JNB3-Build_traingset.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "b588ea61",
   "metadata": {},
   "source": [
    "# JNB3 - Build Training Set "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Welcome to this Jupyter Notebook. Here, we focus on building YAML files that encapsulate data essential for training.\n",
    "\n",
    "This notebook corresponds to the third step in the protocol outlined in the associated article, as detailed in the README.md file. For visual reference, please see Figure 3.\n",
    "\n",
    "**Sections Overview**\n",
    "\n",
    "- [Jobs & Engine Collecting](#jobs_engine_collecting): In this section, we gather DFT (Density Functional Theory) simulation data and the necessary settings for optimization. We maintain consistent settings across all ReaxFF optimization steps. During each step, we run all the systems using ReaxFF and compute the objective function.\n",
    "\n",
    "- [Data Set](#data_set): Here, we process the previously collected simulation data. We extract key quantities like energies, forces, charges, distances, etc., which are crucial for comparing ReaxFF predictions with DFT reference data. Subsequently, we divide this data into training and test sets.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<figure>\n",
    "    <img src=\"assets/img/wf-JNB3.png\" alt=\"JNB2 workflow\">\n",
    "    <figcaption><strong>Figure 3</strong>: Protocol workflow and part covered by this Jupyter Notebook highlighted in red.<figcaption>\n",
    "<figure> "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Settings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section, we define the constants and generators that will be used throughout the notebook:\n",
    "\n",
    "* `MAIN_DIR_JNB3`: The main folder where the results of this Jupyter Notebook will be stored.\n",
    "* `CHEM_FORMULA`: The chemical formula of the SEI component.\n",
    "* `TIMEZONE`: The timezone needed to save the time of post-processing into the files\n",
    "* `SEED`: Seed number for the random number generators\n",
    "\n",
    "These constants and generators play a crucial role in ensuring reproducibility and providing the necessary inputs for various computations and simulations in this notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pytz\n",
    "\n",
    "from tools.misc import MAIN_DIR_JNB1, MAIN_DIR_JNB2, MAIN_DIR_JNB3\n",
    "\n",
    "CHEM_FORMULA = \"LiF\"\n",
    "TIMEZONE = pytz.timezone(\"Europe/Rome\")\n",
    "SEED = 2023"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "93c8beee",
   "metadata": {},
   "outputs": [],
   "source": [
    "import datetime\n",
    "import glob\n",
    "import os\n",
    "import pickle\n",
    "\n",
    "import matplotlib as mp\n",
    "import matplotlib.gridspec as gridspec\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import scm.params as params\n",
    "import scm.plams as scm\n",
    "from prj_utils import (\n",
    "    get_all_angles,\n",
    "    get_all_bonds,\n",
    "    get_all_dihedrals,\n",
    "    get_distances_atom,\n",
    "    get_surf_atoms,\n",
    "    make_dir,\n",
    ")\n",
    "from pymatgen.core import Structure\n",
    "from pymatgen.io.ase import AseAtomsAdaptor\n",
    "from scm.plams.interfaces.molecule.ase import fromASE as ASEtoSCM\n",
    "from scm.plams.interfaces.molecule.ase import toASE as SCMtoASE\n",
    "\n",
    "matget2ase = AseAtomsAdaptor()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Making the output folder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "make_dir(os.path.join(MAIN_DIR_JNB3, CHEM_FORMULA))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aee6c96d",
   "metadata": {},
   "source": [
    "## Jobs & Engine Collecting  <a id='jobs_engine_collecting'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "b12e55b3",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobcollection = params.JobCollection()\n",
    "enginecollection = params.EngineCollection()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b07c76e",
   "metadata": {},
   "source": [
    "### Unit cells `Geometry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "2e59101c",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"0-UnitCell_GO\"\n",
    "TASK = \"GO\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "19512318",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1db9efe2",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "cd1571a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a0f269c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.BAND = settings.input.BAND.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"BAND_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88628255",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1f8a97ab",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"0-materialproject\")\n",
    "\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "537c89f0",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "36a96587",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_0 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_0[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9272f020",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_0)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(70) + \"| Status\")\n",
    "print(\"-\" * 85)\n",
    "for name, simulation in jobs_0.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(70) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d25b321c",
   "metadata": {},
   "source": [
    "### Pure Species `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "24193b2f",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"0.1-PureSpeices_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ff9a1564",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "b84b3945",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "b844180d",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b0768fe0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.BAND = settings.input.BAND.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"BAND_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8887c64a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a9030b81",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "a56c2e0c",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"0.1-pure_species\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc597e45",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "b50a7e17",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_01 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_01[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "37aaf041-0cd5-467f-8fad-420c9de66399",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(jobcollection)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8903916d",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_01)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(70) + \"| Status\")\n",
    "print(\"-\" * 85)\n",
    "for name, simulation in jobs_01.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(70) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d40ae4d8",
   "metadata": {},
   "source": [
