TT-DMD Online Solver

TT-DMD Online Solver is an interactive web-based application for spatiotemporal prediction using Dynamic Mode Decomposition (DMD) and Tensor-Train Dynamic Mode Decomposition (TT-DMD). The solver enables efficient analysis and forecasting of high-dimensional temporal datasets through advanced tensor-compression techniques.

Developed at the Institute of Structural Mechanics, Brno University of Technology (VUT), the software is part of the ERC Synergy Grant project FatResCon (Grant ID: 101167045).

Key Features

• Tensor-Train Dynamic Mode Decomposition (TT-DMD) for high-dimensional tensor data.

• Classical DMD implementation based on singular value decomposition (SVD).

• Interactive eigenvalue spectrum visualization in the complex plane.

• Growth-rate and frequency analysis of DMD modes.

• Relative L1 error evaluation for training and prediction periods.

• Support for NumPy (.npy), MATLAB (.mat), and CSV (.csv) data formats.

• Extended forecasting and result export capabilities.

Intended Applications

The solver is suitable for reduced-order modeling, time-series forecasting, spatiotemporal system identification, and data-driven analysis of engineering and scientific datasets, including infrastructure deterioration processes.

Related Publication

If you use this software in academic research, please cite:

Exponential Data Compression via Tensor-Train Dynamic Mode Decomposition for Predicting Concrete Deterioration.

Authors

• Yue Li (Brno University of Technology)

• Miroslav Vořechovský (Brno University of Technology)

License

MIT License © 2026 Yue Li, Miroslav Vořechovský