Gauss Trsh Fixes (#713)

@Lilachn91 pointed out that `nosymm` should not be part of the opt or
scf combination of parameters but be its own parameter. This has been
implemented. Furthermore, at times we end up creating an input file that
may have `opt=([param1])` and `opt=([param2])` when rather they should
be combined into one `opt=([param1,param2])`

arc/job/adapters/common.py

@@ -522,4 +522,15 @@ def combine_parameters(input_dict: dict, terms: list) -> Tuple[dict, List]:
                         value = re.sub(term, '', value)
             input_dict_copy[key] = value
 
+    # Parameters may appear as one word, so need to split them via comma
+    parameters = [param.split(',') for param in parameters]
+    # Flatten the list of lists
+    parameters = [item for sublist in parameters for item in sublist]
+    # Remove empty spaces from the beginning and end of each parameter
+    parameters = [param.strip() for param in parameters]
+    # Parameters may appear multiple times in the input_dict, so remove duplicates
+    parameters = list(set(parameters))
+    # Sort the list
+    parameters.sort()
+
     return input_dict_copy, parameters

arc/job/adapters/gaussian.py

@@ -300,7 +300,10 @@ def write_input_file(self) -> None:
             input_dict['job_type_2'] = 'freq IOp(7/33=1)'
 
         elif self.job_type == 'sp':
-            input_dict['job_type_1'] = f'scf=(tight, direct) integral=(grid=ultrafine, {integral_algorithm})'
+            input_dict['job_type_1'] = f'integral=(grid=ultrafine, {integral_algorithm})'
+            if input_dict['trsh']:
+                input_dict['trsh'] += ' '
+            input_dict['trsh'] += 'scf=(tight, direct)'
 
         elif self.job_type == 'scan':
             scans, scans_strings = list(), list()
@@ -317,8 +320,11 @@ def write_input_file(self) -> None:
                 self.torsions = torsions_to_scans(scans, direction=-1)
 
             ts = 'ts, ' if self.is_ts else ''
-            input_dict['job_type_1'] = f'opt=({ts}modredundant, calcfc, noeigentest, maxStep=5) scf=(tight, direct) ' \
+            input_dict['job_type_1'] = f'opt=({ts}modredundant, calcfc, noeigentest, maxStep=5)' \
                                        f'integral=(grid=ultrafine, {integral_algorithm})'
+            if input_dict['trsh']:
+                input_dict['trsh'] += ' '
+            input_dict['trsh'] += 'scf=(tight, direct)'
             input_dict['scan'] = '\n\n' if not input_dict['scan'] else input_dict['scan']
             for scan in scans_strings:
                 input_dict['scan'] += f'D {scan} S {int(360 / self.scan_res)} {self.scan_res:.1f}\n'
@@ -361,13 +367,25 @@ def write_input_file(self) -> None:
             input_dict['job_type_1'] += ' guess=read' if self.checkfile is not None and os.path.isfile(self.checkfile) \
                 else ' guess=mix'
 
+        # Fix OPT
+        terms_opt = [r'opt=\((.*?)\)', r'opt=(\w+)']
+        input_dict, parameters_opt = combine_parameters(input_dict, terms_opt)
+        # If 'opt' parameters are found, concatenate and reinsert them
+        if parameters_opt:
+            # Remove duplicate parameters
+            combined_opt_params = ','.join(parameters_opt)
+            input_dict['job_type_1'] = f"opt=({combined_opt_params}) {input_dict['job_type_1']}"
+
         #Fix SCF
         # This may be redundant due to additional fixes in the above code
         terms = ['scf=\((.*?)\)', 'scf=(\w+)']
         input_dict, parameters = combine_parameters(input_dict, terms)
         if parameters:
             input_dict['trsh'] += f" scf=({','.join(parameters)})"
 
+        # Remove double spaces
+        input_dict['job_type_1'] = input_dict['job_type_1'].replace('  ', ' ')
+
         with open(os.path.join(self.local_path, input_filenames[self.job_adapter]), 'w') as f:
             f.write(Template(input_template).render(**input_dict))
 

arc/job/adapters/gaussian_test.py

@@ -187,7 +187,7 @@ def setUpClass(cls):
 
