Fix bug in Collector with Graph data (#456)

* Fix bug in Collector with Graph data
* Add comments in DataModule class and bug fix in collate

pina/collector.py

@@ -1,3 +1,7 @@
+"""
+# TODO
+"""
+from .graph import Graph
 from .utils import check_consistency
 
 
@@ -52,6 +56,8 @@ class Collector:
                 # get data
                 keys = condition.__slots__
                 values = [getattr(condition, name) for name in keys]
+                values = [value.data if isinstance(
+                    value, Graph) else value for value in values]
                 self.data_collections[condition_name] = dict(zip(keys, values))
                 # condition now is ready
                 self._is_conditions_ready[condition_name] = True

pina/data/data_module.py

@@ -2,11 +2,11 @@ import logging
 import warnings
 from lightning.pytorch import LightningDataModule
 import torch
-from ..label_tensor import LabelTensor
-from torch.utils.data import DataLoader, BatchSampler, SequentialSampler, \
-    RandomSampler
+from torch_geometric.data import Data
+from torch.utils.data import DataLoader, SequentialSampler, RandomSampler
 from torch.utils.data.distributed import DistributedSampler
-from .dataset import PinaDatasetFactory
+from ..label_tensor import LabelTensor
+from .dataset import PinaDatasetFactory, PinaTensorDataset
 from ..collector import Collector
 
 
@@ -61,6 +61,10 @@ class Collator:
             max_conditions_lengths is None else (
                 self._collate_standard_dataloader)
         self.dataset = dataset
+        if isinstance(self.dataset, PinaTensorDataset):
+            self._collate = self._collate_tensor_dataset
+        else:
+            self._collate = self._collate_graph_dataset
 
     def _collate_custom_dataloader(self, batch):
         return self.dataset.fetch_from_idx_list(batch)
@@ -73,7 +77,6 @@ class Collator:
         if isinstance(batch, dict):
             return batch
         conditions_names = batch[0].keys()
-
         # Condition names
         for condition_name in conditions_names:
             single_cond_dict = {}
@@ -82,16 +85,28 @@ class Collator:
                 data_list = [batch[idx][condition_name][arg] for idx in range(
                     min(len(batch),
                         self.max_conditions_lengths[condition_name]))]
-                if isinstance(data_list[0], LabelTensor):
-                    single_cond_dict[arg] = LabelTensor.stack(data_list)
-                elif isinstance(data_list[0], torch.Tensor):
-                    single_cond_dict[arg] = torch.stack(data_list)
-                else:
-                    raise NotImplementedError(
-                        f"Data type {type(data_list[0])} not supported")
+                single_cond_dict[arg] = self._collate(data_list)
+
             batch_dict[condition_name] = single_cond_dict
         return batch_dict
 
+    @staticmethod
+    def _collate_tensor_dataset(data_list):
+        if isinstance(data_list[0], LabelTensor):
+            return LabelTensor.stack(data_list)
+        if isinstance(data_list[0], torch.Tensor):
+            return torch.stack(data_list)
+        raise RuntimeError("Data must be Tensors or LabelTensor ")
+
+    def _collate_graph_dataset(self, data_list):
+        if isinstance(data_list[0], LabelTensor):
+            return LabelTensor.cat(data_list)
+        if isinstance(data_list[0], torch.Tensor):
+            return torch.cat(data_list)
+        if isinstance(data_list[0], Data):
+            return self.dataset.create_graph_batch(data_list)
+        raise RuntimeError("Data must be Tensors or LabelTensor or pyG Data")
+
     def __call__(self, batch):
         return self.callable_function(batch)
 
@@ -125,7 +140,7 @@ class PinaDataModule(LightningDataModule):
                  batch_size=None,
                  shuffle=True,
                  repeat=False,
-                 automatic_batching=False,
+                 automatic_batching=None,
                  num_workers=0,
                  pin_memory=False,
                  ):
@@ -158,15 +173,35 @@ class PinaDataModule(LightningDataModule):
         logging.debug('Start initialization of Pina DataModule')
         logging.info('Start initialization of Pina DataModule')
         super().__init__()
-        self.automatic_batching = automatic_batching
+
+        # Store fixed attributes
         self.batch_size = batch_size
         self.shuffle = shuffle
         self.repeat = repeat
+        self.automatic_batching = automatic_batching
+        if batch_size is None and num_workers != 0:
+            warnings.warn(
+                "Setting num_workers when batch_size is None has no effect on "
+                "the DataLoading process.")
+            self.num_workers = 0
+        else:
+            self.num_workers = num_workers
+        if batch_size is None and pin_memory:
+            warnings.warn("Setting pin_memory to True has no effect when "
+                          "batch_size is None.")
+            self.pin_memory = False
+        else:
+            self.pin_memory = pin_memory
+
+        # Collect data
+        collector = Collector(problem)
+        collector.store_fixed_data()
+        collector.store_sample_domains()
 