    "### Supercells `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "9a45bb21",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"1-Supercells_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f5009d53",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "0b788b42",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "d932e4b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25c951d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.ADF = settings.input.ADF.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"ADF_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "63e6c950",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d66c4520",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "da6c2329",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7cff637d",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "903f8172",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_1 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_1[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a69d6452",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_1)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(70) + \"| Status\")\n",
    "print(\"-\" * 85)\n",
    "for name, simulation in jobs_1.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(70) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9d596dfc",
   "metadata": {},
   "source": [
    "### Supercells `Geometry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "c44519cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"1-Supercells_GO\"\n",
    "TASK = \"GO\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "94435c5d",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "a4a41dab",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "9802ef71",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "177acee5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.BAND = settings.input.BAND.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"BAND_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "06a759b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ad19d7e",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "40f5c736",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5bee7680",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "22e138fc",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_1_GO = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_1_GO[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45b1a21e",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_1_GO)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(70) + \"| Status\")\n",
    "print(\"-\" * 85)\n",
    "for name, simulation in jobs_1_GO.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(70) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56d96fd5",
   "metadata": {},
   "source": [
    "### Strain `Single Poin`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "40f0b9c4",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"2-Strain_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aaa0164d",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "8afe9056",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "7ce5d4ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7ae6bfc2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.ADF = settings.input.ADF.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"ADF_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "75361601",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "71919cbc",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "0064e14b",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"2-strain\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b4af112b",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "789fc71e",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_2 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_2[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "21ced63f",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_2)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_2.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab4bc4ae",
   "metadata": {},
   "source": [
    "### Vacancies `Single Poin`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "58641fad",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"3-Vacancies_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51b6a636",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "9f00e798",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "3286a3ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5063613f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.ADF = settings.input.ADF.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"ADF_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "65eb73f9",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "dfb467e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"3-vacancies\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aa6e063d",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "13ae12c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_3 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_3[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "769e76f1",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_3)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_3.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63fc7f93",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Substitution `Single Poin`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "2f6b1e82",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"4-Substitution_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "360714a6",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "64bac96a",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "6d52dc16",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25c951d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.ADF = settings.input.ADF.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"ADF_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b9776c52",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "722296d3",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"4-substitutions\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30fcb188",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "d8562484",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_4 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_4[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c954588f",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_4)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_4.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0fa0b58d",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Interstitial `Geopmetry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "0128d89b",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"5-Interstitial_GO\"\n",
