fix: reject non-finite model deviations

dpgen2/exploration/render/traj_render_lammps.py

@@ -68,6 +68,28 @@
         ):  # In case model-devi.out is 1-dimensional
             dd = dd.reshape((1, len(dd)))  # type: ignore
 
+        # A NaN or infinity would otherwise reach the report and scheduler, where
+        # comparisons silently produce invalid trust levels. Fail at the artifact
+        # boundary so users can inspect the corresponding LAMMPS task directly.
+        deviation_names = (
+            DeviManager.MAX_DEVI_V,
+            DeviManager.MIN_DEVI_V,
+            DeviManager.AVG_DEVI_V,
+            DeviManager.MAX_DEVI_F,
+            DeviManager.MIN_DEVI_F,
+            DeviManager.AVG_DEVI_F,
+        )
+        invalid = np.argwhere(~np.isfinite(dd[:, 1:7]))
+        if invalid.size:
+            locations = ", ".join(
+                f"row {row + 1} ({deviation_names[column]})" for row, column in invalid
+            )
+            raise ValueError(
+                f"Non-finite model-deviation value in {fname}: {locations}. "
+                "Inspect the LAMMPS/DeePMD task output before scheduling the "
+                "next exploration iteration."
+            )
+
         model_devi.add(DeviManager.MAX_DEVI_V, dd[:, 1])  # type: ignore
         model_devi.add(DeviManager.MIN_DEVI_V, dd[:, 2])  # type: ignore
         model_devi.add(DeviManager.AVG_DEVI_V, dd[:, 3])  # type: ignore

tests/exploration/test_traj_render_lammps.py

@@ -1,6 +1,9 @@
 import json
 import os
 import unittest
+from pathlib import (
+    Path,
+)
 
 import dpdata
 import numpy as np
@@ -15,6 +18,16 @@
 
 
 class TestTrajRenderLammps(unittest.TestCase):
+    def test_rejects_non_finite_model_deviation(self):
+        model_devi = Path("model_devi.out")
+        model_devi.write_text("0 0.1 0.0 0.0 0.2 0.0 0.0\n1 0.2 0.0 0.0 nan 0.0 0.0\n")
+
+        with self.assertRaisesRegex(
+            ValueError,
+            r"Non-finite model-deviation value.*row 2 \(max_devi_f\)",
+        ):
+            TrajRenderLammps().get_model_devi([model_devi])
+
     def test_use_ele_temp_1(self):
         with open("job.json", "w") as f:
             json.dump({"ele_temp": 6.6}, f)
@@ -58,5 +71,6 @@ def test_use_ele_temp_2(self):
         np.testing.assert_array_almost_equal(system.data["aparam"], np.array([[[6.6]]]))
 
     def tearDown(self):
-        if os.path.exists("job.json"):
-            os.remove("job.json")
+        for filename in ("job.json", "model_devi.out"):
+            if os.path.exists(filename):
+                os.remove(filename)