flux-sci-lang 0.1.0

FLUX Scientific Computing Language - Domain-Specific Language for PDEs, CFD, and Computational Physics

FLUX Scientific Computing Language

Domain-Specific Language for PDEs, CFD, and Computational Physics

FLUX is a high-performance DSL designed for scientific computing, specializing in partial differential equations, computational fluid dynamics, electromagnetic simulations, and finite element analysis.

Features

Core Scientific Computing Features

• Native PDE Syntax: Write equations in mathematical notation (∂u/∂t = ∇²u)
• Advanced Mesh Support: Structured, unstructured, and adaptive mesh refinement
• Multi-Physics: CFD, electromagnetics, structural analysis, heat transfer
• GPU Acceleration: CUDA kernel generation and parallel execution
• Modern Solvers: Finite element, finite volume, and spectral methods
• Code Generation: Compile to C++, CUDA, Python for maximum performance

Implemented Features (v0.1.0)

• ✅ PDE-specific lexer with Unicode math operators (∇, ∂, ×, ⊗)
• ✅ Parser for PDE definitions and boundary conditions
• ✅ Mesh generation (structured grids, unstructured, AMR)
• ✅ Code generation backends (Python, C++, CUDA)
• ✅ Scientific templates (heat equation, Navier-Stokes, Maxwell)
• ✅ GPU kernel generation for parallel computing
• ✅ Field operations and mathematical operators
• ✅ Boundary condition specification

Installation

# Clone the repository
git clone https://github.com/yourusername/flux-lang.git
cd flux-lang

# Install dependencies
pip install numpy scipy matplotlib

# Optional: CUDA toolkit for GPU acceleration
# Download from https://developer.nvidia.com/cuda-downloads

Quick Start

Compile a FLUX Program

Compile FLUX scientific code to your target backend:

# Compile to Python (default)
python flux_scientific.py examples/heat_equation.flux

# Compile to CUDA for GPU acceleration  
python flux_scientific.py examples/gpu_accelerated_cfd.flux -b cuda

# Compile to C++ for performance
python flux_scientific.py examples/navier_stokes_cavity.flux -b cpp

Interactive Scientific Computing

Start the interactive scientific environment:

python flux_scientific.py -i

Example session:

flux-sci> mesh StructuredGrid 50 50
Created StructuredGrid with 2601 nodes, 2500 cells

flux-sci> compile examples/heat_equation.flux python
Compiling examples/heat_equation.flux to python...
Generated code written to output/generated.py
Compilation successful!

Run Benchmarks

Test FLUX with included scientific benchmarks:

python flux_scientific.py --benchmark

Language Examples

Hello World

function main() {
    print("Hello, FLUX World!")
}

main()

Variables and Types

// Immutable variable
let pi = 3.14159

// Mutable variable
var counter = 0
counter = counter + 1

// Constant
const MAX_SIZE = 100

// Type annotations (optional)
let name: string = "FLUX"
let age: int = 1

Functions

function add(a: int, b: int) -> int {
    return a + b
}

// Async function (syntax supported, async not yet implemented)
async function fetch_data(url: string) -> string {
    // Implementation
}

Vectors and AI Operations

// Create vectors
let embedding1 = embed("artificial intelligence")
let embedding2 = embed("machine learning")

// Semantic similarity using ~~ operator
let similarity = embedding1 ~~ embedding2
print("Similarity: " + str(similarity))

// Vector operations
let v1 = vector(1.0, 2.0, 3.0)
let v2 = vector(4.0, 5.0, 6.0)
let dot_product = v1 @ v2  // Matrix multiplication operator

Tensor Operations

// Create tensors
let matrix = tensor([[1, 2], [3, 4]])
let weights = tensor([[0.1, 0.2], [0.3, 0.4]])

// Matrix multiplication
let result = matrix @ weights

// Special tensors
let zeros_3x3 = zeros(3, 3)
let ones_2x4 = ones(2, 4)
let random_5x5 = random(5, 5)

Quantum Computing (Basic)

// Create a qubit
let q = qubit()

// Apply quantum gates
q = hadamard(q)  // Put in superposition

// Quantum circuits (syntax supported)
quantum circuit bell_pair() {
    classical {
        // Classical preprocessing
    }
    quantum {
        // Quantum operations
    }
    classical {
        // Classical postprocessing
    }
}

Control Flow

// If statement
if x > 0 {
    print("Positive")
} else if x < 0 {
    print("Negative")
} else {
    print("Zero")
}

// While loop
while condition {
    // Loop body
}

// For loop (syntax planned)
for item in collection {
    process(item)
}

// Match expression (syntax supported)
match value {
    1 => print("One"),
    2 => print("Two"),
    _ => print("Other")
}

Examples Directory

The examples/ directory contains several demonstration programs:

• hello_world.flux - Basic hello world and string operations
• fibonacci.flux - Recursive and iterative Fibonacci implementations
• vectors_ai.flux - Vector operations and semantic similarity
• tensors_ml.flux - Tensor operations and ML concepts
• quantum_basic.flux - Basic quantum computing demonstrations

Architecture

flux-lang/
├── src/
│   ├── __init__.py      # Package initialization
│   ├── lexer.py         # Tokenization
│   ├── parser.py        # AST generation
│   └── interpreter.py   # Execution engine
├── examples/            # Example FLUX programs
├── flux.py             # CLI and REPL
└── README.md           # This file

Development Status

FLUX is currently in early development (v0.1.0). The following features are planned:

Near-term Goals

• Complete type system implementation
• Async/await execution
• Import/module system
• Standard library expansion
• Error handling improvements
• Debugger support

Long-term Goals

• MLIR backend for compilation
• WebAssembly target
• GPU acceleration
• Distributed computing primitives
• Full quantum circuit simulation
• Package manager
• IDE plugins (VSCode, etc.)

Contributing

Contributions are welcome! Areas where help is needed:

1. Standard Library: Implementing built-in functions and types
2. Quantum Operations: Expanding quantum computing support
3. AI Integration: Connecting to real LLM APIs
4. Optimization: Performance improvements
5. Documentation: Tutorials and examples
6. Testing: Unit tests and integration tests

License

FLUX is open-source software. License details to be determined.

Acknowledgments

FLUX is inspired by:

• Shopify Liquid (templating philosophy)
• Python (syntax and simplicity)
• Rust (memory safety concepts)
• Julia (scientific computing)
• Q# (quantum computing)
• Mojo (AI compilation)

---

FLUX: Write once, run everywhere, understand everything.