         # Gaussian: Checkfile error and SCF error
         # First SCF error - qc,nosymm
-        job_status = {'keywords': ['SCF']}
+        job_status = {'keywords': ['SCF', 'NoSymm']}
         output_errors, ess_trsh_methods, remove_checkfile, level_of_theory, software, job_type, fine, trsh_keyword, \
             memory, shift, cpu_cores, couldnt_trsh = trsh.trsh_ess_job(label, level_of_theory, server, job_status,
                                                                        job_type, software, fine, memory_gb,
@@ -297,7 +297,7 @@ def test_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc) cbs-qb3   IOp(2/9=2000) IOp(1/12=5,3/44=0)  
+#P opt=(calcfc)  cbs-qb3   IOp(2/9=2000) IOp(1/12=5,3/44=0)  
 
 spc1
 
@@ -333,7 +333,7 @@ def test_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(modredundant, calcfc, noeigentest, maxStep=5)  integral=(grid=ultrafine, Acc2E=12) guess=mix wb97xd/def2tzvp   IOp(2/9=2000)    scf=(tight, direct)
+#P opt=(calcfc,maxStep=5,modredundant,noeigentest) integral=(grid=ultrafine, Acc2E=12) guess=mix wb97xd/def2tzvp   IOp(2/9=2000)    scf=(direct,tight)
 
 ethanol
 
@@ -366,7 +366,7 @@ def test_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P  uwb97xd/def2tzvp freq IOp(7/33=1)  integral=(grid=ultrafine, Acc2E=12)  IOp(2/9=2000)    scf=(tight, direct)
+#P  uwb97xd/def2tzvp freq IOp(7/33=1)  integral=(grid=ultrafine, Acc2E=12)  IOp(2/9=2000)    scf=(direct,tight)
 
 birad_singlet
 
@@ -384,7 +384,7 @@ def test_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc) uwb97xd/def2tzvp   IOp(2/9=2000)   
+#P opt=(calcfc)  uwb97xd/def2tzvp   IOp(2/9=2000)   
 
 anion
 
@@ -478,7 +478,7 @@ def test_trsh_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc, tight, maxstep=5) uwb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)    scf=(tight,direct)
+#P opt=(calcfc,maxstep=5,tight)  uwb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)    scf=(direct,tight)
 
 anion
 
@@ -496,7 +496,7 @@ def test_trsh_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc, tight, maxstep=5) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)    scf=(tight,direct)
+#P opt=(calcfc,maxstep=5,tight) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)    scf=(direct,tight)
 
 ethanol
 
@@ -522,7 +522,7 @@ def test_trsh_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc, tight, maxstep=5) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)     scf=(tight,direct,xqc,nosymm)
+#P opt=(calcfc,maxstep=5,tight) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)     nosymm scf=(direct,tight,xqc)
 
 ethanol
 
@@ -548,7 +548,7 @@ def test_trsh_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc, tight, maxstep=5) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)     scf=(tight,direct,xqc,nosymm,NDump=30)
+#P opt=(calcfc,maxstep=5,tight) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)     nosymm scf=(NDump=30,direct,tight,xqc)
 
 ethanol
 
@@ -574,7 +574,7 @@ def test_trsh_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc, tight, maxstep=5) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)     scf=(tight,direct,xqc,nosymm,NDump=30,NoDIIS)
+#P opt=(calcfc,maxstep=5,tight) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)     nosymm scf=(NDump=30,NoDIIS,direct,tight,xqc)
 
 ethanol
 
@@ -600,7 +600,7 @@ def test_trsh_write_input_file(self):
 %mem=14336mb
 %NProcShared=8
 
-#P opt=(calcfc, tight, maxstep=5) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)   opt=(cartesian,nosymm)   scf=(tight,direct,xqc,nosymm,NDump=30,NoDIIS)
+#P opt=(calcfc,cartesian,maxstep=5,tight) guess=mix wb97xd  integral=(grid=ultrafine, Acc2E=14) IOp(2/9=2000)      nosymm scf=(NDump=30,NoDIIS,direct,tight,xqc)
 
