🚀 ANSA Engineering Problem Solver

An intelligent RAG-powered agent that generates complete engineering solutions using ANSA Python API with mathematical reasoning and problem-solving capabilities.

🎯 Overview

This system combines:

• Mathematical Analysis: Break down engineering problems into mathematical components
• ANSA API Intelligence: Retrieve and apply appropriate ANSA Python functions
• Complete Solution Generation: Create executable Python scripts for complex engineering tasks
• Quality-Driven Approach: Incorporate engineering best practices and validation

The system transforms from simple code generation to comprehensive engineering problem solving with mathematical rigor and ANSA API expertise.

✨ Key Features

🧮 Mathematical Problem Decomposition

• Analyzes engineering requirements mathematically
• Calculates optimal parameters (mesh sizing, element counts, etc.)
• Applies geometric and numerical methods
• Considers constraints and optimization criteria

🔧 ANSA API Integration

• Function-Specific Search: Find exact ANSA API functions needed
• Module-Based Organization: Search within specific ANSA modules (base, mesh, morph, etc.)
• Example-Driven Learning: Retrieve complete usage examples
• Workflow Generation: Create multi-step engineering processes

💡 Engineering Capabilities

• Mesh Generation: Structured/unstructured meshing with quality criteria
• Geometry Processing: CAD manipulation and transformation
• Material & Properties: Assignment and validation
• Boundary Conditions: Setup and optimization
• Morphing & Optimization: Shape optimization workflows
• Analysis Setup: Complete preprocessing pipelines
• Post-Processing: Results analysis and validation
• Metadata-Aware: Tracks source file, chunk index, and content type for traceability.

---

🏗️ Project Structure

src/
├── app.py                  # Streamlit Web Interface
├── ansa_rag/              # Main Package
│   ├── __init__.py
│   ├── chroma_db.py       # ChromaDB handler for insertion and retrieval
│   ├── cli.py             # CLI Entry points
│   └── database/          # Database Operations
│       ├── __init__.py
│       ├── db_insert.py   # Document processing and insertion
│       └── query_db.py    # Interactive querying utilities
├── chroma_db/             # Vector database storage
├── pyproject.toml         # Package configuration
├── requirements.txt       # Python dependencies
├── .env                   # Environment variables
└── README.md             # This file

---

🚀 Quick Start

Installation

1. Clone the repository

git clone <repository-url>
cd ANSA_Docs_RAG

2. Create virtual environment

python3 -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install the package

pip install -e .

4. Configure environment variables Create a .env file:

# Database
DB_NAME=ansa_docs_collection
EMBEDDING_MODEL=sentence-transformers/all-mpnet-base-v2

# API Keys (optional)
ANTHROPIC_API_KEY=your_key_here
OLLAMA_BASE_URL=http://localhost:11434

🎯 Usage

The package provides three CLI commands after installation:

1. Launch Web Interface

ansa-rag

• Opens Streamlit web interface at http://localhost:8501
• Interactive chat interface for engineering problems
• Support for multiple LLM backends (Ollama, Claude)
• Real-time documentation retrieval and code generation

2. Populate Database

ansa-db-insert

• Processes ANSA documentation from ansa_docs_html/
• Extracts and indexes API functions, code examples, and documentation
• Creates vector embeddings for semantic search
• One-time setup (run when adding new documentation)

3. Query Database

ansa-query

• Command-line interface for database queries
• Test documentation retrieval capabilities
• Debug search functionality
• Useful for development and testing

---

🔧 Configuration

LLM Backends

Ollama (Local)

• Install Ollama: curl -fsSL https://ollama.com/install.sh | sh
• Pull models: ollama pull qwen2.5-coder
• Start service: ollama serve

Claude (Cloud)

• Set ANTHROPIC_API_KEY in .env
• Requires API credits from Anthropic

Embedding Models

Supported embedding models (configure in .env):

• sentence-transformers/all-mpnet-base-v2 (default)
• BAAI/bge-large-en-v1.5
• intfloat/e5-large-v2
• nomic-ai/nomic-embed-text-v1.5

---

🎮 Example Usage

Web Interface

1. Launch: ansa-rag
2. Select LLM model (Ollama/Claude)
3. Choose engineering domain and complexity
4. Ask questions like:
   • "Create a hex mesh for a cylindrical geometry with quality criteria"
   • "Set up contact interfaces with friction coefficients"
   • "Generate morphing controls for shape optimization"

Command Line

# Populate database with documentation
ansa-db-insert

# Query specific topics
ansa-query
# Enter search terms like: "mesh generation quality"

---

🛠️ Development

Package Structure

• Modular design with clear separation of concerns
• CLI entry points for easy access to functionality
• Configurable backends for different LLM providers
• Extensible architecture for adding new features

Adding New Features

1. Extend src/ansa_rag/ modules
2. Update CLI commands in src/ansa_rag/cli.py
3. Add new dependencies to pyproject.toml
4. Test with pip install -e .

---

📋 Requirements

• Python 3.9+
• ChromaDB for vector storage
• Streamlit for web interface
• LangChain for LLM integration
• Sentence Transformers for embeddings
• Optional: Ollama (local) or Anthropic API (cloud)

---

🤝 Contributing

1. Fork the repository
2. Create feature branch: git checkout -b feature/new-feature
3. Make changes and test thoroughly
4. Submit pull request with detailed description

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.