    "TASK = \"GO\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1895c123",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "ac939a1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "e1127e87",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25c951d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.ADF = settings.input.ADF.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"ADF_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09d26a6e",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "19a799b2",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"5.1-interstitial_PO\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a95f40b",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "2b5b4ead",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_5 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_5[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6cfa85a2",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_5)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_5.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "38bb9218",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Slabs `Geopmetry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "5b074703",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"6-Slabs_GO\"\n",
    "TASK = \"GO\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "72bd65a6",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "246a129f",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "983b4e2a",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25c951d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.ADF = settings.input.ADF.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"ADF_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3d46a936",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "b15d83da",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"6.1-slabs_PO\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f9e7c965",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "553c1345",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_6 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"BAND_{SIMsNAME}\"\n",
    "    jobs_6[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "03f59a9f",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_6)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_6.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de697242",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Amorphous `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "25f95e77",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"7.1-Amorphous_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c348ddbc",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "9aa40eef",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "14351440",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25c951d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.BAND = settings.input.BAND.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"BAND_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0a8d8626",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "83396ad9",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"7.1-amorphous_frames\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0837fcdd",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "10eb95c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_7 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"ADF_{SIMsNAME}\"\n",
    "    jobs_7[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c7c1c47c",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_7)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_7.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de697242",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Bulk 300K `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "25f95e77",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"8.1-Bulk_300K_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c348ddbc",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "id": "9aa40eef",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "id": "14351440",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25c951d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.BAND = settings.input.BAND.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"BAND_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0a8d8626",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "id": "83396ad9",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"8.1-bulk_300K_frames\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0837fcdd",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "id": "10eb95c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_10 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"ADF_{SIMsNAME}\"\n",
    "    jobs_10[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c7c1c47c",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_10)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_10.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de697242",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Bulk 500 K `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "25f95e77",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIMsNAME = \"9.1-Bulk_500K_SP\"\n",
    "TASK = \"SP\"\n",
    "RUN_DIR = os.path.join(MAIN_DIR_JNB2, CHEM_FORMULA, SIMsNAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c348ddbc",
   "metadata": {},
   "source": [
    "Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "id": "9aa40eef",
   "metadata": {},
   "outputs": [],
   "source": [
    "jce_all = params.JCEntry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "14351440",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_pkl_file = os.path.join(RUN_DIR, \"settings.pkl\")\n",
    "with open(settings_pkl_file, \"rb\") as file:\n",
    "    settings = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25c951d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Engine\n",
    "e_all = params.Engine()\n",
    "e_all.settings.input.BAND = settings.input.BAND.copy()\n",
    "e_all.metadata[\"sim_set\"] = SIMsNAME\n",
    "enginecollection.add_entry(f\"BAND_{SIMsNAME}\", e_all)\n",
    "print(e_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "727366e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Job Settings\n",
    "jce_all.settings.input.AMS = settings.input.AMS.copy()\n",
    "jce_all.settings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0a8d8626",
   "metadata": {},
   "source": [
    "Load *intial configuration* file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "id": "83396ad9",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"9.1-bulk_500K_frames\")\n",
    "systems = []\n",
    "\n",
    "for path in np.sort(glob.glob(os.path.join(cis_dir, \"*.cif\"))):\n",
    "    system_name = path.split(os.sep)[-1].split(\".cif\")[0]\n",
    "    system_mat = Structure.from_file(path)\n",
    "    system_ase = matget2ase.get_atoms(system_mat)\n",
    "    systems.append((system_name, ASEtoSCM(system_ase)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0837fcdd",
   "metadata": {},
   "source": [
    "Load Job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "10eb95c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs_11 = {}\n",
    "for name, system_obj in systems:\n",
    "    sim_name = f\"{TASK}-{name}\"\n",
    "    dir_jobs = os.path.abspath(os.path.join(RUN_DIR, \"jobs\", sim_name))\n",
    "    jce = jce_all.copy()\n",
    "    if not os.path.isfile(os.path.join(dir_jobs, \"ams.rkf\")):\n",
    "        print(f\"NO ams.rkf file in {dir_jobs}\")\n",
    "        continue\n",
    "    job = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    # check\n",
    "    if not job.check():\n",
    "        print(f\"Simulation {dir_jobs} FAILED\")\n",
