mlatom 3.16.1__tar.gz → 3.16.2__tar.gz

{mlatom-3.16.1/src/mlatom.egg-info → mlatom-3.16.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: mlatom
-Version: 3.16.1
+Version: 3.16.2
 Summary: A Package for AI-enhanced computational chemistry
 Home-page: http://mlatom.com
 Author: Pavlo O. Dral
@@ -27,7 +27,7 @@ Requires-Dist: tqdm
 # About Program
 MLatom: a Package for Atomistic Simulations with Machine Learning
 
-Version 3.16.1
+Version 3.16.2
 
 **Official website**: http://mlatom.com/    
 **Documentation**: http://mlatom.com/docs/                     
@@ -56,7 +56,7 @@ When this Software or its derivatives are used in scientific publications, it sh
 * Pavlo O. Dral, Fuchun Ge, Yi-Fan Hou, Peikun Zheng, Yuxinxin Chen, Mario Barbatti, Olexandr Isayev, Cheng Wang, Bao-Xin Xue, Max Pinheiro Jr, Yuming Su, Yiheng Dai, Yangtao Chen, Lina Zhang, Shuang Zhang, Arif Ullah, Quanhao Zhang, Yanchi Ou. [MLatom 3: A Platform for Machine Learning-enhanced Computational Chemistry Simulations and Workflows](https://doi.org/10.1021/acs.jctc.3c01203).   *J. Chem. Theory Comput.* **2024**, *20*, 1193--1213.                                                                                               
 * Pavlo O. Dral, Fuchun Ge, Bao-Xin Xue, Yi-Fan Hou, Max Pinheiro Jr, Jianxing Huang, Mario Barbatti. [MLatom 2: An Integrative Platform for Atomistic Machine Learning](http://doi.org/10.1007/s41061-021-00339-5). *Top. Curr. Chem.* **2021**, *379*, 27.       
 * Pavlo O. Dral. [MLatom: A Program Package for Quantum Chemical Research Assisted by Machine Learning](http://dx.doi.org/10.1002/jcc.26004). *J. Comput. Chem.* **2019**, *40*, 2339--2347.                                                                                      
-* Pavlo O. Dral, Fuchun Ge, Yi-Fan Hou, Peikun Zheng, Yuxinxin Chen, Bao-Xin Xue, Mikolaj Martyka, Max Pinheiro Jr, Yuming Su, Yiheng Dai, Yangtao Chen, Shuang Zhang, Lina Zhang, Arif Ullah, Quanhao Zhang, Sebastian V. Pios, Yanchi Ou, Matheus O. Bispo, Vignesh B. Kumar, MLatom: A Package for Atomistic Simulations with Machine Learning,  version 3.16.1, Xiamen University, Xiamen, China, 2013-2024.
+* Pavlo O. Dral, Fuchun Ge, Yi-Fan Hou, Peikun Zheng, Yuxinxin Chen, Bao-Xin Xue, Mikolaj Martyka, Max Pinheiro Jr, Yuming Su, Yiheng Dai, Yangtao Chen, Shuang Zhang, Lina Zhang, Arif Ullah, Quanhao Zhang, Sebastian V. Pios, Yanchi Ou, Matheus O. Bispo, Vignesh B. Kumar, Xin-Yu Tong, MLatom: A Package for Atomistic Simulations with Machine Learning,  version 3.16.2, Xiamen University, Xiamen, China, 2013-2024.
   [MLatom.com](http://mlatom.com).
                                                                       
 The citations for MLatom's interfaces and features shall be eventually
@@ -105,7 +105,7 @@ Citations mentioned above should be included. For convenience, below we provide
     }
 
     @misc{MLatomProg,
-    author = {Dral, Pavlo O. and Ge, Fuchun and Hou, Yi-Fan and Zheng, Peikun and Chen, Yuxinxin and Xue, Bao-Xin and Martyka, Mikolaj and Pinheiro Jr, Max and Su, Yuming and Dai, Yiheng and Chen, Yangtao and Zhang, Shuang and Zhang, Lina and Ullah, Arif and Zhang, Quanhao and Pios, Sebastian V. and Ou, Yanchi and Bispo, Matheus O. and Kumar, Vignesh B.},
+    author = {Dral, Pavlo O. and Ge, Fuchun and Hou, Yi-Fan and Zheng, Peikun and Chen, Yuxinxin and Xue, Bao-Xin and Martyka, Mikolaj and Pinheiro Jr, Max and Su, Yuming and Dai, Yiheng and Chen, Yangtao and Zhang, Shuang and Zhang, Lina and Ullah, Arif and Zhang, Quanhao and Pios, Sebastian V. and Ou, Yanchi and Bispo, Matheus O. and Kumar, Vignesh B. and Tong, Xin-Yu},
     title = {MLatom: A Package for Atomistic Simulations with Machine Learning},
     year = {2013--2024},
     type = {Computer Program}