         # Check if the splits are correct
         self._check_slit_sizes(train_size, test_size, val_size, predict_size)
 
-        # Begin Data splitting
+        # Split input data into subsets
         splits_dict = {}
         if train_size > 0:
             splits_dict['train'] = train_size
@@ -188,19 +223,6 @@ class PinaDataModule(LightningDataModule):
             self.predict_dataset = None
         else:
             self.predict_dataloader = super().predict_dataloader
-
-        collector = Collector(problem)
-        collector.store_fixed_data()
-        collector.store_sample_domains()
-        if batch_size is None and num_workers != 0:
-            warnings.warn(
-                "Setting num_workers when batch_size is None has no effect on "
-                "the DataLoading process.")
-        if batch_size is None and pin_memory:
-            warnings.warn("Setting pin_memory to True has no effect when "
-                          "batch_size is None.")
-        self.num_workers = num_workers
-        self.pin_memory = pin_memory
         self.collector_splits = self._create_splits(collector, splits_dict)
         self.transfer_batch_to_device = self._transfer_batch_to_device
 
@@ -316,10 +338,10 @@ class PinaDataModule(LightningDataModule):
         if self.batch_size is not None:
             sampler = PinaSampler(dataset, shuffle)
             if self.automatic_batching:
-                collate = Collator(self.find_max_conditions_lengths(split))
-
+                collate = Collator(self.find_max_conditions_lengths(split),
+                                   dataset=dataset)
             else:
-                collate = Collator(None, dataset)
+                collate = Collator(None, dataset=dataset)
             return DataLoader(dataset, self.batch_size,
                               collate_fn=collate, sampler=sampler,
                               num_workers=self.num_workers)

pina/data/dataset.py

@@ -1,10 +1,12 @@
 """
 This module provide basic data management functionalities
 """
+import functools
 import torch
 from torch.utils.data import Dataset
 from abc import abstractmethod
-from torch_geometric.data import Batch
+from torch_geometric.data import Batch, Data
+from pina import LabelTensor
 
 
 class PinaDatasetFactory:
@@ -62,7 +64,7 @@ class PinaTensorDataset(PinaDataset):
         if automatic_batching:
             self._getitem_func = self._getitem_int
         else:
-            self._getitem_func = self._getitem_list
+            self._getitem_func = self._getitem_dummy
 
     def _getitem_int(self, idx):
         return {
@@ -82,7 +84,7 @@ class PinaTensorDataset(PinaDataset):
         return to_return_dict
 
     @staticmethod
-    def _getitem_list(idx):
+    def _getitem_dummy(idx):
         return idx
 
     def get_all_data(self):
@@ -102,15 +104,56 @@ class PinaTensorDataset(PinaDataset):
         }
 
 
+class PinaBatch(Batch):
+    """
+    Add extract function to torch_geometric Batch object
+    """
+    def __init__(self):
+
+        super().__init__(self)
+
+    def extract(self, labels):
+        """
+        Perform extraction of labels on node features (x)
+
+        :param labels: Labels to extract 
+        :type labels: list[str] | tuple[str] | str 
+        :return: Batch object with extraction performed on x
+        :rtype: PinaBatch
+        """
+        self.x = self.x.extract(labels)
+        return self
+
+
 class PinaGraphDataset(PinaDataset):
 
     def __init__(self, conditions_dict, max_conditions_lengths,
                  automatic_batching):
         super().__init__(conditions_dict, max_conditions_lengths)
+        self.in_labels = {}
+        self.out_labels = None
         if automatic_batching:
             self._getitem_func = self._getitem_int
         else:
-            self._getitem_func = self._getitem_list
+            self._getitem_func = self._getitem_dummy
+
+        ex_data = conditions_dict[list(conditions_dict.keys())[
+            0]]['input_points'][0]
+        for name, attr in ex_data.items():
+            if isinstance(attr, LabelTensor):
+                self.in_labels[name] = attr.stored_labels
+        ex_data = conditions_dict[list(conditions_dict.keys())[
+            0]]['output_points'][0]
+        if isinstance(ex_data, LabelTensor):
+            self.out_labels = ex_data.labels
+
+        self._create_graph_batch_from_list = self._labelise_batch(
+            self._base_create_graph_batch_from_list) if self.in_labels \
+            else self._base_create_graph_batch_from_list
+
+        self._create_output_batch = self._labelise_tensor(
+            self._base_create_output_batch) if self.out_labels is not None \
+            else self._base_create_output_batch
 
