implement autoregressive solver

Co-authored-by: GiovanniCanali <giovanni.canali98@yahoo.it>

fixed solver test

Remove old test

fix test data manager

Add pytest fixture for cleaning work directory

remove problem.input_pts

fix batch size

fix batch_size bug

Author: Davide-Miotti

docs/source/_rst/_code.rst

@@ -82,6 +82,8 @@ Solvers
     DeepEnsembleSupervisedSolver <solver/ensemble_solver/ensemble_supervised>
     ReducedOrderModelSolver <solver/supervised_solver/reduced_order_model.rst>
     GAROM <solver/garom.rst>
+    AutoregressiveSolverInterface <solver/autoregressive_solver/autoregressive_solver_interface.rst>
+    AutoregressiveSolver <solver/autoregressive_solver/autoregressive_solver.rst>
 
 
 Models

docs/source/_rst/solver/autoregressive_solver/autoregressive_solver.rst

@@ -0,0 +1,7 @@
+Autoregressive Solver 
+======================
+.. currentmodule:: pina.solver.autoregressive_solver.autoregressive_solver
+
+.. autoclass::  pina._src.solver.autoregressive_solver.autoregressive_solver.AutoregressiveSolver
+   :members:
+   :show-inheritance:

docs/source/_rst/solver/autoregressive_solver/autoregressive_solver_interface.rst

@@ -0,0 +1,7 @@
+Autoregressive Solver Interface
+=================================
+.. currentmodule:: pina.solver.autoregressive_solver.autoregressive_solver_interface
+
+.. autoclass::  pina._src.solver.autoregressive_solver.autoregressive_solver_interface.AutoregressiveSolverInterface
+   :members:
+   :show-inheritance:

pina/_src/data/creator.py

@@ -79,13 +79,16 @@ def _compute_batch_sizes(self, datasets):
         """
         batch_sizes = {}
         if self.batching_mode == "common_batch_size":
+
+            if self.batch_size is None:
+                batch_size = max(
+                    dataset.length for dataset in datasets.values()
+                )
+            else:
+                batch_size = self.batch_size
+
             for name in datasets.keys():
-                if self.batch_size is None:
-                    batch_sizes[name] = len(datasets[name])
-                else:
-                    batch_sizes[name] = min(
-                        self.batch_size, len(datasets[name])
-                    )
+                batch_sizes[name] = min(batch_size, len(datasets[name]))
             return batch_sizes
         if self.batching_mode == "proportional":
             return self._compute_proportional_batch_sizes(datasets)
@@ -168,8 +171,12 @@ def __call__(self, datasets):
         dataloaders = {}
         if self.batching_mode == "common_batch_size":
             max_len = max(len(dataset) for dataset in datasets.values())
+        print(batch_sizes)
         for name, dataset in datasets.items():
-            if self.batching_mode == "common_batch_size":
+            if (
+                self.batching_mode == "common_batch_size"
+                and dataset.length != batch_sizes[name]
+            ):
                 dataset.max_len = max_len
             dataloaders[name] = self.conditions[name].create_dataloader(
                 dataset=dataset,

pina/_src/problem/abstract_problem.py

@@ -43,21 +43,21 @@ def __init__(self):
                     self.domains[cond_name] = cond.domain
                     cond.domain = cond_name
 
-    #  back compatibility 0.1
-    @property
-    def input_pts(self):
-        """
-        Return a dictionary mapping condition names to their corresponding
-        input points. If some domains are not sampled, they will not be returned
-        and the corresponding condition will be empty.
-
-        :return: The input points of the problem.
-        :rtype: dict
-        """
-        to_return = {}
-        for cond_name, data in self.collected_data.items():
-            to_return[cond_name] = data["input"]
-        return to_return
+    # #  back compatibility 0.1
+    # @property
+    # def input_pts(self):
+    #     """
+    #     Return a dictionary mapping condition names to their corresponding
+    #     input points. If some domains are not sampled, they will not be returned
+    #     and the corresponding condition will be empty.
+
+    #     :return: The input points of the problem.
+    #     :rtype: dict
+    #     """
+    #     to_return = {}
+    #     for cond_name, data in self.collected_data.items():
+    #         to_return[cond_name] = data["input"]
+    #     return to_return
 
     @property
     def discretised_domains(self):