| --- |
| license: mit |
| tags: |
| - numpy |
| - Non-interacting random-walkers |
| - Dean-Kawasaki |
| configs: |
| - config_name: default |
| data_files: |
| - split: test |
| path: |
| - "preview.csv" |
| --- |
| |
| # Dataset Card |
|
|
| ## Dataset Description |
|
|
| This dataset consists of number density for 5120 realizations of a 100 cell one-dimensional system subjected to an external potential. |
| The system is modeled with 4 different methods, namely, random-walker particles (`Random-Walkers.npy`), Dean-Kawasaki model (`Dean-Kawasaki.npy`), |
| Markovian ML model (`Markovian-ML.npy`), and non-Markovian ML model (`Non-Markovian-ML.npy`). Simulation time step is 3.0e-6 and cell size is 1.0e-2. |
| The data is saved at 50 time step intervals. |
|
|
| Each NumPy binary file (`.npy`) contains multidimensional arrays of shape `(200, 100, 5120)`. |
| The ordering of data is time, spatial positions, and independent realizations. |
|
|
| The Dataset Viewer is configured against `preview.csv`, which provides one metadata row per `.npy` file. The `.npy` files remain the canonical dataset artifacts. |
|
|
| ## Array Structure |
|
|
| | Axis | Dimension | Description | |
| |------|-----------|-------------| |
| | 0 | 200 | Time steps | |
| | 1 | 100 | System (spatial) direction | |
| | 2 | 5120 | Realizations (ensemble members) | |
|
|
| ## Usage |
|
|
| ```python |
| import numpy as np |
| |
| data = np.load("file_name.npy") |
| print(data.shape) # (200, 100, 5120) |
| |
| n_time, n_space, n_realizations = data.shape |
| dt_saved = 1.5e-4 |
| dx = 0.01 |
| ``` |