     def fetch_from_idx_list(self, idx):
         to_return_dict = {}
@@ -119,17 +162,24 @@ class PinaGraphDataset(PinaDataset):
             condition_len = self.conditions_length[condition]
             if self.length > condition_len:
                 cond_idx = [idx % condition_len for idx in cond_idx]
-            to_return_dict[condition] = {k: Batch.from_data_list([
-                v[i] for i in cond_idx])
+            to_return_dict[condition] = {
+                k: self._create_graph_batch_from_list([v[i] for i in idx])
                 if isinstance(v, list)
-                                         else v[
-                                             cond_idx].reshape(
-                                                 -1, *v[cond_idx].shape[2:])
+                else self._create_output_batch(v[idx])
                 for k, v in data.items()
             }
+
         return to_return_dict
 
-    def _getitem_list(self, idx):
+    def _base_create_graph_batch_from_list(self, data):
+        batch = PinaBatch.from_data_list(data)
+        return batch
+
+    def _base_create_output_batch(self, data):
+        out = data.reshape(-1, *data.shape[2:])
+        return out
+
+    def _getitem_dummy(self, idx):
         return idx
 
     def _getitem_int(self, idx):
@@ -144,3 +194,31 @@ class PinaGraphDataset(PinaDataset):
 
     def __getitem__(self, idx):
         return self._getitem_func(idx)
+
+    def _labelise_batch(self, func):
+        @functools.wraps(func)
+        def wrapper(*args, **kwargs):
+            batch = func(*args, **kwargs)
+            for k, v in self.in_labels.items():
+                tmp = batch[k]
+                tmp.labels = v
+                batch[k] = tmp
+            return batch
+        return wrapper
+
+    def _labelise_tensor(self, func):
+        @functools.wraps(func)
+        def wrapper(*args, **kwargs):
+            out = func(*args, **kwargs)
+            if isinstance(out, LabelTensor):
+                out.labels = self.out_labels
+            return out
+        return wrapper
+
+    def create_graph_batch(self, data):
+        """
+        # TODO
+        """
+        if isinstance(data[0], Data):
+            return self._create_graph_batch_from_list(data)
+        return self._create_output_batch(data)

pina/graph.py

@@ -108,16 +108,14 @@ class Graph:
                                                     x)
 
         # Perform the graph construction
-        self._build_graph_list(x, pos, edge_index, edge_attr, additional_params)
+        self._build_graph_list(
+            x, pos, edge_index, edge_attr, additional_params)
 
     def _build_graph_list(self, x, pos, edge_index, edge_attr,
                           additional_params):
         for i, (x_, pos_, edge_index_) in enumerate(zip(x, pos, edge_index)):
-            if isinstance(x_, LabelTensor):
-                x_ = x_.tensor
             add_params_local = {k: v[i] for k, v in additional_params.items()}
             if edge_attr is not None:
-
                 self.data.append(Data(x=x_, pos=pos_, edge_index=edge_index_,
                                       edge_attr=edge_attr[i],
                                       **add_params_local))
@@ -127,7 +125,8 @@ class Graph:
 
     @staticmethod
     def _build_edge_attr(x, pos, edge_index):
-        distance = torch.abs(pos[edge_index[0]] - pos[edge_index[1]])
+        distance = torch.abs(pos[edge_index[0]] -
+                             pos[edge_index[1]]).as_subclass(torch.Tensor)
         return distance
 
     @staticmethod
@@ -165,7 +164,8 @@ class Graph:
         # If edge_index is a 3D tensor, we split it into a list of 2D tensors
         if edge_index is not None:
             if isinstance(edge_index, torch.Tensor) and edge_index.ndim == 3:
-                edge_index = [edge_index[i] for i in range(edge_index.shape[0])]
+                edge_index = [edge_index[i]
+                              for i in range(edge_index.shape[0])]
             elif not (isinstance(edge_index, list) and all(
                     t.ndim == 2 for t in edge_index)) and not (
                     isinstance(edge_index,
@@ -258,6 +258,8 @@ class RadiusGraph(Graph):
         """
         dist = torch.cdist(points, points, p=2)
         edge_index = torch.nonzero(dist <= r, as_tuple=False).t()
+        if isinstance(edge_index, LabelTensor):
+            edge_index = edge_index.tensor
         return edge_index
 
