feat: filter short-distance FP structures

dpgen/generator/arginfo.py

@@ -654,15 +654,17 @@ def fp_style_vasp_args() -> list[Argument]:
         "Skip configurations with unreasonable simulation box geometries before "
         "first-principles calculations. This parameter accepts a semicolon-separated "
         "string of colon-separated key-value pairs defining geometric criteria. "
-        "Example: 'length_ratio:3;height_ratio:3;wrap_ratio:0.5;tilt_ratio:0.5'. "
+        "Example: 'length_ratio:3;height_ratio:3;min_distance:1.0'. "
         "Available criteria: "
         "(1) 'length_ratio': maximum ratio of cell edge lengths (max/min); "
         "(2) 'height_ratio': ratio of maximum cell edge length to minimum "
         "face-to-face distance; "
         "(3) 'wrap_ratio': maximum absolute ratio of off-diagonal to diagonal "
         "cell matrix elements, controlling triclinic wrapping; "
-        "(4) 'tilt_ratio': maximum absolute tilt ratio for triclinic cells. "
-        "Configurations exceeding any specified threshold are skipped."
+        "(4) 'tilt_ratio': maximum absolute tilt ratio for triclinic cells; "
+        "(5) 'min_distance' (alias 'min_dist'): minimum allowed periodic "
+        "interatomic distance in Angstrom. Configurations violating any "
+        "specified threshold are skipped."
     )
 
     return [

dpgen/generator/run.py

@@ -2319,6 +2319,7 @@ def check_cluster(conf_name, fp_cluster_vacuum, fmt="lammps/dump"):
 
 
 def check_bad_box(conf_name, criteria, fmt="lammps/dump"):
+    """Return whether a configuration violates any FP screening criterion."""
     all_c = criteria.split(";")
     sys = dpdata.System(conf_name, fmt)
     assert sys.get_nframes() == 1
@@ -2353,6 +2354,18 @@ def check_bad_box(conf_name, criteria, fmt="lammps/dump"):
             ]
             if np.max(np.abs(ratio)) > float(value):
                 is_bad = True
+        elif key in {"min_distance", "min_dist"}:
+            from ase.geometry import find_mic
+
+            natoms = sys.get_natoms()
+            if natoms > 1:
+                atom_pairs = np.triu_indices(natoms, k=1)
+                vectors = (
+                    sys["coords"][0][atom_pairs[0]] - sys["coords"][0][atom_pairs[1]]
+                )
+                _, distances = find_mic(vectors, sys["cells"][0], pbc=True)
+                if np.min(distances) < float(value):
+                    is_bad = True
         else:
             raise RuntimeError("unknow key", key)
     return is_bad

tests/generator/test_check_bad_box.py

@@ -23,3 +23,10 @@ def test_height_ratio(self):
         conf_bad = os.path.join(dirname, "check_bad_box", "bad.height.POSCAR")
         self.assertTrue(check_bad_box(conf_bad, "height_ratio:5", fmt="vasp/POSCAR"))
         self.assertFalse(check_bad_box(conf_bad, "length_ratio:5", fmt="vasp/POSCAR"))
+
+    def test_min_distance(self):
+        dirname = os.path.dirname(__file__)
+        conf = os.path.join(dirname, "check_bad_box", "good.lammpstrj")
+        self.assertFalse(check_bad_box(conf, "min_distance:2.0"))
+        self.assertTrue(check_bad_box(conf, "min_distance:2.2"))
+        self.assertTrue(check_bad_box(conf, "min_dist:2.2"))