    "        continue\n",
    "    # job_path = os.path.join(ALLMAIN_DIR_JNB2, f'{sim_name}')\n",
    "    jce.molecule = system_obj.copy()\n",
    "    jce.metadata[\"sim_set\"] = SIMsNAME\n",
    "    jce.metadata[\"dir_job\"] = dir_jobs\n",
    "    jce.reference_engine = f\"ADF_{SIMsNAME}\"\n",
    "    jobs_11[sim_name] = scm.AMSJob.load_external(dir_jobs, settings=jce.settings)\n",
    "    jobcollection.add_entry(sim_name, jce.copy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c7c1c47c",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Jobs added {len(jobs_11)}, total {len(jobcollection)}\\n\")\n",
    "print(\"  (item name) == (simulation name);\".ljust(90) + \"| Status\")\n",
    "print(\"-\" * 105)\n",
    "for name, simulation in jobs_11.items():\n",
    "    print(f\"  {name} == {simulation.name};\".ljust(90) + f\"| Status={simulation.check()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2ca3eb6b",
   "metadata": {},
   "source": [
    "### Storeing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "054bfdfa",
   "metadata": {},
   "source": [
    "save `EngineCollection`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "id": "45f3ccc6",
   "metadata": {},
   "outputs": [],
   "source": [
    "now = datetime.datetime.now(TIMEZONE)\n",
    "comments = dict(\n",
    "    info=f\"\"\"Engine Collection of all jobs\n",
    "@date = {now.strftime(\"%Y-%m-%d %H:%M:%S %Z\")} ({TIMEZONE.zone})\n",
    "@authors = Paolo De Angelis\n",
    "@mail = paolo.deangelis@polito.it\n",
    "@PLAMS = {scm.__version__}\n",
    "@ParAMS = {params.__version__}\"\"\"\n",
    ")\n",
    "enginecollection.header = comments\n",
    "enginecollection.store(os.path.join(MAIN_DIR_JNB3, \"engine_collection.yaml\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e864bb73",
   "metadata": {},
   "source": [
    "save `JobCollection`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "id": "85158f79",
   "metadata": {},
   "outputs": [],
   "source": [
    "now = datetime.datetime.now(TIMEZONE)\n",
    "comments = dict(\n",
    "    info=f\"\"\"Simulations Collection of all jobs\n",
    "@date = {now.strftime(\"%Y-%m-%d %H:%M:%S %Z\")} ({TIMEZONE.zone})\n",
    "@authors = Paolo De Angelis\n",
    "@mail = paolo.deangelis@polito.it\n",
    "@PLAMS = {scm.__version__}\n",
    "@ParAMS = {params.__version__}\"\"\"\n",
    ")\n",
    "jobcollection.header = comments\n",
    "jobcollection.store(os.path.join(MAIN_DIR_JNB3, \"jobs_collection.yaml\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "id": "e2089462",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/paolo/Desktop/Batteries/Tier0-C+EC-DMC+Li/venv_ams2022/AMS2022.1.venv/lib/python3.6/site-packages/ipykernel_launcher.py:1: DeprecationWarning:\n",
      "\n",
      "Pickling will be removed in a future version. Please use the yaml io instead\n",
      "\n"
     ]
    }
   ],
   "source": [
    "jobcollection.pickle_dump(os.path.join(MAIN_DIR_JNB3, \"jobs_collection.pkl\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fee2557e",
   "metadata": {},
   "source": [
    "## Data Set <a id='data_set'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "id": "75e1e38d",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = params.DataSet()\n",
    "dataset_df = pd.DataFrame(columns=[\"chem_formula\", \"systems_name\", \"entry_type\", \"entry_weight\", \"entry_sigma\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "460ea3b7",
   "metadata": {},
   "source": [
    "### Unit cells `Geometry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "id": "3777ae24",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Unit cells (GO)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed386f32",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "id": "5b085cfd",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"0-materialproject\")\n",
    "data_path = os.path.join(cis_dir, \"properties.json\")\n",
    "# data_path = os.path.join(ALLMAIN_DIR_JNB2, f'results_0-UnitCell_SP.json')\n",
    "properties_df = pd.read_json(data_path)\n",
    "referece_j_name = properties_df.loc[properties_df[\"energy_per_atom\"] == properties_df[\"energy_per_atom\"].min()].index\n",
    "data_j_names = properties_df.loc[properties_df[\"energy_per_atom\"] != properties_df[\"energy_per_atom\"].min()].index\n",
    "\n",
    "referece_job_name = \"GO-\" + referece_j_name[0][:-4]\n",
    "Natoms_ref = len(jobs_0[referece_job_name].molecule.atoms)\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "for indx in data_j_names:\n",
    "    data_job_name = \"GO-\" + indx[:-4]\n",
    "    Natoms_data = len(jobs_0[data_job_name].molecule.atoms)\n",
    "    entries_str.append(f\" energy('{data_job_name}') / {Natoms_data} - energy('{referece_job_name}') / {Natoms_ref} \")\n",
    "    weight_all.append(5.0 / (float(indx.split(\"-\")[0]) + 1.0))\n",
    "    sigma_all.append(1e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "97764a4b",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "id": "1279282b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.0\n",
    "# sigma = 1.0\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a925e00",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "id": "18e3fd77",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2bcd066",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "08ead6be",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5af942a2",
   "metadata": {},
   "source": [
    "##### Charge"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "id": "aa6a3fe4",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "for name, job in jobs_0.items():\n",
    "    Natoms_data = len(job.molecule.atoms)\n",
    "    for ai in range(Natoms_data):\n",
    "        entries_str.append(f\"charges('{name}', {ai})\")\n",
    "        weight_all.append(2.0 / (float(name.split(\"-\")[1]) + 1.0))\n",
    "        sigma_all.append(1e-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85c1b1e3",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "id": "456b184e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 0.75\n",
    "# sigma = 1.0\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma)\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Charges\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e641ebd5",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "id": "50ae543c",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ffe7e1d4",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a9596021",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8123d356",
   "metadata": {},
   "source": [
    "##### Lattice Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "id": "accbb6f9",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "for name, job in jobs_0.items():\n",
    "    entries_str.append(f\"lattice('{name}')\")\n",
    "    weight_all.append(10.0 / (float(name.split(\"-\")[1]) + 1.0))\n",
    "    sigma_all.append(1e-2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e110c65e",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "id": "abe95137",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 10.0\n",
