# Installation

This guide covers installing GenesisLab and its dependencies.

## System Requirements

### Hardware
- **GPU**: NVIDIA GPU with CUDA support (RTX 2070 or better recommended)
- **RAM**: 16GB minimum, 32GB+ recommended for large-scale training
- **Storage**: 10GB for software and assets

### Software
- **OS**: Linux (Ubuntu 20.04+), macOS, or Windows
- **Python**: 3.8 or higher
- **CUDA**: 11.8 or higher (for GPU support)

## Installation Steps

### 1. Install PyTorch

First, install PyTorch with CUDA support:

```bash
# For CUDA 11.8
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

# For CUDA 12.1
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
```

Verify installation:
```python
import torch
print(f"PyTorch version: {torch.__version__}")
print(f"CUDA available: {torch.cuda.is_available()}")
print(f"CUDA version: {torch.version.cuda}")
```

### 2. Install Genesis

Install the Genesis physics engine:

```bash
pip install genesis-world
```

Verify installation:
```python
import genesis as gs
gs.init(backend=gs.gpu)
print(f"Genesis version: {gs.__version__}")
```

### 3. Install GenesisLab

Clone the repository and install:

```bash
# Clone repository
git clone https://github.com/yourusername/genesislab.git
cd genesislab

# Install in editable mode
cd source/genesislab
pip install -e .
```

Verify installation:
```python
import genesislab
print(f"GenesisLab installed successfully!")
```

### 4. Install RL Libraries (Optional)

For training with reinforcement learning:

**Stable Baselines3**:
```bash
pip install stable-baselines3[extra]
```

**RSL RL**:
```bash
pip install rsl-rl
```

**Other dependencies**:
```bash
pip install gymnasium
pip install tensorboard  # For logging
pip install wandb  # For experiment tracking (optional)
```

## Verification

Run the verification script to ensure everything is installed correctly:

```python
import torch
import genesis as gs
import gymnasium as gym
import genesislab.envs

# Check PyTorch
assert torch.cuda.is_available(), "CUDA not available!"
print(f"✓ PyTorch {torch.__version__} with CUDA {torch.version.cuda}")

# Check Genesis
gs.init(backend=gs.gpu)
print(f"✓ Genesis {gs.__version__}")

# Check GenesisLab
env = gym.make("GenesisLab-Go2-Flat-v0", num_envs=4)
obs, info = env.reset()
print(f"✓ GenesisLab environment created successfully")
print(f"  Observation shape: {obs.shape}")
env.close()

print("\n✓ All checks passed! You're ready to go!")
```

## Troubleshooting

### CUDA not available

**Problem**: `torch.cuda.is_available()` returns `False`

**Solutions**:
1. Check NVIDIA drivers are installed: `nvidia-smi`
2. Install correct PyTorch version for your CUDA version
3. Verify CUDA toolkit is installed: `nvcc --version`

### Genesis import error

**Problem**: `ImportError: cannot import name 'gs'`

**Solutions**:
1. Reinstall Genesis: `pip install --upgrade genesis-world`
2. Check Python version (requires 3.8+)
3. Try in a fresh virtual environment

### GenesisLab import error

**Problem**: `ImportError: No module named 'genesislab'`

**Solutions**:
1. Ensure you installed from correct directory: `source/genesislab`
2. Use editable install: `pip install -e .`
3. Check you're in the correct Python environment

### Memory errors

**Problem**: `RuntimeError: CUDA out of memory`

**Solutions**:
1. Reduce `num_envs` parameter
2. Reduce model size
3. Use lower-resolution sensors
4. Close other GPU applications

### Slow simulation

**Problem**: Simulation runs slowly

**Solutions**:
1. Ensure CUDA is being used (check `gs.init(backend=gs.gpu)`)
2. Increase `num_envs` for better GPU utilization
3. Use fake sensors instead of Genesis sensors
4. Check no other processes are using GPU

## Docker Installation (Optional)

For a containerized setup:

```dockerfile
FROM pytorch/pytorch:2.1.0-cuda11.8-cudnn8-runtime

# Install Genesis
RUN pip install genesis-world

# Install GenesisLab
COPY . /workspace/genesislab
WORKDIR /workspace/genesislab/source/genesislab
RUN pip install -e .

# Install RL libraries
RUN pip install stable-baselines3[extra] gymnasium

CMD ["/bin/bash"]
```

Build and run:
```bash
docker build -t genesislab .
docker run --gpus all -it genesislab
```

## Virtual Environment Setup

We recommend using virtual environments:

**Using venv**:
```bash
python3 -m venv genesislab-env
source genesislab-env/bin/activate  # Linux/Mac
# or
genesislab-env\Scripts\activate  # Windows
```

**Using conda**:
```bash
conda create -n genesislab python=3.10
conda activate genesislab
```

## Development Installation

For development work on GenesisLab:

```bash
# Clone repository
git clone https://github.com/yourusername/genesislab.git
cd genesislab

# Install in editable mode with dev dependencies
cd source/genesislab
pip install -e ".[dev]"

# Install pre-commit hooks
pre-commit install
```

## Updating GenesisLab

To update to the latest version:

```bash
cd genesislab
git pull origin main
cd source/genesislab
pip install -e . --upgrade
```

## Uninstallation

To uninstall:

```bash
pip uninstall genesislab
pip uninstall genesis-world
```

## Next Steps

- Run your [first environment](first_environment.md)
- Learn [basic concepts](basic_concepts.md)
- Check out [tutorials](../tutorials/index.md)