 
@@ -293,4 +295,6 @@ class KNNGraph(Graph):
         row = torch.arange(points.size(0)).repeat_interleave(k)
         col = knn_indices.flatten()
         edge_index = torch.stack([row, col], dim=0)
+        if isinstance(edge_index, LabelTensor):
+            edge_index = edge_index.tensor
         return edge_index

pina/trainer.py

@@ -105,9 +105,9 @@ class Trainer(lightning.pytorch.Trainer):
         # checking compilation and automatic batching
         if compile is None or sys.platform == "win32":
             compile = False
-        if automatic_batching is None:
-            automatic_batching = False
 
+        self.automatic_batching = automatic_batching if automatic_batching \
+            is not None else False
         # set attributes
         self.compile = compile
         self.solver = solver

tests/test_data/test_datamodule.py

@@ -13,10 +13,10 @@ from torch.utils.data import DataLoader
 input_tensor = torch.rand((100, 10))
 output_tensor = torch.rand((100, 2))
 
-x = torch.rand((100, 50 , 10))
-pos = torch.rand((100, 50 , 2))
+x = torch.rand((100, 50, 10))
+pos = torch.rand((100, 50, 2))
 input_graph = RadiusGraph(x, pos, r=.1, build_edge_attr=True)
-output_graph = torch.rand((100, 50 , 10))
+output_graph = torch.rand((100, 50, 10))
 
 
 @pytest.mark.parametrize(
@@ -30,6 +30,7 @@ def test_constructor(input_, output_):
     problem = SupervisedProblem(input_=input_, output_=output_)
     PinaDataModule(problem)
 
+
 @pytest.mark.parametrize(
     "input_, output_",
     [
@@ -46,14 +47,15 @@ def test_constructor(input_, output_):
 )
 def test_setup_train(input_, output_, train_size, val_size, test_size):
     problem = SupervisedProblem(input_=input_, output_=output_)
-    dm = PinaDataModule(problem, train_size=train_size, val_size=val_size, test_size=test_size)
+    dm = PinaDataModule(problem, train_size=train_size,
+                        val_size=val_size, test_size=test_size)
     dm.setup()
     assert hasattr(dm, "train_dataset")
     if isinstance(input_, torch.Tensor):
         assert isinstance(dm.train_dataset, PinaTensorDataset)
     else:
         assert isinstance(dm.train_dataset, PinaGraphDataset)
-    #assert len(dm.train_dataset) == int(len(input_) * train_size)
+    # assert len(dm.train_dataset) == int(len(input_) * train_size)
     if test_size > 0:
         assert hasattr(dm, "test_dataset")
         assert dm.test_dataset is None
@@ -64,7 +66,8 @@ def test_setup_train(input_, output_, train_size, val_size, test_size):
         assert isinstance(dm.val_dataset, PinaTensorDataset)
     else:
         assert isinstance(dm.val_dataset, PinaGraphDataset)
-    #assert len(dm.val_dataset) == int(len(input_) * val_size)
+    # assert len(dm.val_dataset) == int(len(input_) * val_size)
+
 
 @pytest.mark.parametrize(
     "input_, output_",
@@ -82,7 +85,8 @@ def test_setup_train(input_, output_, train_size, val_size, test_size):
 )
 def test_setup_test(input_, output_, train_size, val_size, test_size):
     problem = SupervisedProblem(input_=input_, output_=output_)
-    dm = PinaDataModule(problem, train_size=train_size, val_size=val_size, test_size=test_size)
+    dm = PinaDataModule(problem, train_size=train_size,
+                        val_size=val_size, test_size=test_size)
     dm.setup(stage='test')
     if train_size > 0:
         assert hasattr(dm, "train_dataset")
@@ -100,7 +104,8 @@ def test_setup_test(input_, output_, train_size, val_size, test_size):
         assert isinstance(dm.test_dataset, PinaTensorDataset)
     else:
         assert isinstance(dm.test_dataset, PinaGraphDataset)
-    #assert len(dm.test_dataset) == int(len(input_) * test_size)
+    # assert len(dm.test_dataset) == int(len(input_) * test_size)
+
 
 @pytest.mark.parametrize(
     "input_, output_",
@@ -112,7 +117,8 @@ def test_setup_test(input_, output_, train_size, val_size, test_size):
 def test_dummy_dataloader(input_, output_):
     problem = SupervisedProblem(input_=input_, output_=output_)
     solver = SupervisedSolver(problem=problem, model=torch.nn.Linear(10, 10))
-    trainer = Trainer(solver, batch_size=None, train_size=.7, val_size=.3, test_size=0.)
+    trainer = Trainer(solver, batch_size=None, train_size=.7,
+                      val_size=.3, test_size=0.)
     dm = trainer.data_module
     dm.setup()
     dm.trainer = trainer
@@ -140,6 +146,7 @@ def test_dummy_dataloader(input_, output_):
         assert isinstance(data[0][1]['input_points'], torch.Tensor)
     assert isinstance(data[0][1]['output_points'], torch.Tensor)
 
