feat: Update environment.yml with PyTorch and add installation guide

- environment.yml: Added PyTorch with CUDA 12.1, PyG (torch-geometric),
  and TensorBoard for neural network training
- INSTALL_INSTRUCTIONS.md: Step-by-step guide for installing Miniconda
  and setting up the Atomizer environment

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

environment.yml

@@ -14,6 +14,8 @@
 name: atomizer
 channels:
   - pytorch
+  - nvidia
+  - pyg
   - conda-forge
   - defaults
 
@@ -41,7 +43,19 @@ dependencies:
   # FEA/Nastran Parsing
   # ============================================================================
   - h5py>=3.0.0
-  - pip
+
+  # ============================================================================
+  # Neural Network / AtomizerField
+  # ============================================================================
+  # PyTorch with CUDA support
+  - pytorch>=2.0.0
+  - pytorch-cuda=12.1
+
+  # PyTorch Geometric for Graph Neural Networks
+  - pyg>=2.3.0
+
+  # TensorBoard for training visualization
+  - tensorboard>=2.13.0
 
   # ============================================================================
   # Web Framework (Dashboard/API)
@@ -67,6 +81,7 @@ dependencies:
   # ============================================================================
   # pip-only packages (not available in conda)
   # ============================================================================
+  - pip
   - pip:
     # Nastran file parser (numpy<2 constraint is handled above)
     - pyNastran>=1.4.0,<1.5.0
@@ -79,11 +94,10 @@ dependencies:
 # ============================================================================
 #
 # 1. Verify installation:
-#    python -c "import optuna; import pyNastran; print('All imports OK')"
+#    python -c "import optuna; import pyNastran; import torch; print('All imports OK')"
 #
-# 2. For NX integration, ensure NX Python stubs are accessible:
+# 2. Check GPU:
+#    python -c "import torch; print(f'CUDA: {torch.cuda.is_available()}')"
+#
+# 3. For NX integration, ensure NX Python stubs are accessible:
 #    See config.py for NX_PYTHON_STUBS path
-#
-# 3. For AtomizerField neural surrogates, install PyTorch and torch-geometric separately:
-#    conda install pytorch pytorch-cuda=12.1 -c pytorch -c nvidia
-#    pip install torch-geometric tensorboard