fix: normalize pair distances only if max_rul is set

rul_datasets/core.py

@@ -1,7 +1,7 @@
 """Basic data modules for experiments involving only a single subset of any RUL
 dataset. """
 
-from typing import Dict, List, Optional, Tuple, Any, Callable
+from typing import Dict, List, Optional, Tuple, Any, Callable, cast, Union
 
 import numpy as np
 import pytorch_lightning as pl
@@ -411,17 +411,17 @@ def __init__(
         elif mode == "labeled":
             self._get_pair_func = self._get_labeled_pair_idx
 
-    def _get_max_rul(self):
+    def _get_max_rul(self) -> Optional[int]:
         max_ruls = [dm.reader.max_rul for dm in self.dms]
         if all(m is None for m in max_ruls):
-            max_rul = 1e10
+            max_rul = None
         elif any(m is None for m in max_ruls):
             raise ValueError(
                 "PairedRulDataset needs a set max_rul for all or none of the readers "
                 "but at least one and not all of them has None."
             )
         else:
-            max_rul = max(max_ruls)
+            max_rul = max(cast(List[int], max_ruls))
 
         return max_rul
 
@@ -470,7 +470,7 @@ def __next__(self) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tens
         else:
             raise StopIteration
 
-    def _get_pair_idx(self) -> Tuple[int, int, int, int, int]:
+    def _get_pair_idx(self) -> Tuple[int, int, int, Union[int, float], int]:
         chosen_run_idx = self._rng.integers(0, len(self._features))
         domain_label = self._run_domain_idx[chosen_run_idx]
         chosen_run = self._features[chosen_run_idx]
@@ -480,7 +480,7 @@ def _get_pair_idx(self) -> Tuple[int, int, int, int, int]:
             low=0,
             high=run_length - self.min_distance,
         )
-        end_idx = min(run_length, anchor_idx + self._max_rul)
+        end_idx = min(run_length, anchor_idx + (self._max_rul or 999999))
         query_idx = self._rng.integers(
             low=anchor_idx + self.min_distance,
             high=end_idx,
@@ -489,7 +489,7 @@ def _get_pair_idx(self) -> Tuple[int, int, int, int, int]:
 
         return chosen_run_idx, anchor_idx, query_idx, distance, domain_label
 
-    def _get_pair_idx_piecewise(self) -> Tuple[int, int, int, int, int]:
+    def _get_pair_idx_piecewise(self) -> Tuple[int, int, int, Union[int, float], int]:
         chosen_run_idx = self._rng.integers(0, len(self._features))
         domain_label = self._run_domain_idx[chosen_run_idx]
         chosen_run = self._features[chosen_run_idx]
@@ -511,7 +511,7 @@ def _get_pair_idx_piecewise(self) -> Tuple[int, int, int, int, int]:
 
         return chosen_run_idx, anchor_idx, query_idx, distance, domain_label
 
-    def _get_labeled_pair_idx(self) -> Tuple[int, int, int, int, int]:
+    def _get_labeled_pair_idx(self) -> Tuple[int, int, int, Union[int, float], int]:
         chosen_run_idx = self._rng.integers(0, len(self._features))
         domain_label = self._run_domain_idx[chosen_run_idx]
         chosen_run = self._features[chosen_run_idx]
@@ -527,7 +527,7 @@ def _get_labeled_pair_idx(self) -> Tuple[int, int, int, int, int]:
             high=run_length,
         )
         # RUL label difference is negative time step difference
-        distance = int(chosen_labels[anchor_idx] - chosen_labels[query_idx])
+        distance = (chosen_labels[anchor_idx] - chosen_labels[query_idx]).item()
 
         return chosen_run_idx, anchor_idx, query_idx, distance, domain_label
 
@@ -536,13 +536,15 @@ def _build_pair(
         run: torch.Tensor,
         anchor_idx: int,
         query_idx: int,
-        distance: int,
+        distance: Union[int, float],
         domain_label: int,
     ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
         anchors = run[anchor_idx]
         queries = run[query_idx]
         domain_tensor = torch.tensor(domain_label, dtype=torch.float)
-        distances = torch.tensor(distance, dtype=torch.float) / self._max_rul
-        distances = torch.clamp_max(distances, max=1)  # max distance is max_rul
+        distances = torch.tensor(distance, dtype=torch.float)
+        if self._max_rul is not None:  # normalize only if max_rul is set
+            distances /= self._max_rul
+            distances = torch.clamp_max(distances, max=1)
 
         return anchors, queries, distances, domain_tensor