+
 @pytest.mark.parametrize(
     "input_, output_",
     [
@@ -147,10 +154,17 @@ def test_dummy_dataloader(input_, output_):
         (input_graph, output_graph)
     ]
 )
-def test_dataloader(input_, output_):
+@pytest.mark.parametrize(
+    "automatic_batching",
+    [
+        True, False
+    ]
+)
+def test_dataloader(input_, output_, automatic_batching):
     problem = SupervisedProblem(input_=input_, output_=output_)
     solver = SupervisedSolver(problem=problem, model=torch.nn.Linear(10, 10))
-    trainer = Trainer(solver, batch_size=10, train_size=.7, val_size=.3, test_size=0.)
+    trainer = Trainer(solver, batch_size=10, train_size=.7, val_size=.3,
+                      test_size=0., automatic_batching=automatic_batching)
     dm = trainer.data_module
     dm.setup()
     dm.trainer = trainer
@@ -176,3 +190,67 @@ def test_dataloader(input_, output_):
         assert isinstance(data['data']['input_points'], torch.Tensor)
     assert isinstance(data['data']['output_points'], torch.Tensor)
 
+from pina import LabelTensor
+
+input_tensor = LabelTensor(torch.rand((100, 3)), ['u', 'v', 'w'])
+output_tensor = LabelTensor(torch.rand((100, 3)), ['u', 'v', 'w'])
+
+x = LabelTensor(torch.rand((100, 50, 3)), ['u', 'v', 'w'])
+pos = LabelTensor(torch.rand((100, 50, 2)), ['x', 'y'])
+input_graph = RadiusGraph(x, pos, r=.1, build_edge_attr=True)
+output_graph = LabelTensor(torch.rand((100, 50, 3)), ['u', 'v', 'w'])
+
+@pytest.mark.parametrize(
+    "input_, output_",
+    [
+        (input_tensor, output_tensor),
+        (input_graph, output_graph)
+    ]
+)
+@pytest.mark.parametrize(
+    "automatic_batching",
+    [
+        True, False
+    ]
+)
+def test_dataloader_labels(input_, output_, automatic_batching):
+    problem = SupervisedProblem(input_=input_, output_=output_)
+    solver = SupervisedSolver(problem=problem, model=torch.nn.Linear(10, 10))
+    trainer = Trainer(solver, batch_size=10, train_size=.7, val_size=.3,
+                      test_size=0., automatic_batching=automatic_batching)
+    dm = trainer.data_module
+    dm.setup()
+    dm.trainer = trainer
+    dataloader = dm.train_dataloader()
+    assert isinstance(dataloader, DataLoader)
+    assert len(dataloader) == 7
+    data = next(iter(dataloader))
+    assert isinstance(data, dict)
+    if isinstance(input_, RadiusGraph):
+        assert isinstance(data['data']['input_points'], Batch)
+        assert isinstance(data['data']['input_points'].x, LabelTensor)
+        assert data['data']['input_points'].x.labels == ['u', 'v', 'w']
+        assert data['data']['input_points'].pos.labels == ['x', 'y']
+    else: 
+        assert isinstance(data['data']['input_points'], LabelTensor)
+        assert data['data']['input_points'].labels == ['u', 'v', 'w']
+    assert isinstance(data['data']['output_points'], LabelTensor)
+    assert data['data']['output_points'].labels == ['u', 'v', 'w']
+
+    dataloader = dm.val_dataloader()
+    assert isinstance(dataloader, DataLoader)
+    assert len(dataloader) == 3
+    data = next(iter(dataloader))
+    assert isinstance(data, dict)
+    if isinstance(input_, RadiusGraph):
+        assert isinstance(data['data']['input_points'], Batch)
+        assert isinstance(data['data']['input_points'].x, LabelTensor)
+        assert data['data']['input_points'].x.labels == ['u', 'v', 'w']
+        assert data['data']['input_points'].pos.labels == ['x', 'y']
+    else:
+        assert isinstance(data['data']['input_points'], torch.Tensor)
+        assert isinstance(data['data']['input_points'], LabelTensor)
+        assert data['data']['input_points'].labels == ['u', 'v', 'w']
+    assert isinstance(data['data']['output_points'], torch.Tensor)
+    assert data['data']['output_points'].labels == ['u', 'v', 'w']
+test_dataloader_labels(input_graph, output_graph, True)