{mlatom-3.16.1 → mlatom-3.16.2}/README.md

@@ -1,7 +1,7 @@
 # About Program
 MLatom: a Package for Atomistic Simulations with Machine Learning
 
-Version 3.16.1
+Version 3.16.2
 
 **Official website**: http://mlatom.com/    
 **Documentation**: http://mlatom.com/docs/                     
@@ -30,7 +30,7 @@ When this Software or its derivatives are used in scientific publications, it sh
 * Pavlo O. Dral, Fuchun Ge, Yi-Fan Hou, Peikun Zheng, Yuxinxin Chen, Mario Barbatti, Olexandr Isayev, Cheng Wang, Bao-Xin Xue, Max Pinheiro Jr, Yuming Su, Yiheng Dai, Yangtao Chen, Lina Zhang, Shuang Zhang, Arif Ullah, Quanhao Zhang, Yanchi Ou. [MLatom 3: A Platform for Machine Learning-enhanced Computational Chemistry Simulations and Workflows](https://doi.org/10.1021/acs.jctc.3c01203).   *J. Chem. Theory Comput.* **2024**, *20*, 1193--1213.                                                                                               
 * Pavlo O. Dral, Fuchun Ge, Bao-Xin Xue, Yi-Fan Hou, Max Pinheiro Jr, Jianxing Huang, Mario Barbatti. [MLatom 2: An Integrative Platform for Atomistic Machine Learning](http://doi.org/10.1007/s41061-021-00339-5). *Top. Curr. Chem.* **2021**, *379*, 27.       
 * Pavlo O. Dral. [MLatom: A Program Package for Quantum Chemical Research Assisted by Machine Learning](http://dx.doi.org/10.1002/jcc.26004). *J. Comput. Chem.* **2019**, *40*, 2339--2347.                                                                                      
-* Pavlo O. Dral, Fuchun Ge, Yi-Fan Hou, Peikun Zheng, Yuxinxin Chen, Bao-Xin Xue, Mikolaj Martyka, Max Pinheiro Jr, Yuming Su, Yiheng Dai, Yangtao Chen, Shuang Zhang, Lina Zhang, Arif Ullah, Quanhao Zhang, Sebastian V. Pios, Yanchi Ou, Matheus O. Bispo, Vignesh B. Kumar, MLatom: A Package for Atomistic Simulations with Machine Learning,  version 3.16.1, Xiamen University, Xiamen, China, 2013-2024.
+* Pavlo O. Dral, Fuchun Ge, Yi-Fan Hou, Peikun Zheng, Yuxinxin Chen, Bao-Xin Xue, Mikolaj Martyka, Max Pinheiro Jr, Yuming Su, Yiheng Dai, Yangtao Chen, Shuang Zhang, Lina Zhang, Arif Ullah, Quanhao Zhang, Sebastian V. Pios, Yanchi Ou, Matheus O. Bispo, Vignesh B. Kumar, Xin-Yu Tong, MLatom: A Package for Atomistic Simulations with Machine Learning,  version 3.16.2, Xiamen University, Xiamen, China, 2013-2024.
   [MLatom.com](http://mlatom.com).
                                                                       
 The citations for MLatom's interfaces and features shall be eventually
@@ -79,7 +79,7 @@ Citations mentioned above should be included. For convenience, below we provide
     }
 
     @misc{MLatomProg,
-    author = {Dral, Pavlo O. and Ge, Fuchun and Hou, Yi-Fan and Zheng, Peikun and Chen, Yuxinxin and Xue, Bao-Xin and Martyka, Mikolaj and Pinheiro Jr, Max and Su, Yuming and Dai, Yiheng and Chen, Yangtao and Zhang, Shuang and Zhang, Lina and Ullah, Arif and Zhang, Quanhao and Pios, Sebastian V. and Ou, Yanchi and Bispo, Matheus O. and Kumar, Vignesh B.},
+    author = {Dral, Pavlo O. and Ge, Fuchun and Hou, Yi-Fan and Zheng, Peikun and Chen, Yuxinxin and Xue, Bao-Xin and Martyka, Mikolaj and Pinheiro Jr, Max and Su, Yuming and Dai, Yiheng and Chen, Yangtao and Zhang, Shuang and Zhang, Lina and Ullah, Arif and Zhang, Quanhao and Pios, Sebastian V. and Ou, Yanchi and Bispo, Matheus O. and Kumar, Vignesh B. and Tong, Xin-Yu},
     title = {MLatom: A Package for Atomistic Simulations with Machine Learning},
     year = {2013--2024},
     type = {Computer Program}

