torch-geopooling 1.0.0__cp311-cp311-macosx_11_0_arm64.whl → 1.0.1__cp311-cp311-macosx_11_0_arm64.whl

torch_geopooling/__init__.py

@@ -1,3 +1,3 @@
 from typing import Final
 
-__version__: Final[str] = "1.0.0"
+__version__: Final[str] = "1.0.1"

torch_geopooling/functional.py

@@ -271,10 +271,12 @@ class AdaptiveFunction(autograd.Function):
         weight, input = ctx.saved_tensors
         tiles, w = cls.sparse_ravel(weight)
 
-        grad_weight_out = cls.backward_impl(grad_values, tiles, w, input, ctx.exterior, *ctx.params)  # type: ignore
-        grad_weight_out = cls.sparse_unravel(tiles, grad_weight_out, size=weight.size())
+        grad_weight_dense = cls.backward_impl(
+            grad_values, tiles, w, input, ctx.exterior, *ctx.params
+        )  # type: ignore
+        grad_weight_sparse = cls.sparse_unravel(tiles, grad_weight_dense, size=weight.size())
 
-        return grad_weight_out.coalesce(), None, None, None, None
+        return grad_weight_sparse.coalesce(), None, None, None, None
 
     @classmethod
     def func(

torch_geopooling/functional_test.py

@@ -17,6 +17,7 @@ import pytest
 import torch
 
 from torch_geopooling.functional import (
+    AdaptiveFunction,
     adaptive_quad_pool2d,
     adaptive_avg_quad_pool2d,
     adaptive_max_quad_pool2d,
@@ -26,6 +27,20 @@ from torch_geopooling.functional import (
 )
 
 
+def test_adaptive_function_ravel() -> None:
+    size = (2, 2, 2, 1)
+    tiles = torch.tensor([[0, 0, 0], [1, 0, 0], [1, 0, 1], [1, 1, 0], [1, 1, 1]], dtype=torch.int64)
+
+    weight = torch.tensor([[1.0], [2.0], [3.0], [4.0], [5.0]], dtype=torch.float64)
+
+    sparse = AdaptiveFunction.sparse_unravel(tiles, weight, size=size)
+    torch.testing.assert_close(sparse.to_dense().to_sparse_coo(), sparse)
+
+    tiles_out, weight_out = AdaptiveFunction.sparse_ravel(sparse)
+    torch.testing.assert_close(tiles_out, tiles)
+    torch.testing.assert_close(weight_out, weight)
+
+
 @pytest.mark.parametrize(
     "function",
     [

torch_geopooling/nn.py

@@ -91,7 +91,12 @@ class _AdaptiveQuadPool(nn.Module):
     def initialize_parameters(self) -> None:
         # The weight for adaptive operation should be sparse, since training operation
         # results in a dynamic change of the underlying quadtree.
-        weight_size = (self.max_depth, 1 << self.max_depth, 1 << self.max_depth, self.feature_dim)
+        weight_size = (
+            self.max_depth + 1,
+            1 << self.max_depth,
+            1 << self.max_depth,
+            self.feature_dim,
+        )
         self.weight = nn.Parameter(torch.sparse_coo_tensor(size=weight_size, dtype=torch.float64))
 
     @property

torch_geopooling/nn_test.py

@@ -19,6 +19,7 @@ import pytest
 import torch
 from shapely.geometry import Polygon
 from torch import nn
+from torch.optim import SGD
 from torch.nn import L1Loss
 
 from torch_geopooling.nn import (
@@ -56,6 +57,43 @@ def test_adaptive_quad_pool2d_gradient(module_class: Type[nn.Module]) -> None:
     assert pool.weight.grad.sum().item() == pytest.approx(-1)
 
 
+def test_adaptive_quad_pool2d_optimize() -> None:
+    pool = AdaptiveQuadPool2d(1, (-180, -90, 360, 180), max_depth=1)
+
+    # Input coordinates are simply centers of the level-1 quads.
+    x_true = torch.tensor(
+        [[90.0, 45.0], [90.0, -45.0], [-90.0, -45.0], [-90.0, 45.0]], dtype=torch.float64
+    )
+    y_true = torch.tensor([[10.0], [20.0], [30.0], [40.0]], dtype=torch.float64)
+    y_tile = [[1, 1, 1], [1, 1, 0], [1, 0, 0], [1, 0, 1]]
+
+    optim = SGD(pool.parameters(), lr=0.01)
+    loss_fn = nn.L1Loss()
+
+    for i in range(20000):
+        optim.zero_grad()
+
+        y_pred = pool(x_true)
+        loss = loss_fn(y_pred, y_true)
+        loss.backward()
+
+        optim.step()
+
+    # Ensure that model converged with a small loss.
+    assert pytest.approx(0.0, abs=1e-1) == loss.detach().item()
+
+    # Ensure that weights that pooling operation learned are the same as in the
+    # target matrix (y_true).
+    weight = pool.weight.to_dense()
+
+    for i, tile in enumerate(y_tile):
+        z, x, y = tile
+        expect_weight = y_true[i].item()
+        actual_weight = weight[z, x, y].detach().item()
+
+        assert pytest.approx(expect_weight, abs=1e-1) == actual_weight, f"tile {tile} is wrong"
+
+
 @pytest.mark.parametrize(
     "module_class",
     [
@@ -83,3 +121,38 @@ def test_quad_pool2d_gradient(module_class: Type[nn.Module]) -> None:
 
