Merge pull request #36 from praksharma/development

basic domain created

DeepINN/__init__.py

@@ -13,4 +13,5 @@
 from . import utils
 from . import constraint
 
+from . import domain
 

DeepINN/constraint/templates.py

@@ -1,8 +1,18 @@
+from .gradients import Jacobian
 import torch
 
-def burgers_pde(X, y):
+# TODO
+def test_function(X, y):
     """
-    The burgers equation.
+    x: array_like, N x D_in
+    y: array_like, N x D_out
+    """
+    return x * np.sin(5 * x)
+def Laplace(X, y):
+    """
+    The laplace equation.
+
+    u_xx + u_yy
     """
     dy_x = dde.grad.jacobian(y, x, i=0, j=0)
     dy_t = dde.grad.jacobian(y, x, i=0, j=1)

DeepINN/domain/__init__.py

@@ -0,0 +1 @@
+from .standard import Generic

DeepINN/domain/base.py

@@ -0,0 +1,23 @@
+import abc
+
+class Dataset(abc.ABC):
+    """
+    Dataset base class
+    """
+    def sample_collocation_points(self):
+        # returns the collocation points as tensor
+        raise NotImplementedError(f"{self.__class__.__name__} must implement the sample_collocation_points() method.")
+
+    def sample_collocation_labels(self):
+        # returns the collocation points using the above function.
+        # also returns the collocation points labels i.e. the residual of the PDE as tensor.
+        raise NotImplementedError(f"{self.__class__.__name__} must implement the sample_collocation_labels() method.")
+
+    def sample_boundary_points(self):
+        # returns a list of boundary points on each boundary as tensor
+        raise NotImplementedError(f"{self.__class__.__name__} must implement the sample_boundary_points method.")
+
+    def sample_boundary_labels(self):
+        # returns the list of boundary points on each boundary using the above function.
+        # returns a list of tensor boundary conditions on each boundary.
+        raise NotImplementedError(f"{self.__class__.__name__} must implement the sample_boundary_labels() method.")

DeepINN/domain/standard.py

@@ -0,0 +1,50 @@
+from .base import Dataset
+
+class Generic(Dataset):
+    """
+    Very generic dataset loader. It loads the entire dataset in one go, which can lead to overfitting in stiff-PDEs.
+
+    Parameters
+    ----------
+    pde_equation: a method which returns the PDE equation as tensor object (not numpy)
+    collocation_object: A PDE sampler. 
+        Instance of DeepINN.constraint.PDE()
+    bc_object: A BC sampler.
+        Instance of dp.constraint.DirichletBC() or Neumann BC (to be implemented).
+    """
+    def __init__(self, pde_equation, collocation_object, bc_object) -> None:
+        super().__init__()
+
+        self.pde = pde_equation
+        self.pde_sampler = collocation_object
+        # if the bcs is not list, then make it a list
+        self.bc_sampler = bc_object if isinstance(bc_object, (list)) else [bc_object]
+
+        self.bc_list_len = len(self.bc_sampler)
+
+    def sample_collocation_points(self):
+        # returns the collocation points as tensor.
+        # the DeepINN.constraint.PDE() takes care of the missing column in 1D tensor. See tutorials/4. dataset/ 1.basic.ipynb 
+        return self.pde_sampler.sampler_object().sample_points().as_tensor
+
+    def sample_collocation_labels(self):
+        # returns the collocation points using the above function.
+        # also returns the collocation points labels i.e. the residual of the PDE as tensor.
+        self.collocation_points = self.sample_collocation_points() # returns a tensor of sampled collocation points
+        return self.collocation_points, self.pde_sampler.sample_labels(self.collocation_points).unsqueeze(1)
+
+    def sample_boundary_points(self):
+        # returns a list of boundary points on each boundary as tensor
+        self.bc_points_list =[]
+        for bc in self.bc_sampler:
+            self.bc_points_list.append(bc.sampler_object().sample_points().as_tensor)
+        return self.bc_points_list
+
+    def sample_boundary_labels(self):
+        # returns the list of boundary points on each boundary using the above function.
+        # returns a list of tensor boundary conditions on each boundary.
+        self.bc_labels_list = []
+        self.bc_points = self.sample_boundary_points() # returns a list of boundary points on each boundary as tensor
+        for bc_number in range(0, self.bc_list_len):
+            self.bc_labels_list.append(self.bc_sampler[bc_number].sample_labels(self.bc_points[bc_number]).unsqueeze(1))
+        return self.bc_points, self.bc_labels_list

DeepINN/utils/plotting/scatter_points.py

@@ -5,7 +5,7 @@
 import matplotlib.pyplot as plt
 
 
-def scatter(subspace, *samplers, dpi=False, save=False):
+def scatter(subspace, *samplers, dpi=100, save=False):
     """Shows (one batch) of used points in the training. If the sampler is
     static, the shown points will be the points for the training. If not
     the points may vary, depending of the sampler. 
@@ -67,7 +67,7 @@ def _choose_scatter_function(space_dim):
 
 
 def _scatter_1D(points, labels, dpi, save):
-    fig = plt.figure()
+    fig = plt.figure(dpi=dpi)
     ax = fig.add_subplot()
     ax.grid()
     ax.scatter(points, np.zeros_like(points))
@@ -77,7 +77,7 @@ def _scatter_1D(points, labels, dpi, save):
 
 
 def _scatter_2D(points, labels, dpi, save):
-    fig = plt.figure()
+    fig = plt.figure(dpi=dpi)
     ax = fig.add_subplot()
     ax.grid()
     ax.scatter(points[:, 0], points[:, 1])
@@ -88,7 +88,7 @@ def _scatter_2D(points, labels, dpi, save):
 
 
 def _scatter_3D(points, labels, dpi, save):
-    fig = plt.figure()
+    fig = plt.figure(dpi=dpi)
     ax = fig.add_subplot(projection='3d')
     ax.grid()
     ax.scatter(points[:, 0], points[:, 1], points[:, 2])