Update pi-deeponet example.

docs/demos/operator/pideeponet_poisson1d.rst

@@ -14,8 +14,8 @@ for the one-dimensional Poisson problem
 
 with zero Dirichlet boundary conditions :math:`u(0) = u(1) = 0`.
 
-The source term :math:`f` is supposed to be an arbitrary continuous function.
-We choose polynomials of degree three.
+The source term :math:`f` is supposed to be an arbitrary continuous function,
+we choose polynomials of degree three.
 
 
 Implementation
@@ -50,8 +50,9 @@ First, we define the PDE with boundary conditions and the domain:
     )
 
 
-Next, we specify the function space for :math:`f` using a Power series and the corresponding evaluation points. Together with
-the PDE, the function space is used to define a PDEOperator ``dde.data.PDEOperatorCartesianProd`` that incorporates the PDE into
+Next, we specify the function space for :math:`f` and the corresponding evaluation points.
+Together with the PDE, the function space is used to define a
+PDEOperator ``dde.data.PDEOperatorCartesianProd`` that incorporates the PDE into
 the loss function.
 
 .. code-block:: python
@@ -70,7 +71,7 @@ the loss function.
     )
 
 
-Next, we define a DeepONet ``dde.nn.DeepONetCartesianProd``.
+The DeepONet can be defined using ``dde.nn.DeepONetCartesianProd``.
 The branch net is chosen as a fully connected neural network of size ``[m, 32, p]`` where ``p=32``,
 and the the trunk net is a fully connected neural network of size ``[dim_x, 32, p]``.
 
@@ -86,16 +87,21 @@ and the the trunk net is a fully connected neural network of size ``[dim_x, 32,
     )
 
 
-We define a ``Model``, and then train the model with Adam and learning rate 0.001 for 5000 iterations:
+We define a ``Model`` and train it with L-BFGS (pytorch) or Adam:
 
 .. code-block:: python
 
     model = dde.Model(pde_op, net)
-    model.compile("adam", lr=0.001, metrics=["l2 relative error"])
-    model.train(iterations=5000)
-
-
-Finally, the trained model can be used to solve the Poisson equation for arbitrary source terms :math:`f`.
+    if dde.backend.backend_name == "pytorch":
+        dde.optimizers.set_LBFGS_options(maxiter=1000)
+        model.compile("L-BFGS", metrics=["l2 relative error"])
+        model.train()
+    else:
+        model.compile("adam", lr=0.001, metrics=["l2 relative error"])
+        model.train(iterations=10000)
+
+Finally, the trained model can be used to predict the solution of the Poisson
+equation for any admissible source term :math:`f`.
 
 ![](pideeponet_poisson1d.png)
 

examples/operator/pideeponet_poisson1d.py

@@ -60,12 +60,16 @@ def boundary(_, on_boundary):
     kernel_initializer="Glorot normal",
 )
 
-
 # Define and train model
 model = dde.Model(pde_op, net)
-model.compile("adam", lr=0.001, metrics=["l2 relative error"])
-model.train(iterations=5000)
 
+if dde.backend.backend_name == "pytorch":
+    dde.optimizers.set_LBFGS_options(maxiter=1000)
+    model.compile("L-BFGS", metrics=["l2 relative error"])
+    model.train()
+else:
+    model.compile("adam", lr=0.001, metrics=["l2 relative error"])
+    model.train(iterations=10000)
 
 # Plot realisations of f(x)
 n = 3