COMSTR V2: Vibration Prediction and Design of FGM Trapezoidal Plates

This package supports the manuscript:
"A Unified Physics-Based Framework for Vibration Prediction and Design
Optimization of Functionally Graded Trapezoidal Thin Plates."

Contents
- python/COMSTR_V2_workflow.ipynb: main ROM, FFT-PINN, ANN-surrogate,
  sensitivity, and NSGA-II workflow.
- abaqus/: four Abaqus validation scripts.
- abaqus_results/: reported outputs of the revised Abaqus checks.
- matlab/: manuscript and Supplementary Information plotting scripts.
- data/: audited numerical outputs and plotting inputs.
- docs/: results-to-code map and MATLAB plotting guide.
- thumbnail.png: square website thumbnail.

Requirements
- Python 3.10+; install packages with: python -m pip install -r requirements.txt
- JupyterLab or Jupyter Notebook
- Abaqus/CAE for abaqus/ scripts
- MATLAB for matlab/ scripts

Quick start
1. From the package root, install the Python dependencies.
2. Change to python/.
3. Start Jupyter and open COMSTR_V2_workflow.ipynb.
4. Run the notebook. Complete execution is computationally intensive; audited
   data needed for figure/table reproduction are already in data/.

Abaqus scripts are run from abaqus/ with, for example:
abaqus cae noGUI=Abaqus_Pareto_design_validation_FGX_v2.py

The selected ANN is 4 x 256 ReLU and is evaluated on held-out ROM-generated
samples. Table 11 includes three limited Abaqus checks, not full-domain
ANN-Abaqus validation. See README.md and docs/RESULTS_TO_CODE_MAP.md for
details.
