Add tests for multiple NEP potentials

tests/gpumd/carbon_average/README.md

@@ -0,0 +1,9 @@
+# `dump_observer`: average
+
+Test case for `average` mode of `dump_observer`. The two potentials are the `C_2022_NEP3.txt` potential, and a modified version of the same potential with the two first cutoffs changed. 
+Two MD runs have been performed with each of the two potentials as the only potential, to generate the files `observer0.xyz` and `observer1.xyz`.
+The test runs with the average of the two potentials. 
+The resulitng `observer.xyz` file shall be compared to the average of the two `observer*.xyz` files.
+This comparison is performed by the `compare_results.py` script, which compares the outputs numerically down to a precision of 1e-4.
+
+Definition of pass: For the test of pass, running the `compare_results.py` script shall return all values to be within the tolerance. 

tests/gpumd/carbon_average/compare_results.py

@@ -0,0 +1,66 @@
+from math import isclose
+
+def isfloat(num):
+    try:
+        float(num)
+        return True
+    except ValueError:
+        return False
+
+
+
+def print_justified(val1, val2, close=True):
+    print(f'{val1}'.ljust(30), end='')
+    print(f'\t', end='')
+    print(f'{val2}'.ljust(30), end='')
+    if not close:
+        print('!!!!')
+    else:
+        print('\n', end='')
+
+
+def get_floats_in_file(file):
+    entries = []
+    with open(file, 'r') as f:
+        for line in f.readlines():
+            linem = line.strip()
+            linem = linem.replace(' ', '=')
+            linem = linem.replace(':', '=')
+            linem = linem.replace('"', '=')
+            split_line = linem.split('=')
+            entries.extend([float(entry) for entry in split_line if isfloat(entry)])
+    return entries
+
+
+entries0 = get_floats_in_file('observer0.xyz')
+entries1 = get_floats_in_file('observer1.xyz')
+assert len(entries0) == len(entries1)
+average_entries = [(entries0[i] + entries1[i])*0.5 for i in range(len(entries0))]
+entries_to_test = get_floats_in_file('observer.xyz')
+assert len(entries0) == len(entries_to_test)
+
+n_total = len(entries_to_test)
+n_close = 0
+header_printed = False
+for i in range(n_total):
+    test = entries_to_test[i]
+    ref = average_entries[i]
+    close = isclose(test, ref, abs_tol=1e-4, rel_tol=1e-4)
+    if not close:
+        if not header_printed:
+            print_justified('Test', 'Ref')
+            header_printed = True
+        print_justified(test, ref, close=close)
+        raise ValueError(f"`observer.xyz` does not match average of `observer0.xyz` and `observer1.xyz` for value {i}")
+    n_close += 1 if close else 0
+
+print(f'{n_close}/{n_total} within tolerance')
+print("Test passed")
+
+
+
+
+
+
+
+

tests/gpumd/carbon_average/run.in

@@ -0,0 +1,16 @@
+potential       ../../../potentials/nep/C_2022_NEP3.txt
+potential       C_2022_NEP3_MODIFIED.txt
+velocity       	300
+
+time_step       1.0
+
+ensemble        nve
+dump_observer 	average 1 1 1
+run             1
+
+
+
+
+
+
+