TorchRDIT

https://github.com/yi-huang-1/torchrdit

Admin

TorchRDIT is a GPU-accelerated PyTorch framework for inverse design of meta-optics devices using...

Tokens:1,100

Snippets:9

Trust Score:5.4

License:GPL-3.0

Update:6 months ago

Show doc for...

Tokens:

Raw

### Install TorchRDIT via Pip

Source: https://github.com/yi-huang-1/torchrdit/blob/main/README.md

Standard installation command for the TorchRDIT package using the Python package manager.

```bash
pip install torchrdit
```

--------------------------------

### GDS Export/Import with TorchRdit

Source: https://github.com/yi-huang-1/torchrdit/blob/main/README.md

Demonstrates the comprehensive GDS (Generic Design System) file functionality, including exporting generated masks to GDS format and importing them back. This is useful for integrating TorchRdit designs with other GDS-based tools.

```python
from torchrdit.shapes import ShapeGenerator
from torchrdit.gds import mask_to_gds, gds_to_mask

# Create shape and export to GDS
shape_gen = ShapeGenerator(X, Y, grids)
mask = shape_gen.generate_circle_mask(center=(0, 0), radius=0.5)
mask_to_gds(mask, shape_gen.get_layout(), "DEVICE", "output.gds")

# Import from GDS
reconstructed = gds_to_mask("output.json", shape_gen)
```

--------------------------------

### Compute Tangent Field for Fast Fourier Factorization

Source: https://github.com/yi-huang-1/torchrdit/blob/main/README.md

This snippet demonstrates how to compute the tangent field for FFF using the compute_tangent_field function. It supports multiple schemes like POL, NORMAL, and JONES to ensure compatibility with various Fourier factorization workflows.

```python
from torchrdit.vector import compute_tangent_field

tx, ty = compute_tangent_field(
    mask.float(),
    XO=solver.XO,
    YO=solver.YO,
    lattice_t1=solver.lattice_t1,
    lattice_t2=solver.lattice_t2,
    harmonics=tuple(int(k) for k in solver.harmonics),
    scheme="POL",  # POL, NORMAL, JONES, or JONES_DIRECT
)
```

Tokens:

Raw