real-data analysis of the congressional voting dataset
- **Data files** (`.csv`): for real-world data analysis. It is available at [Dropbox](https://www.dropbox.com/scl/fo/zbfrhxm60y8hhrzufhno2/AJzjVAZiJHK8AhrrBS6xxUw?rlkey=fpjf3h5awrki1cik5ypy5pqg9&st=8qwgpufi&dl=0).
## File Descriptions
### Main R scripts
- `simu_degree.R` — empirical sample complexity analysis with respect to the degree.
- `simu_beta.R` — empirical sample complexity analysis with respect to the maximum signal.
- `simu_high.R` — experiments for high-dimensional cases.
- `simu_p.R` — empirical sample complexity analysis with respect to the dimension.
- `simu_ws.R` — empirical sample complexity analysis with respect to the weakest signal.
- `DataAnalysis.R` — real-data analysis: data cleaning, estimation of the graphical structure among senators, and visualization.
#### Utility R scripts (automatically used by the main scripts)
- `simulation_main.R` — runs one method on a given simulated dataset.
- `method_implementation.R` — implementations of baseline methods (RPLE, RISE, logRISE, ELASSO, RLRF).
- `evaluation.R` — evaluation metrics (e.g., Frobenius norm, true positive rate).
## Reproducing Results
The scripts reproduce the results presented in the paper as follows:
- **Figure 1 and Table S1** → `simu_degree.R`
- **Figure 2 and Figure S1** → `simu_beta.R`
- **Figure 3 and Figure S2** → `simu_high.R`
- **Figure S3** → `simu_p.R`
- **Figure S4** → `simu_ws.R`
- **Figure 4** → `DataAnalysis.R`
The simplest procedure on reproduction:
1. Use the provided bash scripts (`batch.sh`) to execute the full set of simulation automatically.
2. Run `DataAnalysis.R` to reproduce the real-data analysis (Figure 4).