An open API service indexing awesome lists of open source software.

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

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).