adding cp2k/restart_aimd_ouput

.gitignore

@@ -20,4 +20,5 @@ dist
 dpdata.egg-info
 _version.py
 !tests/cp2k/aimd/cp2k.log
+!tests/cp2k/restart_aimd/ch4.log
 __pycache__

dpdata/cp2k/output.py

@@ -14,36 +14,42 @@
 delimiter_patterns.append(delimiter_p1)
 delimiter_patterns.append(delimiter_p2)
 avail_patterns = []
-
 avail_patterns.append(re.compile(r'^ INITIAL POTENTIAL ENERGY'))
 avail_patterns.append(re.compile(r'^ ENSEMBLE TYPE'))
 
 class Cp2kSystems(object):
     """
     deal with cp2k outputfile
     """
-    def __init__(self, log_file_name, xyz_file_name):
+    def __init__(self, log_file_name, xyz_file_name, restart=False):
         self.log_file_object = open(log_file_name, 'r')
         self.xyz_file_object = open(xyz_file_name, 'r')
         self.log_block_generator = self.get_log_block_generator()
         self.xyz_block_generator = self.get_xyz_block_generator()
+        self.restart_flag = restart
         self.cell=None
+
+        if self.restart_flag:
+            self.handle_single_log_frame(next(self.log_block_generator))
+
     def __del__(self):
         self.log_file_object.close()
         self.xyz_file_object.close()
+
     def __iter__(self):
         return self
+
     def __next__(self):
         info_dict = {}
         log_info_dict = self.handle_single_log_frame(next(self.log_block_generator))
         xyz_info_dict = self.handle_single_xyz_frame(next(self.xyz_block_generator))
         eq1 = [v1==v2 for v1,v2 in zip(log_info_dict['atom_numbs'], xyz_info_dict['atom_numbs'])]
         eq2 = [v1==v2 for v1,v2 in zip(log_info_dict['atom_names'], xyz_info_dict['atom_names'])]
         eq3 = [v1==v2 for v1,v2 in zip(log_info_dict['atom_types'], xyz_info_dict['atom_types'])]
-        assert all(eq1), (log_info_dict,xyz_info_dict,'There may be errors in the file')
-        assert all(eq2), (log_info_dict,xyz_info_dict,'There may be errors in the file')
-        assert all(eq3), (log_info_dict,xyz_info_dict,'There may be errors in the file')
-        assert log_info_dict['energies']==xyz_info_dict['energies'], (log_info_dict['energies'],xyz_info_dict['energies'],'There may be errors in the file')
+        assert all(eq1), (log_info_dict,xyz_info_dict,'There may be errors in the file. If it is a restart task; use restart=True')
+        assert all(eq2), (log_info_dict,xyz_info_dict,'There may be errors in the file. If it is a restart task; use restart=True')
+        assert all(eq3), (log_info_dict,xyz_info_dict,'There may be errors in the file. If it is a restart task; use restart=True')
+        assert log_info_dict['energies']==xyz_info_dict['energies'], (log_info_dict['energies'], xyz_info_dict['energies'],'There may be errors in the file')
         info_dict.update(log_info_dict)
         info_dict.update(xyz_info_dict)
         return info_dict
@@ -111,6 +117,7 @@ def handle_single_log_frame(self, lines):
                     force_lines.append(line)
             if energy_pattern_1.match(line):
                 energy = float(energy_pattern_1.match(line).groupdict()['number']) * AU_TO_EV
+                #print('1to', energy)
             if energy_pattern_2.match(line):
                 energy = float(energy_pattern_2.match(line).groupdict()['number']) * AU_TO_EV
             if cell_length_pattern.match(line):

dpdata/system.py

@@ -1026,7 +1026,7 @@ def __init__ (self,
                 - ``gaussian/log``: gaussian logs
                 - ``gaussian/md``: gaussian ab initio molecular dynamics
                 - ``cp2k/output``: cp2k output file
-                - ``cp2k/aimd_output``: cp2k aimd output  dir(contains *pos*.xyz and *.log file)
+                - ``cp2k/aimd_output``: cp2k aimd output  dir(contains *pos*.xyz and *.log file); optional `restart=True` if it is a cp2k restarted task.
                 - ``pwmat/movement``: pwmat md output file
                 - ``pwmat/out.mlmd``: pwmat scf output file
 