{mlatom-3.16.1 → mlatom-3.16.2}/setup.py

@@ -43,7 +43,7 @@ with open(os.path.join(here, 'README.md')) as f:
 
 setup(
     name = pkg_name,
-    version = "3.16.1",
+    version = "3.16.2",
     author = "Pavlo O. Dral",
     author_email = "admin@mlatom.com",
     license = 'MIT (modified)',

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/MLatom.py

@@ -4,7 +4,7 @@
   !---------------------------------------------------------------------------!
   !                                                                           !
   !     MLatom: a Package for Atomistic Simulations with Machine Learning     !
-  !                             MLatom 3.16.1                                 !
+  !                             MLatom 3.16.2                                 !
   !                                   @                                       !
   !                 Xiamen Atomistic Computing Suite (XACS)                   !
   !                                                                           !
@@ -36,8 +36,9 @@
   ! Bao-Xin Xue, Mikolaj Martyka, Max Pinheiro Jr, Yuming Su, Yiheng Dai,     !
   ! Yangtao Chen, Shuang Zhang, Lina Zhang, Arif Ullah, Quanhao Zhang,        !
   ! Sebastian V. Pios, Yanchi Ou, Matheus O. Bispo, Vignesh B. Kumar,         !
+  ! Xin-Yu Tong,                                                              !
   ! MLatom: A Package for Atomistic Simulations with Machine Learning,        !
-  ! version 3.16.1, Xiamen University, Xiamen, China, 2013-2024.              !
+  ! version 3.16.2, Xiamen University, Xiamen, China, 2013-2024.              !
   !                                                                           !
   ! The citations for MLatom's interfaces and features shall be eventually    !
   ! included too. See header.py, ref.json and http://mlatom.com.              !

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/MLtasks.py

@@ -364,7 +364,15 @@ def slice(args):
     from . import sliceData
     sliceData.sliceDataCls(argsSD = args.args2pass)
 
-def geomopt(args):
+def optfreq(args):
+    molDB = geomopt(args, return_moldb=True)
+    freq(args, molDB=molDB)
+    
+def tsfreq(args):
+    molDB = ts(args, return_moldb=True)
+    freq(args, molDB=molDB)
+
+def geomopt(args, return_moldb=False):
     from . import simulations
     molDB = loading_data(XYZfile=args.XYZfile, charges=args.charges, multiplicities=args.multiplicities)
     model = loading_model(args)
@@ -419,13 +427,20 @@ def geomopt(args):
         else:
             print('\n Final energy of molecule %6d: %25.13f Hartree\n\n' % (imol+1, geomopt.optimized_molecule.energy))
     db_opt.write_file_with_xyz_coordinates(filename=fname)
+    if return_moldb:
+        return db_opt
 
-def ts(args):
-    geomopt(args)
+def ts(args, return_moldb=False):
+    if return_moldb:
+        molDB = geomopt(args, return_moldb=return_moldb)
+        return molDB
+    else:
+        geomopt(args, return_moldb=return_moldb)
 
-def freq(args):
+def freq(args, molDB=None):
     from . import simulations
-    molDB = loading_data(args.XYZfile, charges=args.charges, multiplicities=args.multiplicities)
+    if molDB is None:
+        molDB = loading_data(args.XYZfile, charges=args.charges, multiplicities=args.multiplicities)
     model = loading_model(args)
     kwargs = {}
     if args.freqProg:       kwargs['program'] = args.freqProg

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/args_class.py

@@ -695,6 +695,10 @@ class mlatom_args(ArgsBase):
             elif len(tasks) > 1:
                 if self.selfCorrect:
                     self._task = 'selfCorrect'
+                elif 'geomopt' in tasks and 'freq' in tasks:
+                    self._task = 'optfreq'
+                elif 'ts' in tasks and 'freq' in tasks:
+                    self._task = 'tsfreq'
                 else:
                     if 'useMLmodel' in tasks:
                         tasks.remove('useMLmodel')