 ethanol
 

arc/job/trsh.py

@@ -97,7 +97,7 @@ def determine_ess_status(output_path: str,
                         error = 'The blank line after the coordinate section is missing, ' \
                                 'or charge/multiplicity was not specified correctly.'
                     elif 'l103.exe' in line:
-                        keywords = ['InternalCoordinateError', 'GL103']
+                        keywords = ['InternalCoordinateError', 'GL103','NoSymm']
                         error = 'Internal coordinate error'
                     elif 'l108.exe' in line:
                         keywords = ['InputError', 'GL108']
@@ -112,7 +112,7 @@ def determine_ess_status(output_path: str,
                     elif 'l401.exe' in line:
                         keywords = ['GL401']
                     elif 'l502.exe' in line:
-                        keywords = ['SCF', 'GL502']
+                        keywords = ['SCF', 'GL502', 'NoSymm']
                         error = 'Unconverged SCF'
                     elif 'l508.exe' in line:
                         keywords = ['no_xqc', 'GL508']
@@ -867,10 +867,10 @@ def trsh_ess_job(label: str,
         if remove_checkfile:
              logger_info.append('that failed with "Basis set data is not on the checkpoint file" by removing the checkfile.')
 
-        # Check if InternalCoordinateError is in the keyword or opt=(cartesian,nosymm)
+        # Check if InternalCoordinateError is in the keyword or opt=(cartesian)
         ess_trsh_methods, trsh_keyword, couldnt_trsh = trsh_keyword_cartesian(job_status, ess_trsh_methods, job_type, trsh_keyword,couldnt_trsh)
         if 'cartesian' in ess_trsh_methods:
-            logger_info.append('using opt=cartesian with nosyym')
+            logger_info.append('using opt=cartesian')
         ess_trsh_methods, trsh_keyword, couldnt_trsh = trsh_keyword_intaccuracy(ess_trsh_methods, trsh_keyword, couldnt_trsh)
         if 'int=(Acc2E=14)' in ess_trsh_methods:
            logger_info.append('using int=(Acc2E=14)')
@@ -879,11 +879,16 @@ def trsh_ess_job(label: str,
         ess_trsh_methods, trsh_keyword, couldnt_trsh = trsh_keyword_scf(job_status, ess_trsh_methods, trsh_keyword, couldnt_trsh)
         if 'scf=(NDump=30)' in ess_trsh_methods:
             logger_info.append('using scf=(NDump=30)')
-        if 'scf=(qc,nosymm)' in ess_trsh_methods:
-            logger_info.append('using scf=(qc,nosymm)')
+        if 'scf=(qc)' in ess_trsh_methods:
+            logger_info.append('using scf=(qc)')
         if 'scf=(NoDIIS)' in ess_trsh_methods:
             logger_info.append('using scf=(NoDIIS)')
 
+        # Check if NoSymm
+        ess_trsh_methods, trsh_keyword, couldnt_trsh = trsh_keyword_nosymm(job_status, ess_trsh_methods, trsh_keyword, couldnt_trsh)
+        if 'NoSymm' in ess_trsh_methods:
+            logger_info.append('using nosymm')
+
         # Check if unconverged is in the keyword
         ess_trsh_methods, trsh_keyword, fine, couldnt_trsh = trsh_keyword_unconverged(job_status, ess_trsh_methods, trsh_keyword, couldnt_trsh, fine)
         if fine:
@@ -1540,18 +1545,25 @@ def trsh_keyword_checkfile(job_status, ess_trsh_methods, couldnt_trsh) -> Tuple[
     """
     Check if the job requires removal of checkfile
     """
-    if 'CheckFile' in job_status.get('keywords', '') or 'checkfile=None' in ess_trsh_methods:
+    if 'CheckFile' in job_status.get('keywords', '') and 'checkfile=None' not in ess_trsh_methods:
         ess_trsh_methods.append('checkfile=None')
         couldnt_trsh = False
         return True, ess_trsh_methods, couldnt_trsh
+    elif 'checkfile=None' in ess_trsh_methods:
+        couldnt_trsh = False
+        return True, ess_trsh_methods, couldnt_trsh
 
     return False, ess_trsh_methods, couldnt_trsh
 
 def trsh_keyword_intaccuracy(ess_trsh_methods, trsh_keyword, couldnt_trsh) -> Tuple[List, List, bool]:
     """
     Check if the job requires change of 2 electron integral accuracy
     """
-    if 'int=(Acc2E=14)' in ess_trsh_methods:
+    if 'int=(Acc2E=14)' not in ess_trsh_methods:
+        ess_trsh_methods.append('int=(Acc2E=14)')
+        trsh_keyword.append('int=(Acc2E=14)')
+        couldnt_trsh = False
+    elif 'int=(Acc2E=14)' not in trsh_keyword and 'int=(Acc2E=14)' in ess_trsh_methods:
         trsh_keyword.append('int=(Acc2E=14)')
         couldnt_trsh = False
 
