Skip to content

readme.md

Latest commit

聽

History

History
78 lines (54 loc) 路 2.45 KB

readme.md

File metadata and controls

78 lines (54 loc) 路 2.45 KB
Raw

Dr. NIMA - Nuclear Imaging with Multi-Agents

A Streamlit application for particle and nuclear physics research using a transparent multi-agent approach with Google's Gemini API.

Features

  • PDF document management and retrieval (RAG)
  • Physics tools integration and discovery
  • Python code execution and debugging
  • Multi-agent coordination for complex physics questions
  • Testing mode right now

Installation

  1. Clone the repository:
git clone https://github.com/StarNetLaboratory/dr_nima.git
cd dr_nima
  1. Install requirements:
pip install -r requirements.txt
  1. Place your PDF documents in the pdfs directory and Python tools in the tools directory.

Usage

  1. Run the Streamlit application:
streamlit run app.py

  1. Enter your Google API key in the sidebar.

  2. Use the interface to:

    • Upload PDF documents for the knowledge base
    • Ask questions about particle or nuclear physics
    • View the inner workings of the multi-agent system
    • See code execution results and generated visualizations

Multi-Agent System Architecture

Dr. NIMA uses a transparent multi-agent approach with three specialized agents:

  1. Manager Agent (馃) - Plans the overall approach, breaks down complex problems, and delegates tasks.
  2. Research Agent (馃摎) - Retrieves and analyzes information from the knowledge base.
  3. Code Agent (馃捇) - Writes, executes, and debugs Python code for simulations and visualizations.

The new transparent communications panel shows the interactions between these agents in real-time, providing insights into the problem-solving process.

Agent Transparency Features

  • Color-Coded Agent Logs: Each agent has a distinct color for easy identification

    • Manager Agent: Blue 馃
    • Research Agent: Green 馃摎
    • Code Agent: Purple 馃捇
    • System: Orange 馃攧

  • Real-Time Status Updates: See what each agent is currently working on

  • Filterable Log View: Filter the log view to focus on specific agents

  • Exportable Logs: Download logs in CSV or JSON format for analysis

Example Queries

  • "How are GPDs, QCFs and CFFs related to each other? Show me equations and plots."
  • "For a toy problem involving 1-D QCFs, can you generate events and plot the histrograms?"
  • "In 2D, what do the events from DIS, IDIS, SIDIS look like? Show me 3 2D histograms of the events."

Requirements

  • Python 3.8+
  • Streamlit
  • Google Gemini API key
  • Other dependencies listed in requirements.txt