Parse VASP-ML output

pyiron_atomistics/vasp/parser/outcar.py

@@ -80,6 +80,10 @@ def from_file(self, filename="OUTCAR"):
         elapsed_time = self.get_elapsed_time(filename=filename, lines=lines)
         memory_used = self.get_memory_used(filename=filename, lines=lines)
         vasp_version = self.get_vasp_version(filename=filename, lines=lines)
+        full_array, ml_index, dft_index = self.get_index_type(
+            filename=filename,
+            lines=lines,
+        )  # ml and dft indices from all outcar indices
 
         try:
             (
@@ -97,14 +101,38 @@ def from_file(self, filename="OUTCAR"):
         )
         broyden_mixing = self.get_broyden_mixing_mesh(filename=filename, lines=lines)
         self.parse_dict["vasp_version"] = vasp_version
+
+        self.parse_dict["ml_index"] = ml_index
+        self.parse_dict["dft_index"] = dft_index
+        outcar_index = [steps[i] for i in np.union1d(dft_index, ml_index)]
+        is_ml_vec = full_array[outcar_index]
+        self.parse_dict["steps"] = np.arange(len(is_ml_vec))
+
+        self.parse_dict["ml_steps"] = self.parse_dict["steps"][is_ml_vec]
+        self.parse_dict["dft_steps"] = self.parse_dict["steps"][~is_ml_vec]
+
+        ## quantities from both DFT and ML
+        self.parse_dict["energies"] = [
+            energies[i] for i in np.union1d(dft_index, ml_index)
+        ]  # remove zeroes
+
+        ## quantites from DFT
+        self.parse_dict["dft_energies"] = [energies[i] for i in dft_index]
+        self.parse_dict["energies_int"] = [energies_int[i] for i in dft_index]
+        self.parse_dict["energies_zero"] = [energies_zero[i] for i in dft_index]
+        self.parse_dict["scf_energies"] = [scf_energies[i] for i in dft_index]
+
+        ## quantites from ML
+        self.parse_dict["ml_energies"] = [energies[i] for i in ml_index]
         self.parse_dict["energies"] = energies
         self.parse_dict["energies_int"] = energies_int
         self.parse_dict["energies_zero"] = energies_zero
         self.parse_dict["scf_energies"] = scf_energies
+
         self.parse_dict["forces"] = forces
         self.parse_dict["positions"] = positions
         self.parse_dict["cells"] = cells
-        self.parse_dict["steps"] = steps
+        # self.parse_dict["steps"] = steps
         self.parse_dict["temperatures"] = temperatures
         self.parse_dict["time"] = time
         self.parse_dict["fermi_level"] = fermi_level
@@ -443,7 +471,7 @@ def get_energy_without_entropy(filename="OUTCAR", lines=None):
         """
 
         def get_energy_without_entropy_from_line(line):
-            return float(_clean_line(line.strip()).split()[3])
+            return float(_clean_line(line.strip()).split("=")[1].split()[0])
 
         trigger_indices, lines = _get_trigger(
             lines=lines,
@@ -457,6 +485,34 @@ def get_energy_without_entropy_from_line(line):
             ]
         )
 
+    @staticmethod  # -> new function
+    def get_index_type(filename="OUTCAR", lines=None):
+        """
+        - In the outcar file you have two output for each DFT step (one of which is always zero for the energy and labelled ML) and one output for each ML step
+        - This function read all steps from the outcar file and then returns back what is the index of the DFT steps and the index of each ML step
+        - return indices of each DFT steps and index of DFT steps (excluding the empty ML output before DFT) from all steps
+        """
+        trigger_indices, lines = _get_trigger(
+            lines=lines,
+            filename=filename,
+            trigger="FREE ENERGIE OF THE ION-ELECTRON SYSTEM (eV)",
+        )
+        full_array = np.array(
+            ##boolean array, length: ml_steps + 2 x dft steps, values: true if ml false if dft and additional true for dummy ml steps printed withing dft steps
+            ["ML" in lines[j] for j in trigger_indices]
+        )
+        outcar_ml_index = np.setdiff1d(
+            np.where(full_array)[0], np.where(~full_array)[0] - 1
+        )  ## ml indices of of the full_array dummy ml in dft is removed
+        outcar_dft_index = np.where(~full_array)[0]  ## dft indiceds of the full_array
+
+        # outcar_index = [steps[i] for i in np.union1d(dft_index, ml_index)]
+        return full_array, outcar_ml_index, outcar_dft_index  # ml indices, dft indices
+
+        # full_array, ml_index, dft_index = self.get_index_type(filename=filename, lines=lines)  # ml and dft indices from all outcar indices
+        # outcar_index = [steps[i] for i in np.union1d(dft_index, ml_index)]
+        # is_ml_vec = full_array[outcar_index]
+
     @staticmethod
     def get_energy_sigma_0(filename="OUTCAR", lines=None):
         """