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/data.py

@@ -5,7 +5,7 @@
   !---------------------------------------------------------------------------! 
   ! data: Module for working with data                                        ! 
   ! Implementations by: Pavlo O. Dral, Fuchun Ge,                             !
-  !                     Shuang Zhang, Yi-Fan Hou, Yanchi Ou                   !
+  !                     Shuang Zhang, Yi-Fan Hou, Yanchi Ou, Mikołaj Martyka  !
   !---------------------------------------------------------------------------! 
 '''
 
@@ -13,6 +13,7 @@ from __future__ import annotations
 from typing import Any, Union, Dict, List, Optional, Iterable
 import uuid, copy, os, json
 import numpy as np
+import math
 import h5py
 import functools
 from . import constants
@@ -142,6 +143,10 @@ def load_return_molecule(filename=None, format='json'):
         jsonfile = open(filename, 'r')
         moldict = json.load(jsonfile)
         newmol = dict_to_molecule_class_instance(moldict)
+    elif format.casefold() == 'xyz'.casefold():
+        newmol = molecule.from_xyz_file(filename)
+    elif format.casefold() == 'xyzstring'.casefold():
+        newmol = molecule.from_xyz_string(filename)
     elif format.casefold() == 'gaussian'.casefold():
         from .interfaces import gaussian_interface
         newmol = gaussian_interface.parse_gaussian_output(filename=filename)
@@ -663,7 +668,78 @@ class molecule:
             new_molecule = copy.deepcopy(self)
         new_molecule.id = str(uuid.uuid4())
         return new_molecule
+    def bond_length(self, a1, a2):
+        """Return the distance between atom numbers a1 and a2.
     
+        Atoms are numbered from zero.
+    
+        """
+        diff = self.atoms[a1].xyz_coordinates - self.atoms[a2].xyz_coordinates
+        return np.linalg.norm(diff)
+        
+    def bond_angle(self, a1, a2, a3, degrees=True):
+            """Return the bond angle a1-a2-a3.
+    
+            The angle is defined by the vectors a1-a2 and a2-a3.
+            Atoms are numbered from zero.
+            based on Tom Keal MNDOtools.py, October 2007
+            degrees - if true, return angle in degrees, else radians.
+    
+            """
+            # Based on the bond angle routine from geoman.f90 by Eduardo Fabiano
+            # vector 1 = 1->2
+            v1 = self.atoms[a2].xyz_coordinates - self.atoms[a1].xyz_coordinates
+            v1 = v1/np.linalg.norm(v1)
+    
+            # vector 2 = 2->3
+            v2 = self.atoms[a3].xyz_coordinates - self.atoms[a2].xyz_coordinates
+            v2 = v2/np.linalg.norm(v2)
+    
+            # dot product
+            dotp = np.dot(v1, v2)
+            # angle in radians
+            ang = math.pi - math.acos(dotp)
+            if degrees:
+                ang *= (180.0 / math.pi)
+            return ang
+    def dihedral_angle(self, a1, a2, a3, a4, degrees=True):
+        """Return the dihedral angle a1-a2-a3-a4.
+    
+        The angle is defined between the planes a1-a2-a3 and a2-a3-a4.
+        Atoms are numbered from zero.
+        based on Tom Keal MNDOtools.py, October 2007
+    
+        degrees - if true, return angle in degrees, else radians.
+    
+        """
+        # Based on the dihedral routine from geoman.f90 by Eduardo Fabiano
+        # vector 1 = 1->2
+        v1 = self.atoms[a2].xyz_coordinates - self.atoms[a1].xyz_coordinates
+        v1 = v1/np.linalg.norm(v1)
+    
+        # vector 2 = 2->3
+        v2 = self.atoms[a3].xyz_coordinates - self.atoms[a2].xyz_coordinates
+        v2 = v2/np.linalg.norm(v2)
+    
+        # vector 3 = 3->4
+        v3 = self.atoms[a4].xyz_coordinates - self.atoms[a3].xyz_coordinates
+        v3 = v3/np.linalg.norm(v3)
+    
+        # vector product 1 = v1^v2
+        w1 = np.cross(v1,v2)
+        w1 = w1/np.linalg.norm(w1)
+    
+        # vector product 2 = v3^v2
+        w2 = np.cross(v3,v2)
+        w2 = w2/np.linalg.norm(w2)
+    
+        # dot product
+        dotp = np.dot(w1,w2)
+        # angle in radians
+        ang = math.pi - math.acos(dotp)
+        if degrees:
+            ang *= (180.0 / math.pi)
+        return ang    
     def proliferate(
             self, 
             shifts: Optional[Iterable] = None, 
@@ -1663,14 +1739,21 @@ class molecular_database:
             jsonfile = open(filename, 'r')
             data = json.load(jsonfile)
             self.molecules = []
-            for molecule in data['molecules']:
-                self.molecules.append(dict_to_molecule_class_instance(molecule))
-        if format.casefold() == 'npz'.casefold():
+            for mol in data['molecules']:
+                self.molecules.append(dict_to_molecule_class_instance(mol))
+        elif format.casefold() == 'npz'.casefold():
             with np.load(filename, allow_pickle=True) as npz:
                 data = dict(npz)
                 self.molecules = []
-                for molecule in data['molecules']:
-                    self.molecules.append(dict_to_molecule_class_instance(molecule))
+                for mol in data['molecules']:
+                    self.molecules.append(dict_to_molecule_class_instance(mol))
+        elif format.casefold() == 'gaussian'.casefold():
+            mol = molecule.load(filename, format=format)
+            if 'molecular_database' in mol.__dict__.keys():
+                for key in mol.molecular_database.__dict__.keys():
+                    self.__dict__[key] = mol.molecular_database.__dict__[key]
+            else:
+                self.molecules.append(mol)
         return self
     