    "# sigma = 1.0\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Lattice\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1e0a18e0",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "id": "6ce47466",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eba56bdc",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "97ea5fbf",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b355ccd",
   "metadata": {},
   "source": [
    "##### Hessian"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "id": "cc525681",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for name, job in jobs_0.items():\n",
    "    entries_str.append(f\"hessian('{name}')\")\n",
    "    weight_all.append(5.0 / (float(name.split(\"-\")[1]) + 1.0))\n",
    "    sigma_all.append(1.0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b5466f9",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "id": "9761502c",
   "metadata": {},
   "outputs": [],
   "source": [
    "unit = (\"au/Ang^2\", scm.Units.convert(1, \"au/Ang^2\", \"au/Ang^2\"))\n",
    "\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Hessian\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2b2583a6",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "id": "006eb5de",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed051ce2",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5f594340",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bff69338",
   "metadata": {},
   "source": [
    "##### Stress Tensor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "id": "f2887f25",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for name, job in jobs_0.items():\n",
    "    Natoms_data = len(job.molecule.atoms)\n",
    "    entries_str.append(f\"stresstensor('{name}')/{Natoms_data}\")\n",
    "    weight_all.append(10.0 / (float(name.split(\"-\")[1]) + 1))\n",
    "    sigma_all.append(1e-2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a8144be2",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "id": "2d8a18da",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.5\n",
    "# sigma = 1.0\n",
    "# unit = ('kJ/mol/Ang^3', scm.Units.convert(1, 'au/Ang^3', 'kJ/mol/Ang^3'))\n",
    "unit = (\"au/Ang^3\", scm.Units.convert(1, \"au/Ang^3\", \"au/Ang^3\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Stress Tensor\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f30ac260",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "423a56d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "33742d43",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9ee64a1d",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "76c3c997",
   "metadata": {},
   "source": [
    "##### Stress Tensor (relative)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "69460e96",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"0-materialproject\")\n",
    "data_path = os.path.join(cis_dir, \"properties.json\")\n",
    "# data_path = os.path.join(ALLMAIN_DIR_JNB2, f'results_0-UnitCell_SP.json')\n",
    "properties_df = pd.read_json(data_path)\n",
    "referece_j_name = properties_df.loc[properties_df[\"energy_per_atom\"] == properties_df[\"energy_per_atom\"].min()].index\n",
    "data_j_names = properties_df.loc[properties_df[\"energy_per_atom\"] != properties_df[\"energy_per_atom\"].min()].index\n",
    "\n",
    "referece_job_name = \"GO-\" + referece_j_name[0][:-4]\n",
    "Natoms_ref = len(jobs_0[referece_job_name].molecule.atoms)\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "for indx in data_j_names:\n",
    "    data_job_name = \"GO-\" + indx[:-4]\n",
    "    Natoms_data = len(jobs_0[data_job_name].molecule.atoms)\n",
    "    entries_str.append(\n",
    "        f\"stresstensor('{data_job_name}') /{Natoms_data} - stresstensor('{referece_job_name}') /  {Natoms_ref} \"\n",
    "    )\n",
    "    weight_all.append(2.5 / (float(indx.split(\"-\")[0]) + 1))\n",
    "    sigma_all.append(1e-2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "739014c8",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "id": "b4f2e41a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 2.5\n",
    "# sigma = 1.0\n",
    "# # unit = ('kJ/mol/Ang^3', scm.Units.convert(1, 'au/Ang^3', 'kJ/mol/Ang^3'))\n",
    "# unit = ('au/Ang^3', scm.Units.convert(1, 'au/Ang^3', 'au/Ang^3'))\n",
    "\n",
    "\n",
    "# for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "#     dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "#     dataset_df.loc[N_df, 'entry_sigma'] = sigma\n",
    "#     dataset_df.loc[N_df, 'chem_formula'] = CHEM_FORMULA\n",
    "#     dataset_df.loc[N_df, 'systems_name'] = systems_name\n",
    "#     dataset_df.loc[N_df, 'entry_type'] = 'Stress Tensor'\n",
    "#     dataset_df.loc[N_df, 'entry_weight'] = weight\n",
    "#     N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5622326c",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "id": "8b65b51e",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8b695085",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6cf0c8da",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7f02019f",
   "metadata": {},
   "source": [
    "##### Bond distance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "34139ce3",
   "metadata": {},
   "source": [
    "get intila configurations for each job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "id": "48fc92b9",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol = {}\n",
    "\n",
    "for name, job in jobs_0.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed8d57f4",
   "metadata": {},
   "source": [
    "get molecule bonds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "id": "603b86f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol_bond = {}\n",
    "\n",
    "for name, ase_mol in data_mol.items():\n",
    "    data_mol_bond[name] = get_all_bonds(ase_mol, verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a979059c",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol_bond"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "id": "9a137fb8",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "max_bond_per_type = 10\n",
    "\n",
    "data_bond_indx = {}\n",
    "for name, bond_df in data_mol_bond.items():\n",
    "    data_bond_indx[name] = []\n",
    "    for indx in bond_df.index:\n",
    "        bonds = bond_df.loc[indx, \"bonds\"]\n",
    "        data_bond_indx[name] += bonds"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99d86d1d",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "id": "e68377b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_0\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_bonds in data_bond_indx.items():\n",
    "    for ai, aj in data_bonds:\n",
    "        entries_str.append(f\"distance('{data_name}', {ai}, {aj})\")\n",
    "        weight_all.append(0.2 / (float(data_name.split(\"-\")[1]) + 1.0))\n",
    "        sigma_all.append(1e-2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b44c052e",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "id": "4b3444ac",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .75\n",
    "# sigma = 1.\n",
    "# unit = ('Angstrom', scm.Units.convert(1, 'au', 'Angstrom'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Distance\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ff7a305",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "id": "31d3695d",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "97f900b8",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "248904df",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "87e6f421",