     assert pool.weight.grad is not None
     assert pool.weight.grad.sum().item() == pytest.approx(-1)
+
+
+def test_quad_pool2d_optimize() -> None:
+    poly = Polygon([(-180, -90), (-180, 90), (180, 90), (180, -90)])
+    pool = QuadPool2d(1, poly, (-180, -90, 360, 180), max_depth=1)
+
+    x_true = torch.tensor(
+        [[90.0, 45.0], [90.0, -45.0], [-90.0, -45.0], [-90.0, 45.0]], dtype=torch.float64
+    )
+    y_true = torch.tensor([[10.0], [20.0], [30.0], [40.0]], dtype=torch.float64)
+    y_tile = [(1, 1, 1), (1, 1, 0), (1, 0, 0), (1, 0, 1)]
+
+    optim = SGD(pool.parameters(), lr=0.01)
+    loss_fn = nn.L1Loss()
+
+    for i in range(20000):
+        optim.zero_grad()
+
+        y_pred = pool(x_true)
+        loss = loss_fn(y_pred, y_true)
+        loss.backward()
+
+        optim.step()
+
+    # Ensure that model converged with a small loss.
+    assert pytest.approx(0.0, abs=1e-1) == loss.detach().item()
+
+    actual_tiles = {}
+    for i in range(pool.tiles.size(0)):
+        tile = tuple(pool.tiles[i].detach().tolist())
+        actual_tiles[tile] = pool.weight[i, 0].detach().item()
+
+    for tile, expect_weight in zip(y_tile, y_true[:, 0].tolist()):
+        actual_weight = actual_tiles[tile]
+        assert pytest.approx(expect_weight, abs=1e-1) == actual_weight, f"tile {tile} is wrong"

{torch_geopooling-1.0.0.dist-info → torch_geopooling-1.0.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: torch-geopooling
-Version: 1.0.0
+Version: 1.0.1
 Summary: The geospatial pooling modules for neural networks in PyTorch
 Author-email: Yakau Bubnou <girokompass@gmail.com>
 License: GNU GENERAL PUBLIC LICENSE
@@ -699,17 +699,21 @@ Classifier: Operating System :: MacOS
 Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: shapely >=2.0.0
-Requires-Dist: torch <2.3.0,>=2.2.0
+Requires-Dist: shapely>=2.0.0
+Requires-Dist: torch<2.3.0,>=2.2.0
 Provides-Extra: test
-Requires-Dist: pytest >=8.0.0 ; extra == 'test'
-Requires-Dist: numpy >=1.4.0 ; extra == 'test'
+Requires-Dist: pytest>=8.0.0; extra == "test"
+Requires-Dist: numpy>=1.4.0; extra == "test"
 
 # Torch Geopooling - The geospatial pooling library for PyTorch
 
 The Torch Geopooling library is an extension for PyTorch library that provide extra layers for
 building geospatial neural networks.
 
+Here is an example of how you can use modules from Torch Geopooling library to train neural
+networks predicting geospatial features:
+![example](docs/index.png)
+
 ## Installation
 
 The library is distributed as PyPI package, to install that package, execute the following