     @classmethod
@@ -1887,6 +1970,14 @@ def dict_to_molecule_class_instance(dd):
                 mol.atoms.append(dict_to_atom_class_instance(aa))
         elif key == 'electronic_states':
             mol.electronic_states = [dict_to_molecule_class_instance(state_dict) for state_dict in dd[key]]
+        elif key == 'molecular_database':
+            mol.molecular_database = molecular_database()
+            mol.molecular_database.molecules = [dict_to_molecule_class_instance(state_dict) for state_dict in dd[key]['molecules']]
+        elif key == 'optimization_trajectory':
+            mol.optimization_trajectory = molecular_trajectory()
+            mol.optimization_trajectory.steps = [molecular_trajectory_step(step=state_dict['step'],
+                                                                           molecule=dict_to_molecule_class_instance(state_dict['molecule']))
+                                                 for state_dict in dd[key]['steps']]
         elif type(dd[key]) == dict:
             if 'parent' in dd[key].keys():
                 dict_to_properties_tree_node_class_instance(dd, key, mol)    
@@ -2134,6 +2225,13 @@ class molecular_trajectory():
                     if not key in ['step', 'molecule']:
                         self.steps[-1].__dict__[key] = step[key]
     
+        elif format.casefold() == 'gaussian'.casefold():
+            mol = molecule.load(filename, format=format)
+            if 'optimization_trajectory' in mol.__dict__.keys():
+                for key in mol.optimization_trajectory.__dict__.keys():
+                    self.__dict__[key] = mol.optimization_trajectory.__dict__[key]
+            else:
+                raise ValueError('No optimization trajectory could be parsed from the Gaussian output file')
    
     def get_xyz_string(self) -> str:
         '''

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/interfaces/ase_interface.py

@@ -129,9 +129,11 @@ def dimer_method(initial_molecule, model,
     atoms.calc = MLatomCalculator(model=model,  model_predict_kwargs= model_predict_kwargs, save_optimization_trajectory=True)
 
     from ase.dimer import DimerControl, MinModeAtoms, MinModeTranslate
+    
+    random_seed = kwargs.pop('random_seed') if 'random_seed' in kwargs else None
 
     with DimerControl(**kwargs) as d_control:
-        d_atoms = MinModeAtoms(atoms, d_control, random_seed = kwargs['random_seed'] if 'random_seed' in kwargs else 0)
+        d_atoms = MinModeAtoms(atoms, d_control, random_seed = random_seed)
         d_atoms.displace()
         with MinModeTranslate(d_atoms) as dim_rlx:
             dim_rlx.run(fmax=convergence_criterion_for_forces,

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/interfaces/gaussian_interface.py

