Sovereign 8s.is mirror of standardgalactic/TASC-DMD
- MATLAB 100%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
| docs | ||
| examples | ||
| src | ||
| LICENSE | ||
| README.md | ||
TASC-DMD
MATLAB implementation of Time-Augmented, Space-Contracted Dynamic Mode Decomposition (TASC-DMD).
Paper: [TASC DMD Paper — title and citation to be added after publication]
Repository Structure
TASC-DMD/
├── src/
│ ├── DMD_exact.m Core DMD solver with time-delay embedding
│ ├── fbDMD.m Forward-Backward DMD (noise-robust)
│ └── nls_rhs.m ODE RHS for the NLS benchmark
├── examples/
│ ├── generate_NLS_data.m Generate NLS benchmark data [paper]
│ ├── Schrodinger_equ_TASCDMD.m NLS benchmark — reproduces paper results [paper]
│ ├── generate_AdvDiff_data.m Generate advection-diffusion data [NOT in paper]
│ └── AdvDiff_TASCDMD.m Advection-diffusion benchmark [NOT in paper]
├── docs/
│ └── index.html GitHub Pages project site
└── README.md
Quick Start
addpath('src');
cd examples
% --- NLS benchmark (reproduces paper results) ---
run generate_NLS_data.m
run Schrodinger_equ_TASCDMD.m
% --- Advection-Diffusion (NOT in paper — additional example) ---
run generate_AdvDiff_data.m
run AdvDiff_TASCDMD.m
Benchmarks
NLS — Nonlinear Schrödinger Equation (paper)
Two-soliton solution. TASC-DMD vs Exact DMD vs fbDMD under 20% noise.
Advection-Diffusion (NOT in paper — additional example)
Linear PDE u_t + U0*u_x = nu*u_xx with analytically known eigenvalues:
omega_n = -nu*k_n^2 - i*U0*k_n
These lie on the parabola Re(ω) = -(ν/U₀²)·Im(ω)², which is overlaid on the
eigenvalue plot as ground truth. Tested at 50% noise. Cv reported for all methods.
Key Parameters
| Parameter | Symbol | Description | NLS | AdvDiff |
|---|---|---|---|---|
d |
d | Time-delay levels (Hankel depth) | 11 | 11 |
r0 |
r₀ | Spatial basis rank per delay | 11 | 11 |
r1 |
r₁ | Final DMD rank | 9 | 9 |
noise_level |
— | Gaussian noise fraction | 0.20 | 0.50 |
Citation
[To be populated after publication]
License
MIT License. See LICENSE for details.