   "metadata": {},
   "source": [
    "##### Forces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "id": "bc6cefc4",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for name, job in jobs_0.items():\n",
    "    Natoms_data = len(job.molecule.atoms)\n",
    "    for ai in range(Natoms_data):\n",
    "        entries_str.append(f\"forces('{name}', {ai})\")\n",
    "        weight_all.append(0.2 / (float(name.split(\"-\")[1]) + 1))\n",
    "        sigma_all.append(5e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "648f994c",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "id": "43757adc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .15\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol/Ang', scm.Units.convert(1, 'au/Ang', 'kJ/mol/Ang'))\n",
    "unit = (\"au/Ang\", scm.Units.convert(1, \"au/Ang\", \"au/Ang\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Forces\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a68e7f99",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "id": "07cdb6a8",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1d29544a",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e4a00766",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "110e6888",
   "metadata": {},
   "source": [
    "### Supercells `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "id": "12b4628b",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Supercells (SP)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3e09734b",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "id": "6de431ec",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = jobs_0\n",
    "data_j = jobs_1\n",
    "entries_str = []\n",
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_path = os.path.join(cis_dir, \"properties.json\")\n",
    "# data_path = os.path.join(ALLMAIN_DIR_JNB2, f'results_0-UnitCell_SP.json')\n",
    "properties_df = pd.read_json(data_path)\n",
    "\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[1])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            scale = properties_df.loc[data_name[3:] + \".cif\", \"unitcell_repetition\"]\n",
    "            entries_str.append(f\"energy('{data_name}')/{scale} - energy('{ref_name}') \")\n",
    "            weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4ddb189c",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "id": "2cd9b8a2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.0\n",
    "# sigma = 1.0\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fb70516f",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "id": "926bbb95",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b4efbfb",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9b479ffc",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "69f816f0",
   "metadata": {},
   "source": [
    "##### Stress Tensor (relative)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "id": "aacf4e9f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "referece_j = jobs_0\n",
    "data_j = jobs_1\n",
    "entries_str = []\n",
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_path = os.path.join(cis_dir, \"properties.json\")\n",
    "# data_path = os.path.join(ALLMAIN_DIR_JNB2, f'results_0-UnitCell_SP.json')\n",
    "properties_df = pd.read_json(data_path)\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[1])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            scale = properties_df.loc[data_name[3:] + \".cif\", \"unitcell_repetition\"]\n",
    "            entries_str.append(f\"stresstensor('{data_name}'){scale} - stresstensor('{ref_name}')\")\n",
    "            weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(1.0e-2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "41ed079f",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "id": "a8e890da",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1/1.5\n",
    "# sigma = 1.0\n",
    "# # unit = ('kJ/mol/Ang^3', scm.Units.convert(1, 'au/Ang^3', 'kJ/mol/Ang^3'))\n",
    "# unit = ('au/Ang^3', scm.Units.convert(1, 'au/Ang^3', 'au/Ang^3'))\n",
    "\n",
    "# for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "#     dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "#     dataset_df.loc[N_df, 'entry_sigma'] = sigma\n",
    "#     dataset_df.loc[N_df, 'chem_formula'] = CHEM_FORMULA\n",
    "#     dataset_df.loc[N_df, 'systems_name'] = systems_name\n",
    "#     dataset_df.loc[N_df, 'entry_type'] = 'Stress Tensor'\n",
    "#     dataset_df.loc[N_df, 'entry_weight'] = weight\n",
    "#     N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a496475e",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "id": "ad9e5891",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(jobs_0)\n",
    "all_jobs.update(jobs_1)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "81c715db",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "284360cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09ddc223",
   "metadata": {},
   "source": [
    "### Supercells `Geometry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "id": "d0866982",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Supercells (GO)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c53c297",
   "metadata": {},
   "source": [
    "##### Bond distance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f93394ec",
   "metadata": {},
   "source": [
    "get intila configurations for each job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "id": "9f969890",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = jobs_0\n",
    "data_j = jobs_1_GO\n",
    "\n",
    "data_mol = {}\n",
    "\n",
    "for name, job in data_j.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74c9f6c9",
   "metadata": {},
   "source": [
    "get molecule bonds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "id": "050a5187",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol_bond = {}\n",
    "\n",
    "for name, ase_mol in data_mol.items():\n",
    "    data_mol_bond[name] = get_all_bonds(ase_mol, verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "id": "404cfa57",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "max_bond_per_type = 10\n",
    "\n",
    "data_bond_indx = {}\n",
    "for name, bond_df in data_mol_bond.items():\n",
    "    for indx in bond_df.index:\n",
    "        bonds = bond_df.loc[indx, \"bonds\"]\n",
    "        if len(bonds) > max_bond_per_type:\n",
    "            rnd_indexs = np.arange(bond_df.loc[indx, \"Nbonds\"])\n",
    "            rg.shuffle(rnd_indexs)\n",
    "            rnd_indexs = rnd_indexs[:max_bond_per_type]\n",
    "            data_bond_indx[name] = [bonds[bi] for bi in rnd_indexs]\n",
    "        else:\n",
    "            data_bond_indx[name] = bonds"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "567a8fd0",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "id": "a0c37c8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_1_GO\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_bonds in data_bond_indx.items():\n",
    "    for ai, aj in data_bonds:\n",
    "        entries_str.append(f\"distance('{data_name}', {ai}, {aj})\")\n",
