---
title: Sadim Sky Explorer
emoji: 🚀
colorFrom: indigo
colorTo: blue
sdk: streamlit
sdk_version: 1.31.0
app_file: app.py
pinned: false
license: apache-2.0
---

# SADIM-77M: Stellar Radar & AI Anomaly Detector

## About the Project
SADIM-77M is an interactive analytical tool designed to investigate stars from the **ESA Gaia DR3** catalog. It combines real-time data retrieval (via DuckDB and ESA TAP API) with a custom deep learning model (Variational Autoencoder) to analyze stellar kinematics. The goal is to classify stars and instantly identify anomalies such as hypervelocity stars, unresolved binary systems, or data artifacts.

## How to Use
1. Obtain a valid **Gaia DR3 `SOURCE_ID`**.
2. Paste the ID into the search bar at the bottom left of the screen.
3. Click **SCAN**.
4. The system will retrieve the star's data, plot it on an interactive deep-space map (Aladin Lite), and display a detailed diagnostic report.

*Example ID to try:* `38655544960`

## Understanding the Scan Report
When you scan a star, a report pops up containing both physical calculations and AI metrics. Here is what each value means:

* **SOURCE_ID:** The unique numerical identifier for the star in the Gaia DR3 database.
* **DIST_LY:** The estimated distance to the star from Earth, measured in Light-Years.
* **PARALLAX:** The apparent shift of the star against the background (measured in milliarcseconds - mas), along with its error margin. Used to calculate distance.
* **VEL_KMS (TAN):** Tangential Velocity. The star's speed moving across our line of sight, measured in km/s. (Values > 600 km/s may trigger a Hypervelocity alert).
* **RUWE (ESA_API):** Renormalised Unit Weight Error. A quality metric for the Gaia data. A value around 1.0 is normal. Values > 1.4 indicate a poor fit, which often points to an unseen binary companion or data noise.
* **RADIAL_VEL:** The speed of the star moving directly toward or away from us (if measured by Gaia).
* **LOCAL_COUNT:** The number of neighboring stars fetched within the immediate spatial cone around the target.
* **MSE_DEV (AI Metric):** Mean Squared Error. Shows how much the star's kinematic profile deviates from the AI model's expected baseline.
* **KL_SURP (AI Metric):** Kullback-Leibler Divergence. Acts as a "surprise" metric for the AI; higher values mean the star's data distribution is highly unusual.
* **STATUS:** The final automated classification of the star based on strict physics and AI rules. Categories include:
  * `[NOMINAL (NORMAL)]`: Standard star, no anomalies detected.
  * `[DATA ARTIFACT: BAD ASTROMETRY (RUWE)]`: Likely a binary system or noisy observation.
  * `[PHYSICAL ANOMALY: HYPERVELOCITY]`: Star moving at extreme speeds.
  * `[AI GLOBAL ANOMALY]`: Flagged by the deep learning model due to unusual patterns.

---
**Data Sources:** ESA Gaia DR3, P/DSS2 Space Survey.