@@ -381,6 +381,12 @@ def parse_gaussian_output(filename=None, molecule=None):
     es_mults = []
     es_contribs = []
     readES = False
+    i_read_input_orientation = -1
+    i_read_standard_orientation = -1
+    xyz_input_orientation = []
+    xyz_standard_orientation = []
+    scf_energies = []
+    energy_gradients = []
     # flags to check whether a property was never read before - by default overwrite everything
     read_already_freq = False
     read_already_redmass = False
@@ -422,8 +428,35 @@ def parse_gaussian_output(filename=None, molecule=None):
                 chkfilename = line.split('=')[-1].strip()
                 if not '.chk' in chkfilename:
                     chkfilename += '.chk'
+            if 'Input orientation:' in line:
+                i_read_input_orientation = 0
+                xyz_input_orientation.append([])
+                continue
+            if i_read_input_orientation > -1:
+                i_read_input_orientation += 1
+                if i_read_input_orientation >= 5:
+                    if '-------------------------' in line:
+                        i_read_input_orientation = -1
+                        continue
+                    xyz_input_orientation[-1].append([float(xx) for xx in line.split()[-3:]])
+                    continue
+            if 'Standard orientation:' in line:
+                i_read_standard_orientation = 0
+                xyz_standard_orientation.append([])
+                continue
+            if i_read_standard_orientation > -1:
+                i_read_standard_orientation += 1
+                if i_read_standard_orientation >= 5:
+                    if '-------------------------' in line:
+                        i_read_standard_orientation = -1
+                        continue
+                    xyz_standard_orientation[-1].append([float(xx) for xx in line.split()[-3:]])
+                    continue
+            if 'SCF Done:' in line:
+                scf_energies.append(float(line.split()[4]))
             if 'Forces (Hartrees/Bohr)' in line:
                 forces_iatom = -3
+                energy_gradients.append([])
                 continue
             if forces_iatom > -4:
                 forces_iatom += 1
@@ -431,9 +464,7 @@ def parse_gaussian_output(filename=None, molecule=None):
                 if '------------------' in line:
                     forces_iatom = -4
                     continue
-                if len(mol.atoms) < forces_iatom+1:
-                    mol.atoms.append(data.atom(atomic_number=int(line.split()[1])))
-                mol.atoms[forces_iatom].energy_gradients = -data.array([float(each) for each in line.split()[2:]]) / constants.Bohr2Angstrom
+                energy_gradients[-1].append(-data.array([float(each) for each in line.split()[2:]]) / constants.Bohr2Angstrom)
                 continue
             if 'The second derivative matrix:' in line:
                 i_read_hessian = 0
@@ -815,10 +846,13 @@ def parse_gaussian_output(filename=None, molecule=None):
                 break
         if '\\\\@' in archiveText:
             break
+    optjob = False
     if archiveText != r'':
         method_names = ['HF', 'MP2', 'MP3', 'MP4D', 'MP4DQ', 'MP4SDQ', 'MP4SDTQ', 'MP5', 'CISD', 'QCISD', 'QCISD(T)', 'CCSD', 'CCSD(T)']
         method_energies = []
         archiveTextSplit = archiveText.split('\\\\') # here it splits with '\\' as a delimiter
+        if 'opt'.casefold() in archiveTextSplit[0].casefold():
+            optjob = True
         if 'freq'.casefold() in archiveTextSplit[1].casefold() and chkfilename is not None:
             dir_path = os.path.dirname(os.path.realpath(filename))
             if os.path.exists(f'{dir_path}/{chkfilename}'):
@@ -931,10 +965,76 @@ def parse_gaussian_output(filename=None, molecule=None):
     if 'infrared_intensities' in mol.__dict__.keys():
         if np.all(mol.infrared_intensities == 0):
             del mol.__dict__['infrared_intensities']
-        
-    # to-do: check whether gradients for electronic states are read correctly
-    # to-do: check whether the infrared_intensities have the correct length after anharmonic frequencies calculation
     
