Davout 0.1.1.dev135__tar.gz → 0.1.1.dev136__tar.gz

{davout-0.1.1.dev135/source/Davout.egg-info → davout-0.1.1.dev136}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Davout
-Version: 0.1.1.dev135
+Version: 0.1.1.dev136
 Summary: Scientific computing utilities combining FEM, ANN and tooling
 Author: Matheus Janczkowski
 Author-email: Matheus Janczkowski <matheusj2009@hotmail.com>

{davout-0.1.1.dev135 → davout-0.1.1.dev136}/pyproject.toml

@@ -63,7 +63,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "Davout"
-version = "0.1.1.dev135"
+version = "0.1.1.dev136"
 description = "Scientific computing utilities combining FEM, ANN and tooling"
 readme = "README.md"
 authors = [

{davout-0.1.1.dev135 → davout-0.1.1.dev136}/source/Davout/DeepMech/aa_tests/mecsol_2026_quotient_space/box_with_imposed_displacement_gradient.py

@@ -35,10 +35,6 @@ lagrange_multiplier_file_name, save_snapshot=False):
     "file name": displacement_file_name, "saving method": "binary", 
     "visualization copy for readable xdmf": False}
 
-    post_processes[0][1]["HomogenizeFieldsGradient"] = {"directory path":
-    results_path, "file name": "homogenized_displacement_gradient_impo"+
-    "sed_stress.txt", "subdomain": "volume 1"}
-
     post_processes[1][1]["SaveField"] = {"directory path": results_path, 
     "file name": lagrange_multiplier_file_name, "saving method": "bina"+
     "ry", "txt copy": False}

{davout-0.1.1.dev135 → davout-0.1.1.dev136}/source/Davout/DeepMech/aa_tests/mecsol_2026_quotient_space/neural_network_trainer.py

@@ -9,6 +9,8 @@ from time import time
 
 from .....Davout.DeepMech.tool_box import ANN_tools, training_tools
 
+from .....Davout.DeepMech.tool_box.loss_assembler_classes import MaximumAbsoluteError
+
 from .....Davout.PythonicUtilities.path_tools import get_parent_path_of_file
 
 # Defines a function to train the neural network model
@@ -125,7 +127,20 @@ quotient_space_dimension):
     test_loss = loss_metric(test_true_values, output_model)
 
     print("\nLoss function on test set:", format(test_loss.numpy(),'.5'+
-    'e'))
+    'e')+"\n")
+
+    # Verifies with the maximum absolute error
+
+    maximum_absolute_error = MaximumAbsoluteError()
+
+    maximum_absolute_value = maximum_absolute_error(test_true_values, 
+    output_model)
+
+    print("\nMaximum absolute error on test set:"+str(format(
+    maximum_absolute_value.numpy(),'.5e'))+"\nwhereas the minimum abso"+
+    "lute error is "+str(format(
+    maximum_absolute_error.minimum_absolute_error(test_true_values, 
+    output_model).numpy(), '.5e'))+"\n")
 
 # Testing block
 
@@ -143,9 +158,15 @@ if __name__=="__main__":
 
     quotient_space_dimension = 9
 
-    train_surrogate_model(displacement_data_file, input_data_file, 
-    saved_model_file, results_path, n_training_samples,
-    quotient_space_dimension)
+    # Trains a new model
+
+    training_flag = False 
+
+    if training_flag:
+
+        train_surrogate_model(displacement_data_file, input_data_file, 
+        saved_model_file, results_path, n_training_samples,
+        quotient_space_dimension)
 
     test_surrogate_model(displacement_data_file, input_data_file,
     saved_model_file, results_path, n_training_samples,

{davout-0.1.1.dev135 → davout-0.1.1.dev136}/source/Davout/DeepMech/tool_box/loss_assembler_classes.py

@@ -945,4 +945,24 @@ class QuadraticLossOverAnyResidualAssembler(tf.keras.losses.Loss):
         self.dR_du.numpy().tolist(), "dtype":
         self.dR_du.dtype.name})
 
-        return config
+        return config
+    
+########################################################################
+#                          Conventional losses                         #
+########################################################################
+
+# Defines a function for the maximum absolute error
+
+class MaximumAbsoluteError(tf.keras.losses.Loss):
+
+    # Defines just the call method to return the maximum absolute value
+
+    def call(self, y_true, y_pred):
+
+        return tf.reduce_max(tf.abs(y_true - y_pred))
+    
+    # Defines a function to get the minimum absolute value
+
+    def minimum_absolute_error(self, y_true, y_pred):
+
+        return tf.reduce_min(tf.abs(y_true - y_pred))

{davout-0.1.1.dev135 → davout-0.1.1.dev136/source/Davout.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Davout
-Version: 0.1.1.dev135
+Version: 0.1.1.dev136
 Summary: Scientific computing utilities combining FEM, ANN and tooling
 Author: Matheus Janczkowski
 Author-email: Matheus Janczkowski <matheusj2009@hotmail.com>