check train test (#111)

* Update setup.py
* scheduler for all torch versions
---------

Co-authored-by: Dario Coscia <dariocoscia@cli-10-110-0-208.WIFIeduroamSTUD.units.it>
Co-authored-by: Nicola Demo <demo.nicola@gmail.com>

pina/pinn.py

@@ -1,7 +1,11 @@
 """ Module for PINN """
 import torch
-import torch.optim.lr_scheduler as lrs
+try:
+    from torch.optim.lr_scheduler import LRScheduler  # torch >= 2.0
+except ImportError:
+    from torch.optim.lr_scheduler import _LRScheduler as LRScheduler # torch < 2.0
 
+from torch.optim.lr_scheduler import ConstantLR
 
 from .solver import SolverInterface
 from .label_tensor import LabelTensor
@@ -23,7 +27,7 @@ class PINN(SolverInterface):
                  loss = torch.nn.MSELoss(),
                  optimizer=torch.optim.Adam,
                  optimizer_kwargs={'lr' : 0.001},
-                 scheduler=lrs.ConstantLR,
+                 scheduler=ConstantLR,
                  scheduler_kwargs={"factor": 1, "total_iters": 0},
                  ):
         '''
@@ -46,7 +50,7 @@ class PINN(SolverInterface):
         # check consistency 
         check_consistency(optimizer, torch.optim.Optimizer, 'optimizer', subclass=True)
         check_consistency(optimizer_kwargs, dict, 'optimizer_kwargs')
-        check_consistency(scheduler, lrs.LRScheduler, 'scheduler', subclass=True)
+        check_consistency(scheduler, LRScheduler, 'scheduler', subclass=True)
         check_consistency(scheduler_kwargs, dict, 'scheduler_kwargs')
         check_consistency(loss, (LossInterface, _Loss), 'loss', subclass=False)
 
@@ -116,4 +120,4 @@ class PINN(SolverInterface):
         # TODO Fix the bug, tot_loss is a label tensor without labels
         # we need to pass it as a torch tensor to make everything work
         total_loss = sum(condition_losses)
-        return total_loss
+        return total_loss

setup.py

@@ -15,7 +15,7 @@ VERSION = meta['__version__']
 KEYWORDS = 'physics-informed neural-network'
 
 REQUIRED = [
-    'numpy', 'matplotlib', 'torch'
+    'numpy', 'matplotlib', 'torch', 'lightning'
 ]
 
 EXTRAS = {

tests/test_pinn.py

@@ -86,24 +86,24 @@ def test_constructor_extra_feats():
     model_extra_feats = FeedForward(len(poisson_problem.input_variables)+1,len(poisson_problem.output_variables))
     PINN(problem = poisson_problem, model=model_extra_feats, extra_features=extra_feats)
 
-def test_train():
+def test_train_cpu():
     poisson_problem = Poisson()
     boundaries = ['gamma1', 'gamma2', 'gamma3', 'gamma4']
     n = 10
     poisson_problem.discretise_domain(n, 'grid', locations=boundaries)
     poisson_problem.discretise_domain(n, 'grid', locations=['D'])
     pinn = PINN(problem = poisson_problem, model=model, extra_features=None, loss=LpLoss())
-    trainer = Trainer(solver=pinn, kwargs={'max_epochs' : 5})
+    trainer = Trainer(solver=pinn, kwargs={'max_epochs' : 5, 'accelerator':'cpu'})
     trainer.train()
 
-def test_train_extra_feats():
+def test_train_extra_feats_cpu():
     poisson_problem = Poisson()
     boundaries = ['gamma1', 'gamma2', 'gamma3', 'gamma4']
     n = 10
     poisson_problem.discretise_domain(n, 'grid', locations=boundaries)
     poisson_problem.discretise_domain(n, 'grid', locations=['D'])
     pinn = PINN(problem = poisson_problem, model=model_extra_feats, extra_features=extra_feats)
-    trainer = Trainer(solver=pinn, kwargs={'max_epochs' : 5})
+    trainer = Trainer(solver=pinn, kwargs={'max_epochs' : 5, 'accelerator':'cpu'})
     trainer.train()
 
 """