torch_geopooling-1.0.1.dist-info/RECORD

@@ -0,0 +1,17 @@
+torch_geopooling/_C.cpython-311-darwin.so,sha256=EDFwAndoKc5inME-p5rlD64GPJu4-6Fnn52KcEcqpeI,534584
+torch_geopooling/__init__.py,sha256=g3zAIxXO6CdrT0zoTlfM3K86LeTWJcgIsqPTqYmhBMQ,60
+torch_geopooling/functional.py,sha256=MuJfPW_RC4slgzWBO2ngQIBTzlZXeW7DBqV3ooIOSw4,12060
+torch_geopooling/functional_test.py,sha256=Sfv6bSe4DIB-y1xYVwavZoyIdu0bwARGdr8woOx7F7U,2997
+torch_geopooling/nn.py,sha256=_NnGB45dWTZy4MQokFfmJL3as-EPVU3liwOuAmWbv1Y,14219
+torch_geopooling/nn_test.py,sha256=zKoyK1f9SUJGhR8pt4wvocmVICoA6nPWPeUMvKKbwY8,4786
+torch_geopooling/return_types.py,sha256=QhIKf_qxpppzuZzUrT-VacfbvZ6zKmSRu-WZOqL8wuE,898
+torch_geopooling/tiling.py,sha256=f1g0Gf__5zIOafZzDkwO62t9OqMDMBEYVtM2HXB5X0Y,2553
+torch_geopooling/transforms.py,sha256=NHPBgx1L4-1745xNO_PvLuvbMH6DWi2rs1ViuRyBQAY,3635
+torch_geopooling/transforms_test.py,sha256=Cqp4VVObBVrc8HQ7uxkpe1EsQ5NxXrCI77BiUqngLXw,1196
+torch_geopooling/__bind__/python_module.cc,sha256=5K04tRT62XTVsZhY2EE4r4bBTz4mA4IwEkw5adPzo58,576
+torch_geopooling-1.0.1.dist-info/LICENSE,sha256=jLc4eyvG8hqxH4AdKAjK2DwXoP68qjHiGfq8eP5ubBI,35069
+torch_geopooling-1.0.1.dist-info/METADATA,sha256=CgOBHIw38QEgF4TIAVvnVYWz-RlPgW_60AbdJkK_J-E,44364
+torch_geopooling-1.0.1.dist-info/WHEEL,sha256=LFVzND6nAdWMS-norKkn42oL86bk-j1PiLvh1xzptX0,110
+torch_geopooling-1.0.1.dist-info/dependency_links.txt,sha256=JqLDcYHtEaQB51V72n3gAJvRd36bpoPk9qgTbot-Lx4,37
+torch_geopooling-1.0.1.dist-info/top_level.txt,sha256=3geTL2nsLvybdtr1psWIE6h63B1LuyIIyWWv0rDafTk,17
+torch_geopooling-1.0.1.dist-info/RECORD,,

{torch_geopooling-1.0.0.dist-info → torch_geopooling-1.0.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: bdist_wheel (0.43.0)
+Generator: bdist_wheel (0.44.0)
 Root-Is-Purelib: false
 Tag: cp311-cp311-macosx_11_0_arm64
 

torch_geopooling-1.0.0.dist-info/RECORD

@@ -1,17 +0,0 @@
-torch_geopooling/_C.cpython-311-darwin.so,sha256=fKlQs3WTHYmFfAPcdV8pbZ6e2qVykm4goZRt2HjmkRc,518504
-torch_geopooling/__init__.py,sha256=ryF0V7tLAeu0WwmadDc-gtjcPRpHBN9M1rDKYphqEYc,60
-torch_geopooling/functional.py,sha256=rZrh30K4IlxG3lGQ1_C8wrL7MKS9yIT-NSEkcM21A_k,12028
-torch_geopooling/functional_test.py,sha256=WZ7GePj1f8jp8cxNrnrDoHXnJeEeBUNVnFySPgEDl2I,2406
-torch_geopooling/nn.py,sha256=A_pozkqIfjhmOL7BDvTV2bJBv7nXD3Wi8nAlbl_rSEs,14156
-torch_geopooling/nn_test.py,sha256=4Aj3CcI8KG0parOyruAHZCCnC51CvSacTZ0AxTs16Gc,2359
-torch_geopooling/return_types.py,sha256=QhIKf_qxpppzuZzUrT-VacfbvZ6zKmSRu-WZOqL8wuE,898
-torch_geopooling/tiling.py,sha256=f1g0Gf__5zIOafZzDkwO62t9OqMDMBEYVtM2HXB5X0Y,2553
-torch_geopooling/transforms.py,sha256=NHPBgx1L4-1745xNO_PvLuvbMH6DWi2rs1ViuRyBQAY,3635
-torch_geopooling/transforms_test.py,sha256=Cqp4VVObBVrc8HQ7uxkpe1EsQ5NxXrCI77BiUqngLXw,1196
-torch_geopooling/__bind__/python_module.cc,sha256=5K04tRT62XTVsZhY2EE4r4bBTz4mA4IwEkw5adPzo58,576
-torch_geopooling-1.0.0.dist-info/LICENSE,sha256=jLc4eyvG8hqxH4AdKAjK2DwXoP68qjHiGfq8eP5ubBI,35069
-torch_geopooling-1.0.0.dist-info/METADATA,sha256=8IjAad92llXV6wN59x-x9GK-xj501ZQM17f1PUzIIss,44209
-torch_geopooling-1.0.0.dist-info/WHEEL,sha256=sieEctgmsyAnWfDYOiunmkigyyjGmYuUaApm_YItwoI,110
-torch_geopooling-1.0.0.dist-info/dependency_links.txt,sha256=JqLDcYHtEaQB51V72n3gAJvRd36bpoPk9qgTbot-Lx4,37
-torch_geopooling-1.0.0.dist-info/top_level.txt,sha256=3geTL2nsLvybdtr1psWIE6h63B1LuyIIyWWv0rDafTk,17
-torch_geopooling-1.0.0.dist-info/RECORD,,