+    # delete the last geom if there were fewer SCF energies than geometries
+    # (can be due to failed SCF calculations or duplicated printing after successful geom opt)
+    if len(scf_energies) < len(xyz_input_orientation) and len(xyz_input_orientation) > 1:
+        if np.max(np.abs(data.array(xyz_input_orientation[-1]) - data.array(xyz_input_orientation[-2]))) < 1e-6:
+            del xyz_input_orientation[-1]
+    if len(scf_energies) < len(xyz_standard_orientation) and len(xyz_standard_orientation) > 1:
+        if np.max(np.abs(data.array(xyz_standard_orientation[-1]) - data.array(xyz_standard_orientation[-2]))) < 1e-6:
+            del xyz_standard_orientation[-1]
+    # if it was opt freq job, then geometries and scf energies are duplicated at the end too
+    if len(scf_energies) > 1:
+        if len(scf_energies) == len(xyz_input_orientation) and len(scf_energies) == len(xyz_standard_orientation):
+            if abs(scf_energies[-2] - scf_energies[-1]) < 1e-6 and np.max(np.abs(data.array(xyz_input_orientation[-1]) - data.array(xyz_input_orientation[-2]))) < 1e-6:
+                del scf_energies[-1]
+                del xyz_input_orientation[-1]
+                del xyz_standard_orientation[-1]
+            if len(energy_gradients) > len(scf_energies):
+                if np.max(np.abs(data.array(energy_gradients[-1]) - data.array(energy_gradients[-2]))) < 1e-6:
+                    del energy_gradients[-1]
+        elif len(scf_energies) == len(xyz_input_orientation):
+            if abs(scf_energies[-2] - scf_energies[-1]) < 1e-6 and np.max(np.abs(data.array(xyz_input_orientation[-1]) - data.array(xyz_input_orientation[-2]))) < 1e-6:
+                del scf_energies[-1]
+                del xyz_input_orientation[-1]
+            if len(energy_gradients) > len(scf_energies):
+                if np.max(np.abs(data.array(energy_gradients[-1]) - data.array(energy_gradients[-2]))) < 1e-6:
+                    del energy_gradients[-1]
+        elif len(scf_energies) == len(xyz_standard_orientation):
+            if abs(scf_energies[-2] - scf_energies[-1]) < 1e-6 and np.max(np.abs(data.array(xyz_standard_orientation[-1]) - data.array(xyz_standard_orientation[-2]))) < 1e-6:
+                del scf_energies[-1]
+                del xyz_standard_orientation[-1]
+            if len(energy_gradients) > len(scf_energies):
+                if np.max(np.abs(data.array(energy_gradients[-1]) - data.array(energy_gradients[-2]))) < 1e-6:
+                    del energy_gradients[-1]
+
+    # get molecular database for multiple single-point calculations in the output file
+    if len(scf_energies) > 1:
+        db = data.molecular_database()
+        db.molecules = [mol.copy(atomic_labels=[], molecular_labels=[]) for ii in range(len(scf_energies))]
+        for iscf in range(len(scf_energies)):
+            db.molecules[iscf].energy = scf_energies[iscf]
+        if len(scf_energies) == len(xyz_input_orientation):
+            for iscf in range(len(scf_energies)):
+                db.molecules[iscf].xyz_coordinates = data.array(xyz_input_orientation[iscf])
+        # to-do: reorient to input orientation using mol.xyz_coordinates which are already in input orientation for the last geometry
+        #elif len(scf_energies) == len(xyz_standard_orientation)
+        #    for iscf in range(len(scf_energies)):
+        #            db.molecules[iscf].xyz_coordinates = data.array(xyz_standard_orientation[iscf])
+        if len(scf_energies) == len(energy_gradients):
+            for iscf in range(len(scf_energies)):
+                db.molecules[iscf].energy_gradients = data.array(energy_gradients[iscf])
+        mol.molecular_database = db
+        if optjob:
+            moltraj = data.molecular_trajectory()
+            for istep, mol_step in enumerate(db):
+                moltraj.steps.append(data.molecular_trajectory_step(step=istep, molecule=mol_step))
+            mol.optimization_trajectory = moltraj
+
+    # get energy gradients from the last place they are calculated
+    if len(energy_gradients) > 0:
+        mol.energy_gradients = data.array(energy_gradients[-1])
+    
+    if 'molecular_database' in mol.__dict__.keys():
+        if np.max(np.abs(db[-1].xyz_coordinates - mol.xyz_coordinates)) < 1e-6:
+            tmpmol = mol.copy()
+            del tmpmol.__dict__['molecular_database']
+            mol.molecular_database.molecules[-1] = tmpmol
+            if 'optimization_trajectory' in tmpmol.__dict__.keys():
+                del tmpmol.__dict__['optimization_trajectory']
+                mol.optimization_trajectory.steps[-1].molecule = tmpmol
+                
     if molecule is None:
         return mol
 

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/interfaces/pyscf_interface.py

@@ -101,7 +101,7 @@ class pyscf_methods(OMP_pyscf, metaclass=models.meta_method, available_methods=[
         except:
             return False
             
-    def predict_for_molecule(self, molecule=None, calculate_energy=True, calculate_energy_gradients=True, calculate_hessian=False, calculate_dipole_derivatives=False, nstates=None, current_state=None, **kwargs):
+    def predict_for_molecule(self, molecule=None, calculate_energy=True, calculate_energy_gradients=True, calculate_hessian=False, calculate_dipole_derivatives=False, nstates=1, current_state=0, **kwargs):
         from pyscf import gto, scf
         from pyscf.dft.libxc import parse_xc
         pyscf_mol = gto.Mole()
@@ -367,7 +367,7 @@ class pyscf_methods(OMP_pyscf, metaclass=models.meta_method, available_methods=[
             raise ValueError(errmsg)
 
