|
|
--- |
|
|
license: mit |
|
|
--- |
|
|
|
|
|
**Data Description:** |
|
|
|
|
|
1. `gamma_5666_half.npy`: 5,666 heat sink design fields in [0,1] of shape (5666, 1, 256, 256). Please note these are the left sides of each design only, because there is symmetry about the vertical axis. |
|
|
Recommended loading technique in Python and NumPy: `X = np.load(args.dataset_path).reshape(-1, 1, 256, 256).astype(np.float32)`. This can easily be converted to a Tensor for training a machine learning model in PyTorch as well. |
|
|
|
|
|
2. `inp_paras_5666.npy`: The corresponding 5,666 input parameter vectors, which are 3 scalars per design in the following order: (0) Inlet velocity in the range [-0.095, -0.025], (1) Maximum fluid power dissipation in the range [50J1, 75J1] where J1 = 1.58e-7, and (2) Volume Fraction in the range [0.25, 0.7]. |
|
|
|
|
|
3. `dissP_5666.npy`: The corresponding measured fluid power dissipation for each sample as a factor of J1. |
|
|
|
|
|
4. `meanT_5666.npy`: The corresponding mean temperature across the domain in degrees C. |
