added resampling mode and cleaned up notebook for resasmpling

README.md

@@ -205,7 +205,8 @@ Detailed usage of torch-harmonics, alongside helpful analysis provided in a seri
 5. [Conditioning of the Gramian](./notebooks/conditioning_sht.ipynb)
 6. [Solving the Helmholtz equation](./notebooks/helmholtz.ipynb)
 7. [Solving the shallow water equations](./notebooks/shallow_water_equations.ipynb)
-8. [Training Spherical Fourier Neural Operators](./notebooks/train_sfno.ipynb)
+8. [Training Spherical Fourier Neural Operators (SFNO)](./notebooks/train_sfno.ipynb)
+9. [Resampling signals on the sphere](./notebooks/resample_sphere.ipynb)
 
 ## Remarks on automatic mixed precision (AMP) support
 

torch_harmonics/resampling.py

@@ -48,13 +48,17 @@ def __init__(
         nlon_out: int,
         grid_in: Optional[str] = "equiangular",
         grid_out: Optional[str] = "equiangular",
+        mode: Optional[str] = "bilinear",
     ):
 
-        assert nlat_in <= nlat_out
-        assert nlon_in <= nlon_out
-
         super().__init__()
 
+        # currently only bilinear is supported
+        if mode == "bilinear":
+            self.mode = mode
+        else:
+            raise NotImplementedError(f"unknown interpolation mode {mode}")
+
         self.nlat_in, self.nlon_in = nlat_in, nlon_in
         self.nlat_out, self.nlon_out = nlat_out, nlon_out