    "        weight_all.append(0.2 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "        sigma_all.append(1e-2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e646878",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "id": "d3c0bf0e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .1\n",
    "# sigma = 1.\n",
    "# unit = ('Angstrom', scm.Units.convert(1, 'au', 'Angstrom'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Distance\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f265af2c",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "id": "0a76d5d3",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "38cd8ccb",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ee9e7e7f",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c120e05b",
   "metadata": {},
   "source": [
    "##### Angles"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "df0d273e",
   "metadata": {},
   "source": [
    "get intila configurations for each job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "id": "2078d0a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = jobs_0\n",
    "data_j = jobs_1_GO\n",
    "\n",
    "data_mol = {}\n",
    "\n",
    "for name, job in data_j.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9993a0bc",
   "metadata": {},
   "source": [
    "get molecule angle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "id": "4f7dbc23",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol_angle = {}\n",
    "\n",
    "for name, ase_mol in data_mol.items():\n",
    "    data_mol_angle[name] = get_all_angles(ase_mol, verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "id": "c1e3a891",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "max_angle_per_type = 10\n",
    "\n",
    "data_angle_indx = {}\n",
    "for name, angle_df in data_mol_angle.items():\n",
    "    for indx in angle_df.index:\n",
    "        angles = angle_df.loc[indx, \"angles\"]\n",
    "        if len(angles) > max_angle_per_type:\n",
    "            rnd_indexs = np.arange(angle_df.loc[indx, \"Nangles\"])\n",
    "            rg.shuffle(rnd_indexs)\n",
    "            rnd_indexs = rnd_indexs[:max_angle_per_type]\n",
    "            data_angle_indx[name] = [angles[ani] for ani in rnd_indexs]\n",
    "        else:\n",
    "            data_angle_indx[name] = angles"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4e1852c8",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "id": "05738b43",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_1_GO\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_angles in data_angle_indx.items():\n",
    "    for ai, aj, ak in data_angles:\n",
    "        entries_str.append(f\"angles('{data_name}', ({ai}, {aj}, {ak}) )\")\n",
    "        weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "        sigma_all.append(3.14)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2e45470f",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "id": "ef659648",
   "metadata": {},
   "outputs": [],
   "source": [
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Angles\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c57ccc95",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "id": "be98999b",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8610945f",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f98ca2f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "44e3acab",
   "metadata": {},
   "source": [
    "##### Dihedrals"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c9783629",
   "metadata": {},
   "source": [
    "get intila configurations for each job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "id": "54452445",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = jobs_0\n",
    "data_j = jobs_1_GO\n",
    "\n",
    "data_mol = {}\n",
    "\n",
    "for name, job in data_j.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e18ca5fa",
   "metadata": {},
   "source": [
    "get molecule dihedrals"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "id": "dbfadd73",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol_dihedrals = {}\n",
    "\n",
    "for name, ase_mol in data_mol.items():\n",
    "    data_mol_dihedrals[name] = get_all_dihedrals(ase_mol, verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "id": "e25dec1e",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "max_dihedral_per_type = 12\n",
    "\n",
    "data_dihedral_indx = {}\n",
    "for name, dihedral_df in data_mol_dihedrals.items():\n",
    "    for indx in dihedral_df.index:\n",
    "        dihedrals = dihedral_df.loc[indx, \"dihedrals\"]\n",
    "        if len(dihedrals) > max_dihedral_per_type:\n",
    "            rnd_indexs = np.arange(dihedral_df.loc[indx, \"Ndihedrals\"])\n",
    "            rg.shuffle(rnd_indexs)\n",
    "            rnd_indexs = rnd_indexs[:max_dihedral_per_type]\n",
    "            data_dihedral_indx[name] = [dihedrals[di] for di in rnd_indexs]\n",
    "        else:\n",
    "            data_dihedral_indx[name] = dihedrals"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f0485853",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "id": "c97b705d",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_1_GO\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_dihedral in data_dihedral_indx.items():\n",
    "    for ai, aj, ak, al in data_dihedral:\n",
    "        entries_str.append(f\"dihedral('{data_name}', {ai}, {aj}, {ak}, {al} )\")\n",
    "        weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1) ** 2)\n",
    "        sigma_all.append(5e0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5103da7f",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "id": "3afcff36",
   "metadata": {},
   "outputs": [],
   "source": [
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Dihedrals\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "62e6d271",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e5bdbb4c",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c1a3ec8",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ea58e366",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e05b51ea",
   "metadata": {},
   "source": [
    "### Strain `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "id": "1197e27a",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Strain (SP)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3207d653",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b6c66f5",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "id": "d245b0a9",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"2-strain\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "id": "3ff45010",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_2\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            entries_str.append(f\"energy('{data_name}') - energy('{ref_name}') \")\n",
    "            weight_all.append(10000.0 / (float(data_name.split(\"-\")[2]) + 1.0) ** 2)\n",
    "            sigma_all.append(1e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cf724391",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "id": "e71027d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 2.5\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "81758621",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "id": "588676d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93776a4e",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d273488d",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c31cd302",