     @doc_inherit
-    def predict(self, molecule=None, molecular_database=None, calculate_energy=True, calculate_energy_gradients=False, calculate_hessian=False, nstates = None, current_state =None, **kwargs):
+    def predict(self, molecule=None, molecular_database=None, calculate_energy=True, calculate_energy_gradients=False, calculate_hessian=False, nstates=1, current_state=1, **kwargs):
         '''
             **kwargs: ``# needs to be documented``.
         '''

{mlatom-3.16.1 → mlatom-3.16.2}/src/mlatom/interfaces/torchani_interface.py

@@ -1138,9 +1138,11 @@ class msani(models.ml_model, models.torchani_model):
                         energy_loss= median(predicted_energies,true_energies)
                     else:
                         energy_loss = (loss_function(predicted_energies, true_energies, weightings_e) / num_atoms.sqrt()).nanmean()
-                    auxnumber = int(len(predicted_energies)/self.nstates)
-
-                    gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[:-auxnumber].sqrt()).nanmean()
+                    if self.hyperparameters.gap_coefficient != 0.0:
+                        gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[0].sqrt()).nanmean()
+                    else:
+                        gap_loss = 0
+                    # true_forces[true_
                     # true_forces[true_forces.isnan()]=forces[true_forces.isnan()]
                     force_loss = (loss_function(true_forces, forces, weightings_f).sum(dim=(1, 2)) / num_atoms).nanmean()
                     loss = energy_loss + self.hyperparameters.force_coefficient * force_loss + self.hyperparameters.gap_coefficient * gap_loss
@@ -1157,9 +1159,11 @@ class msani(models.ml_model, models.torchani_model):
                     predicted_gaps = torch.FloatTensor(predicted_gap_list).to(self.device)
                     
                     energy_loss = (loss_function(predicted_energies, true_energies, weightings_e) / num_atoms.sqrt()).nanmean()
-                    auxnumber = int(len(predicted_energies)/self.nstates)
-
-                    gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[:-auxnumber].sqrt()).nanmean()
+                    if self.hyperparameters.gap_coefficient != 0.0:
+                        gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[0].sqrt()).nanmean()
+                    else:
+                        gap_loss = 0
+                    # true_forces[true_
                     loss = energy_loss + self.hyperparameters.gap_coefficient * gap_loss
                     total_mse += loss.item()
             return total_mse
@@ -1234,8 +1238,11 @@ class msani(models.ml_model, models.torchani_model):
                         energy_loss = (loss_function(predicted_energies, true_energies, weightings_e) / num_atoms.sqrt()).nanmean()
                    # print(predicted_gaps.shape, true_gaps.shape)
                     #print(num_atoms.sqrt())
-                    auxnumber = int(len(predicted_energies)/self.nstates)
-                    gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[:-auxnumber].sqrt()).nanmean()
+                    if self.hyperparameters.gap_coefficient != 0.0:
+                        gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[0].sqrt()).nanmean()
+                    else:
+                        gap_loss = 0
+                    # true_forces[true_
                     # true_forces[true_forces.isnan()]=forces[true_forces.isnan()]
                     force_loss = (loss_function(true_forces, forces, weightings_f).sum(dim=(1, 2)) / num_atoms).nanmean()
                     loss = energy_loss + self.hyperparameters.force_coefficient * force_loss + self.hyperparameters.gap_coefficient * gap_loss
@@ -1251,8 +1258,11 @@ class msani(models.ml_model, models.torchani_model):
                     #print(predicted_gaps)
 
                     energy_loss = (loss_function(predicted_energies, true_energies, weightings_e) / num_atoms.sqrt()).nanmean()
-                    auxnumber = int(len(predicted_energies)/self.nstates)
-                    gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[:-auxnumber].sqrt()).nanmean()
+                    if self.hyperparameters.gap_coefficient != 0.0:
+                        gap_loss = (loss_function(predicted_gaps,true_gaps, 1) / num_atoms[0].sqrt()).nanmean()
+                    else:
+                        gap_loss = 0
+                    # true_forces[true_
                     loss = energy_loss + self.hyperparameters.gap_coefficient * gap_loss
 
                 self.AdamW.zero_grad()