diff --git a/source/lmp/tests/test_lammps_pd.py b/source/lmp/tests/test_lammps_pd.py
index c495071551..8ccfe478de 100644
--- a/source/lmp/tests/test_lammps_pd.py
+++ b/source/lmp/tests/test_lammps_pd.py
@@ -30,6 +30,9 @@
 data_type_map_file = Path(__file__).parent / "data_type_map.lmp"
 md_file = Path(__file__).parent / "md.out"
 
+ATOL = 1e-7
+RTOL = 1e-6
+
 # this is as the same as python and c++ tests, test_deeppot_a.py
 expected_ae = np.array(
     [
@@ -334,7 +337,7 @@ def test_pair_deepmd_virial(lammps):
     for ii in range(9):
         assert np.array(
             lammps.variables[f"virial{ii}"].value
-        ) / constants.nktv2p == pytest.approx(expected_v[idx_map, ii], 1e-6, 1e-9)
+        ) / constants.nktv2p == pytest.approx(expected_v[idx_map, ii], RTOL, ATOL)
 
 
 def test_pair_deepmd_model_devi(lammps):
@@ -376,30 +379,30 @@ def test_pair_deepmd_model_devi_virial(lammps):
         "1 all custom 1 dump id " + " ".join([f"v_virial{ii}" for ii in range(9)])
     )
     lammps.run(0)
-    assert lammps.eval("pe") == pytest.approx(expected_e, 1e-6, 1e-9)
+    assert lammps.eval("pe") == pytest.approx(expected_e, RTOL, ATOL)
     for ii in range(6):
         assert lammps.atoms[ii].force == pytest.approx(
-            expected_f[lammps.atoms[ii].id - 1], 1e-6, 1e-9
+            expected_f[lammps.atoms[ii].id - 1], RTOL, ATOL
         )
     idx_map = lammps.lmp.numpy.extract_atom("id") - 1
     for ii in range(9):
         assert np.array(
             lammps.variables[f"virial{ii}"].value
-        ) / constants.nktv2p == pytest.approx(expected_v[idx_map, ii], 1e-6, 1e-9)
+        ) / constants.nktv2p == pytest.approx(expected_v[idx_map, ii], RTOL, ATOL)
     # load model devi
     md = np.loadtxt(md_file.resolve())
     expected_md_f = np.linalg.norm(np.std([expected_f, expected_f2], axis=0), axis=1)
-    assert md[7:] == pytest.approx(expected_md_f, 1e-6, 1e-9)
-    assert md[4] == pytest.approx(np.max(expected_md_f), 1e-6, 1e-9)
-    assert md[5] == pytest.approx(np.min(expected_md_f), 1e-6, 1e-9)
-    assert md[6] == pytest.approx(np.mean(expected_md_f), 1e-6, 1e-9)
+    assert md[7:] == pytest.approx(expected_md_f, RTOL, ATOL)
+    assert md[4] == pytest.approx(np.max(expected_md_f), RTOL, ATOL)
+    assert md[5] == pytest.approx(np.min(expected_md_f), RTOL, ATOL)
+    assert md[6] == pytest.approx(np.mean(expected_md_f), RTOL, ATOL)
     expected_md_v = (
         np.std([np.sum(expected_v, axis=0), np.sum(expected_v2, axis=0)], axis=0) / 6
     )
-    assert md[1] == pytest.approx(np.max(expected_md_v), 1e-6, 1e-9)
-    assert md[2] == pytest.approx(np.min(expected_md_v), 1e-6, 1e-9)
+    assert md[1] == pytest.approx(np.max(expected_md_v), RTOL, ATOL)
+    assert md[2] == pytest.approx(np.min(expected_md_v), RTOL, ATOL)
     assert md[3] == pytest.approx(
-        np.sqrt(np.mean(np.square(expected_md_v))), 1e-6, 1e-9
+        np.sqrt(np.mean(np.square(expected_md_v))), RTOL, ATOL
     )
 