@@ -1091,13 +1091,21 @@ def has_virial(self) :
         return ('virials' in self.data)
 
     @register_from_funcs.register_funcs('cp2k/aimd_output')
-    def from_cp2k_aimd_output(self, file_dir):
+    def from_cp2k_aimd_output(self, file_dir, restart=False):
         xyz_file=sorted(glob.glob("{}/*pos*.xyz".format(file_dir)))[0]
         log_file=sorted(glob.glob("{}/*.log".format(file_dir)))[0]
-        for info_dict in Cp2kSystems(log_file, xyz_file):
+        for info_dict in Cp2kSystems(log_file, xyz_file, restart):
             l = LabeledSystem(data=info_dict)
             self.append(l)
 
+    # @register_from_funcs.register_funcs('cp2k/restart_aimd_output')
+    # def from_cp2k_aimd_output(self, file_dir, restart=True):
+    #     xyz_file = sorted(glob.glob("{}/*pos*.xyz".format(file_dir)))[0]
+    #     log_file = sorted(glob.glob("{}/*.log".format(file_dir)))[0]
+    #     for info_dict in Cp2kSystems(log_file, xyz_file, restart):
+    #         l = LabeledSystem(data=info_dict)
+    #         self.append(l)
+
     @register_from_funcs.register_funcs('fhi_aims/md')
     def from_fhi_aims_output(self, file_name, md=True, begin=0, step =1):
         self.data['atom_names'], \

dpdata/vasp/outcar.py

@@ -128,5 +128,5 @@ def analyze_block(lines, ntot, nelm) :
                 tmp_l = lines[jj]
                 info = [float(ss) for ss in tmp_l.split()]
                 coord.append(info[:3])
-                force.append(info[3:])
+                force.append(info[3:6])
     return coord, cell, energy, force, virial, is_converge

tests/cp2k/restart_aimd/CH4-pos-1.xyz

@@ -0,0 +1,35 @@
+       5
+ i =        6, time =        3.000, E =        -8.0715742982
+  H         5.3857900416        4.0742097705        3.5811120132
+  H         3.9199512866        4.7900150110        4.3285090053
+  H         5.6002847426        5.6077224621        4.4496425595
+  H         5.3479559215        4.1480369706        5.3848299963
+  C         5.0220694622        4.6784611518        4.4577943630
+       5
+ i =        7, time =        3.500, E =        -8.0711045977
+  H         5.3848978757        4.0776224380        3.5814113731
+  H         3.9208809250        4.7887262191        4.3220837446
+  H         5.6114006618        5.6027985800        4.4525980866
+  H         5.3547898958        4.1463117535        5.3804926360
+  C         5.0205568908        4.6788387910        4.4584233243
+       5
+ i =        8, time =        4.000, E =        -8.0707608306
+  H         5.3835297528        4.0817338403        3.5830596000
+  H         3.9234320791        4.7879031911        4.3168681896
+  H         5.6214054135        5.5987495057        4.4552276547
+  H         5.3599417537        4.1448123874        5.3752072465
+  C         5.0191820720        4.6790259231        4.4589444566
+       5
+ i =        9, time =        4.500, E =        -8.0705194159
+  H         5.3817673668        4.0863967978        3.5858600160
+  H         3.9273863696        4.7876326936        4.3129364024
+  H         5.6300924745        5.5954927653        4.4575136286
+  H         5.3633363130        4.1434413688        5.3691425751
+  C         5.0179801128        4.6790427107        4.4593553989
+       5
+ i =       10, time =        5.000, E =        -8.0703422670
+  H         5.3797263472        4.0914106472        3.5895251624
+  H         3.9324173150        4.7880020164        4.3103298482
+  H         5.6372698671        5.5929282695        4.4594480504
+  H         5.3649502724        4.1420701065        5.3625404656
+  C         5.0169867769        4.6789178539        4.4596571950