Features

PySEQM can take in data in either a .xyz file or in vector format.

species, coordinates = read_xyz(['../../data_one.xyz', '../../data_two.xyz', '../../data_three.xyz'])


species, coordinates = read_xyz([['../../data_one.xyz'], ['../../data_two.xyz'], ['../../data_three.xyz']])

species = torch.as_tensor([
                [8,6,1,1],
                [8,6,1,1],
                ],dtype=torch.int64, device=device)

coordinates = torch.tensor([
                [
                [0.00,    0.00,    0.00],
                [1.22,    0.00,    0.00],
                [1.82,    0.94,    0.00],
                [1.82,   -0.94,    0.00]
                ],
                [
                [0.00,    0.00,    0.00],
                [1.22,    0.00,    0.20],
                [1.82,    0.94,    0.00],
                [1.81,   -0.93,    -0.20]
                ],
                ],device=device)

In PySEQM, 0 padding must be used when working with batched molecules that have a different number of atoms.

[0.0,0.0,0.0]

PySEQM requires 64-bit precision for accurate and stable calculations. .. code-block:: python

dtype=torch.int64

PySEQM uses PyTorch, which allows you to run calculations on the GPU.

We set the device to either GPU or CPU.

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

Then we can set device=device so that calculations are run on the correct hardware.

([],device=device)