 
@@ -504,14 +507,14 @@ def test_pair_deepmd_virial_real(lammps_real):
     lammps_real.run(0)
     assert lammps_real.eval("pe") == pytest.approx(
         expected_e * constants.ener_metal2real,
-        1e-6,
-        1e-9,
+        RTOL,
+        ATOL,
     )
     for ii in range(6):
         assert lammps_real.atoms[ii].force == pytest.approx(
             expected_f[lammps_real.atoms[ii].id - 1] * constants.force_metal2real,
-            1e-6,
-            1e-9,
+            RTOL,
+            ATOL,
         )
     idx_map = lammps_real.lmp.numpy.extract_atom("id") - 1
     for ii in range(9):
@@ -519,8 +522,8 @@ def test_pair_deepmd_virial_real(lammps_real):
             lammps_real.variables[f"virial{ii}"].value
         ) / constants.nktv2p_real == pytest.approx(
             expected_v[idx_map, ii] * constants.ener_metal2real,
-            1e-6,
-            1e-9,
+            RTOL,
+            ATOL,
         )
 
 
@@ -569,14 +572,14 @@ def test_pair_deepmd_model_devi_virial_real(lammps_real):
     lammps_real.run(0)
     assert lammps_real.eval("pe") == pytest.approx(
         expected_e * constants.ener_metal2real,
-        1e-6,
-        1e-9,
+        RTOL,
+        ATOL,
     )
     for ii in range(6):
         assert lammps_real.atoms[ii].force == pytest.approx(
             expected_f[lammps_real.atoms[ii].id - 1] * constants.force_metal2real,
-            1e-6,
-            1e-9,
+            RTOL,
+            ATOL,
         )
     idx_map = lammps_real.lmp.numpy.extract_atom("id") - 1
     for ii in range(9):
@@ -584,37 +587,37 @@ def test_pair_deepmd_model_devi_virial_real(lammps_real):
             lammps_real.variables[f"virial{ii}"].value
         ) / constants.nktv2p_real == pytest.approx(
             expected_v[idx_map, ii] * constants.ener_metal2real,
-            1e-6,
-            1e-9,
+            RTOL,
+            ATOL,
         )
     # load model devi
     md = np.loadtxt(md_file.resolve())
     expected_md_f = np.linalg.norm(np.std([expected_f, expected_f2], axis=0), axis=1)
     assert md[7:] == pytest.approx(
-        expected_md_f * constants.force_metal2real, 1e-6, 1e-9
+        expected_md_f * constants.force_metal2real, RTOL, ATOL
     )
     assert md[4] == pytest.approx(
-        np.max(expected_md_f) * constants.force_metal2real, 1e-6, 1e-9
+        np.max(expected_md_f) * constants.force_metal2real, RTOL, ATOL
     )
     assert md[5] == pytest.approx(
-        np.min(expected_md_f) * constants.force_metal2real, 1e-6, 1e-9
+        np.min(expected_md_f) * constants.force_metal2real, RTOL, ATOL
     )
     assert md[6] == pytest.approx(
-        np.mean(expected_md_f) * constants.force_metal2real, 1e-6, 1e-9
+        np.mean(expected_md_f) * constants.force_metal2real, RTOL, ATOL
     )
     expected_md_v = (
         np.std([np.sum(expected_v, axis=0), np.sum(expected_v2, axis=0)], axis=0) / 6
     )
     assert md[1] == pytest.approx(
-        np.max(expected_md_v) * constants.ener_metal2real, 1e-6, 1e-9
+        np.max(expected_md_v) * constants.ener_metal2real, RTOL, ATOL
     )
     assert md[2] == pytest.approx(
-        np.min(expected_md_v) * constants.ener_metal2real, 1e-6, 1e-9
+        np.min(expected_md_v) * constants.ener_metal2real, RTOL, ATOL
     )
     assert md[3] == pytest.approx(
         np.sqrt(np.mean(np.square(expected_md_v))) * constants.ener_metal2real,
-        1e-6,
-        1e-9,
+        RTOL,
+        ATOL,
     )