Merge remote-tracking branch 'origin/dev' into feature/tuple_units

Author: teubert

src/progpy/data_models/lstm_model.py

@@ -538,9 +538,7 @@ def from_data(cls, inputs, outputs, event_states=None, t_met=None, **kwargs):
         from tensorflow import keras
 
         # Build model
-        callbacks = [
-            keras.callbacks.ModelCheckpoint("best_model.keras", save_best_only=True)
-        ]
+        callbacks = [ ]
 
         if params['early_stop']:
             callbacks.append(keras.callbacks.EarlyStopping(**params['early_stop.cfg']))
@@ -593,8 +591,6 @@ def from_data(cls, inputs, outputs, event_states=None, t_met=None, **kwargs):
             workers=params['workers'],
             use_multiprocessing=(params['workers'] > 1))
 
-        model = keras.models.load_model("best_model.keras")
-
         # Split model into separate models
         n_state_layers = params['layers'] + 1 + (params['dropout'] > 0) + (params['normalize'])
         output_layer_input = keras.layers.Input(model.layers[n_state_layers-1].output.shape[1:])

src/progpy/prognostics_model.py

@@ -787,7 +787,7 @@ def simulate_to_threshold(self, future_loading_eqn: abc.Callable = None, first_o
         integration_method: str, optional
             Integration method, e.g. 'rk4' or 'euler' (default: 'euler')
         save_freq : float, optional
-            Frequency at which output is saved (s), e.g., save_freq = 10 \n
+            Frequency at which output is saved (s), e.g., save_freq = 10. A save_freq of 0 will save every step. \n
         save_pts : list[float], optional
             Additional ordered list of custom times where output is saved (s), e.g., save_pts= [50, 75] \n
         eval_pts : list[float], optional
@@ -883,8 +883,8 @@ def simulate_to_threshold(self, future_loading_eqn: abc.Callable = None, first_o
             raise ValueError("'dt' must be positive, was {}".format(config['dt']))
         if not isinstance(config['save_freq'], Number) and not isinstance(config['save_freq'], tuple):
             raise TypeError("'save_freq' must be a number, was a {}".format(type(config['save_freq'])))
-        if (isinstance(config['save_freq'], Number) and config['save_freq'] <= 0) or \
-            (isinstance(config['save_freq'], tuple) and config['save_freq'][1] <= 0):
+        if (isinstance(config['save_freq'], Number) and config['save_freq'] < 0) or \
+            (isinstance(config['save_freq'], tuple) and config['save_freq'][1] < 0):
             raise ValueError("'save_freq' must be positive, was {}".format(config['save_freq']))
         if not isinstance(config['save_pts'], abc.Iterable):
             raise TypeError("'save_pts' must be list or array, was a {}".format(type(config['save_pts'])))
@@ -1013,7 +1013,8 @@ def next_time(t, x):
             def next_time(t, x=None):
                 next_save_pt = save_pts[save_pt_index] if save_pt_index < len(save_pts) else float('inf')
                 next_eval_pt = eval_pts[eval_pt_index] if eval_pt_index < len(eval_pts) else float('inf')
-                return min(dt, next_save-t, next_save_pt-t, next_eval_pt-t)
+                opts = (item for item in (dt, next_save-t, next_save_pt-t, next_eval_pt-t) if item > 0)
+                return min(*opts)
         elif dt_mode != 'function':
             raise ValueError(f"'dt' mode {dt_mode} not supported. Must be 'constant', 'auto', or a function")
 

tests/test_base_models.py

@@ -918,6 +918,14 @@ def load(t, x=None):
         result = m.simulate_to_threshold(load, dt=2, save_freq=0.75, save_pts=[1.5, 2.5])
         self.assertListEqual(result.times, [0, 2, 4])  
 
+        # With save_freq==0
+        result = m.simulate_to_threshold(load, dt=2, save_freq=0)
+        self.assertListEqual(result.times, [0, 2, 4])
+
+        # With save_freq==0 and auto step size
+        result = m.simulate_to_threshold(load, dt=('auto', 2), save_freq=0)
+        self.assertListEqual(result.times, [0, 2, 4])
+
         result = m.simulate_to_threshold(load, dt=2, save_pts=[2.5])
         self.assertListEqual(result.times, [0, 4])