# Estimating Stellar Parameters from LAMOST Low-resolution
This repo contains the code, trained models, experimental data, and catalogs for our paper **Estimating Stellar Parameters from LAMOST Low-resolution**.

## Requirements

- tensorflow
- numpy
- pandas
- matplotlib
- sklearn
- jupyter

## Code directories

* Unzip `1_FITS_files_download_and_preprocessing.zip`.
* Unzip `2_BGANet_StarGRUNet.zip`.
* Unzip `3_Model uncertainty.zip`.
* Unzip `4_Comparison of StarGRUNet with GALAH DR3.zip`.
* Unzip `5_Observation Uncertainty.zip`.

## Experimental data

* Reference spectra data: 

  * Unzip `BR_Flux_Preprocessing_lamost_apogee_between_5_50.zip.001` and  unzip `BR_Flux_Preprocessing_lamost_apogee_above_50.zip.001` to `./1_FITS_files_download_and_preprocessing/spectra_after_processing/3sigma/` directory.
* Reference labels:

  * `./1_FITS_files_download_and_preprocessing/LABELS/between_5_50.npy`
  * `./1_FITS_files_download_and_preprocessing/LABELS/above_50.npy`
* Test spectra data: 
  * Unzip `X_test_above_50.zip.001` and unzip `X_test_between_5_50.zip.001` to `./data` directory.

##  Catalogs

* StarGRUNet catalog: unzip `StarGRUNet_catalog.zip.001`.
* Repeat observation catalog: unzip `dr8_v1.0_LRS_mec2.zip` to `/5_Observation Uncertainty/observation_catalogs/` directory.
* LAMOST DR8-APOGEE DR17 cross-matching catalog is in `./1_FITS_files_download_and_preprocessing` directory.

## Usage

* Please Training, testing, and validating new models:

  `./2_BGANet_StarGRUNet/1_BGANet_and_StarGRUNet_training_on_snrg_5_to_50_data.ipynb` `./2_BGANet_and_StarGRUNet_training_on_snrg_above_50_data.ipynb`

* More examples can be found in the other code directories.