LJ Fusion: Lennard-Jones Molecular Dynamics Simulator

A high-performance, GPU-accelerated molecular dynamics simulator for studying the Lennard-Jones system with real-time 3D visualization and comprehensive thermodynamic analysis.

---

Overview

LJ Fusion is a sophisticated computational chemistry application that simulates the behavior of particles interacting via the Lennard-Jones potential. The simulator leverages NVIDIA CUDA for GPU acceleration, enabling real-time simulation of systems with thousands of particles at ~60 fps.

The project demonstrates:

• CUDA GPU Computing: High-performance force calculations on NVIDIA GPUs
• Molecular Dynamics: Velocity Verlet integration with multiple ensemble supports
• Real-time Visualization: OpenGL 3D rendering with interactive controls
• Thermodynamic Analysis: Computation of thermodynamic properties and statistical distributions
• Multi-platform Support: Cross-platform deployment via CMake

Core Capabilities

Feature	Description
Ensembles	Microcanonical (NVE), Canonical (TVN) with thermostat
Boundary Conditions	Periodic, Hard-wall, and Open expansion modes
GPU Acceleration	CUDA-optimized force calculations for 10-100x speedup
Visualization	Real-time 3D particle rendering with OpenGL
Analysis Tools	RDF, velocity distributions, thermodynamic properties
CLI Tasks	Batch simulations for phase diagrams and fluctuation analysis

---

Visual Results

GUI Application Demo

The main Qt5-based GUI provides interactive control and real-time visualization:

Phase Diagram Visualization

The simulator can explore all major phases of the Lennard-Jones system:

Gaseous Phase (T* = 1.5, ρ* = 0.02)

Low density, high temperature regime showing diffuse particle distribution.

Mixed Phase (T* = 1, ρ* = 0.3)

Coexistence of gas and liquid phases.

Liquid Phase (T* = 1, ρ* = 0.85)

High density, low temperature showing dense particle packing.

---

Lennard-Jones Physics

The Lennard-Jones potential describes the interaction between two particles:

$$V_{LJ}(r) = 4\epsilon \left[ \left(\frac{\sigma}{r}\right)^{12} - \left(\frac{\sigma}{r}\right)^{6} \right]$$

Where:

• ε (epsilon): Depth of the potential well (energy scale)
• σ (sigma): Finite distance at which inter-particle potential is zero
• r: Distance between particles

Dimensionless Units

The simulator uses reduced units:

• T* = k_B T / ε (Reduced temperature)
• ρ* = ρ σ³ (Reduced density)
• t* = t √(mε/σ²) (Reduced time)

This allows study across different materials without unit conversion.

---

Quick Start

Prerequisites

• CMake 3.18+
• C++11 compatible compiler
• Qt5 (for GUI) - optional for CLI tools
• NVIDIA CUDA Toolkit (optional, for GPU acceleration)
• OpenGL development libraries

Linux/macOS Installation

git clone https://github.com/ellay21/LJFusion.git
cd LJFusion
mkdir build && cd build
cmake .. -DCMAKE_BUILD_TYPE=Release
make -j$(nproc)

Running the Application

GUI Application (Qt5 Required)

./src/gui/QtLennardJones

Command-Line Tools

Isotherm Simulation - Generate phase diagram data:

./src/tasks/run-isotherm <temperature> <density> <steps>

Fluctuations Analysis - Study system fluctuations:

./src/tasks/run-fluctuations <temperature> <density> <duration>

Semi-GCE Analysis - Grand canonical ensemble analysis:

./src/tasks/semiGCEfluctuations <parameters>

---

Project Structure

LJFusion/
├── CMakeLists.txt              # Root CMake configuration
├── src/
│   ├── library/                # Core MD library (CPU/GPU)
│   │   ├── MDSystem.h/cpp      # Main molecular dynamics engine
│   │   ├── MDSystem.cu         # CUDA GPU kernels
│   │   └── splinefunction.*    # Utility functions
│   ├── gui/                    # Qt5 GUI application
│   │   ├── main.cpp           # Application entry point
│   │   ├── mainwindow.*        # Main window UI
│   │   ├── glwidget.*          # OpenGL 3D viewport
│   │   └── constants.*         # UI constants
│   ├── tasks/                  # Command-line analysis tools
│   │   ├── run-isotherm/       # Isotherm simulation task
│   │   ├── run-fluctuations/   # Fluctuations analysis
│   │   └── semiGCEfluctuations/ # Semi-GCE analysis
│   └── extra/                  # Utility modules
├── thirdparty/                 # External dependencies
│   ├── QCustomPlot/            # Plot visualization
│   └── MersenneTwister/        # RNG implementation
├── input/                      # Sample input data files
├── .github/workflows/          # GitHub Actions CI/CD
└── LICENSE                     # GPLv3 License

---

Build Options

Custom CMake options:

cmake .. \
  -DCMAKE_BUILD_TYPE=Release      # Release or Debug
  -DCMAKE_CUDA_COMPILER=OFF       # Disable GPU acceleration
  -DCMAKE_PREFIX_PATH=/path/to/Qt5 # Custom Qt5 location

---

Performance

Typical performance (NVIDIA RTX 2080 Ti):

• Single precision (float): ~15 GFLOPS sustained
• Particle count: Up to 100,000 particles at 60 FPS
• Integration accuracy: Velocity Verlet, O(dt²)
• Speedup over CPU: 10-100x depending on system size

---

Algorithm Details

Velocity Verlet Integration

The simulator uses the Velocity Verlet algorithm for stable, reversible dynamics:

$$v(t + dt/2) = v(t) + \frac{dt}{2m}F(t)$$ $$r(t + dt) = r(t) + dt \cdot v(t + dt/2)$$ $$v(t + dt) = v(t + dt/2) + \frac{dt}{2m}F(t+dt)$$

Boundary Conditions

1. Periodic: Particles wrap around domain boundaries
2. Hard-wall: Elastic collisions with fixed walls
3. Open: Particles can leave the domain

Ensembles

• NVE (Microcanonical): Constant volume & energy
• NVT (Canonical): Constant volume & temperature with Berendsen thermostat

---

Testing & Validation

Sample data files in /input/ directory test various scenarios:

• N400.Tst1.4.isotherm - Low temperature isotherm
• N400.Tst1.708.rho0.05 - Gas phase data
• N400.Tst1.4.rho0.05 - Transition region

---

License

This project is licensed under the GNU General Public License v3.0. See LICENSE for details.

---

Contributing

Contributions are welcome! Areas for enhancement:

• Additional ensemble implementations (NPT, μVT)
• GPU optimization techniques
• Web-based visualization
• Performance profiling and optimization

---

Support

For issues, feature requests, or questions:

• Open an issue on GitHub Issues
• Check existing documentation in the project

---