@@ -1563,11 +1575,12 @@ def trsh_keyword_cartesian(job_status, ess_trsh_methods, job_type, trsh_keyword:
     """
     if 'InternalCoordinateError' in job_status['keywords'] \
                 and 'cartesian' not in ess_trsh_methods and job_type == 'opt':
-        trsh_keyword.append('opt=(cartesian,nosymm)')
+        ess_trsh_methods.append('cartesian')
+        trsh_keyword.append('opt=(cartesian)')
         couldnt_trsh = False
-    elif 'opt=(cartesian,nosymm)' in ess_trsh_methods and \
-            job_type == 'opt':
-        trsh_keyword.append('opt=(cartesian,nosymm)')
+    elif 'cartesian' in ess_trsh_methods and \
+            job_type == 'opt' and 'cartesian' not in trsh_keyword:
+        trsh_keyword.append('opt=(cartesian)')
         couldnt_trsh = False
 
     return ess_trsh_methods, trsh_keyword, couldnt_trsh
@@ -1577,9 +1590,9 @@ def trsh_keyword_scf(job_status, ess_trsh_methods, trsh_keyword, couldnt_trsh) -
     Check if the job requires change of scf
     """
     scf_pattern = r"scf=\((.*?)\)" # e.g., scf=(xqc,MaxCycle=1000), will match xqc,MaxCycle=1000
-    if 'SCF' in job_status['keywords'] and 'scf=(qc,nosymm)' not in ess_trsh_methods:
+    if 'SCF' in job_status['keywords'] and 'scf=(qc)' not in ess_trsh_methods:
         # try both qc and nosymm
-        ess_trsh_methods.append('scf=(qc,nosymm)')
+        ess_trsh_methods.append('scf=(qc)')
         couldnt_trsh = False
     elif 'SCF' in job_status['keywords'] and 'scf=(NDump=30)' not in ess_trsh_methods:
         # Switching off Pulay's Direct Inversion
@@ -1607,3 +1620,17 @@ def trsh_keyword_unconverged(job_status, ess_trsh_methods, trsh_keyword, couldnt
         fine = False
 
     return ess_trsh_methods, trsh_keyword, fine, couldnt_trsh
+
+def trsh_keyword_nosymm(job_status, ess_trsh_methods, trsh_keyword, couldnt_trsh) -> Tuple[List, List, bool]:
+    """
+    Check if the job requires change of nosymm
+    """
+    if 'NoSymm' in job_status['keywords'] and 'nosymm' not in ess_trsh_methods:
+        ess_trsh_methods.append('NoSymm')
+        trsh_keyword.append('nosymm')
+        couldnt_trsh = False
+    elif 'nosymm' not in trsh_keyword and any('NoSymm' in keyword for keyword in ess_trsh_methods):
+        trsh_keyword.append('nosymm')
+        couldnt_trsh = False
+
+    return ess_trsh_methods, trsh_keyword, couldnt_trsh

arc/job/trsh_test.py

@@ -304,14 +304,14 @@ def test_trsh_ess_job(self):
         self.assertFalse(couldnt_trsh)
 
         # Gaussian: test 2
-        job_status = {'keywords': ['InternalCoordinateError']}
+        job_status = {'keywords': ['InternalCoordinateError', 'NoSymm']}
         output_errors, ess_trsh_methods, remove_checkfile, level_of_theory, software, job_type, fine, trsh_keyword, \
             memory, shift, cpu_cores, couldnt_trsh = trsh.trsh_ess_job(label, level_of_theory, server, job_status,
                                                                        job_type, software, fine, memory_gb,
                                                                        num_heavy_atoms, cpu_cores, ess_trsh_methods)
 
         self.assertTrue(remove_checkfile)
-        self.assertEqual(trsh_keyword, ['opt=(cartesian,nosymm)', 'int=(Acc2E=14)'] )
+        self.assertEqual(trsh_keyword, ['opt=(cartesian)', 'int=(Acc2E=14)', 'nosymm'] )
 
         # Test Q-Chem
         software = 'qchem'