   "metadata": {},
   "source": [
    "##### Forces"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "28737b7f",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "id": "53bda048",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"2-strain\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "id": "a5ba37ae",
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "max_forces_per_type = 8\n",
    "\n",
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_2\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    Natoms_data = len(ref_job.molecule.atoms)\n",
    "    if Natoms_data > max_forces_per_type:\n",
    "        rnd_indexs = np.arange(Natoms_data)\n",
    "        rg.shuffle(rnd_indexs)\n",
    "        rnd_indexs = rnd_indexs[:max_forces_per_type]\n",
    "    else:\n",
    "        rnd_indexs = np.arange(Natoms_data)\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            for ai in rnd_indexs:\n",
    "                entries_str.append(f\"forces('{data_name}', {ai}) - forces('{ref_name}', {ai})\")\n",
    "                weight_all.append(0.2 / (float(data_name.split(\"-\")[2]) + 1))\n",
    "                sigma_all.append(5e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "65c07711",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "id": "2388b8f8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .25\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol/Ang', scm.Units.convert(1, 'au/Ang', 'kJ/mol/Ang'))\n",
    "unit = (\"au/Ang\", scm.Units.convert(1, \"au/Ang\", \"au/Ang\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Forces\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1e3ae754",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "id": "8524f074",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8771e865",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "78563a67",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7bd6d952",
   "metadata": {},
   "source": [
    "### Vacancies `Single Poin`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "id": "c3ad5b12",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Vacancies (SP)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a6f61c70",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d263aec",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "id": "d7ca6eea",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"3-vacancy\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"0.1-pure_species\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_elem_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "id": "681012d4",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_3\n",
    "elemets_j = jobs_01\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            vacancy_atoms_str = \"\"\n",
    "            for sim_name, el in zip(elemets_j.keys(), properties_elem_df[\"element\"]):\n",
    "                N_v = properties_data_df.loc[data_name[3:] + \".cif\", \"vacancy_element\"].count(el)\n",
    "                N_el = properties_elem_df.loc[sim_name[3:] + \".cif\", \"N\"]\n",
    "                vacancy_atoms_str += f\" +energy('{sim_name}')*{N_v}/{N_el}\"\n",
    "            entries_str.append(f\"energy('{data_name}') {vacancy_atoms_str} -energy('{ref_name}')\")\n",
    "            weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3bd09879",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "id": "b04835ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "13de59b6",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "id": "52bf7363",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_jobs.update(elemets_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "036880e5",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "280c2a7e",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b112db67",
   "metadata": {},
   "source": [
    "##### Charge"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79c31717",
   "metadata": {},
   "source": [
    "Get metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "id": "8335b7fe",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"3-vacancy\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "586e0091",
   "metadata": {},
   "source": [
    "get systems molecules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "id": "bb9cbd51",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_3\n",
    "\n",
    "data_mol = {}\n",
    "\n",
    "for name, job in data_j.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ad182509",
   "metadata": {},
   "source": [
    "get vacancies neighbors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "id": "95431054",
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "cut_off = 2.5  # [Ang]\n",
    "data_vacacies_neighbors = {}\n",
    "max_vacacies_per_type = 5\n",
    "\n",
    "for index_v in properties_data_df.index:\n",
    "    sim_name = \"SP-\" + index_v[:-4]\n",
    "    data_vacacies_neighbors[sim_name] = []\n",
    "    for vacancies_r in properties_data_df.loc[index_v, \"vacancy_site\"]:\n",
    "        dist = get_distances_atom(data_mol[sim_name], vacancies_r, mic=True)\n",
    "        indx = np.where(dist <= cut_off)[0].tolist()\n",
    "        data_vacacies_neighbors[sim_name] += indx\n",
    "    data_vacacies_neighbors[sim_name] = np.unique(data_vacacies_neighbors[sim_name])\n",
    "    rg.shuffle(data_vacacies_neighbors[sim_name])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bf6492ab",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "id": "f7b51d7f",
   "metadata": {},
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_job in data_j.items():\n",
    "    neighbors = data_vacacies_neighbors[data_name]\n",
    "    if len(neighbors) > max_vacacies_per_type:\n",
    "        neighbors = neighbors[:max_vacacies_per_type]\n",
    "    for ai in neighbors:\n",
    "        entries_str.append(f\"charges('{data_name}', {ai})\")\n",
    "        weight_all.append(0.2 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "        sigma_all.append(1e-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a5187155",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "id": "dccbd4c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .1\n",
    "# sigma = 1.\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Charges\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f32e4883",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "id": "cdab5964",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "03c075db",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cd3a3d34",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a5162ec",
   "metadata": {},
   "source": [
    "### Substitution `Single Poin`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "id": "926af371",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Substitution (SP)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db2c439a",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2b355b74",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "id": "b6a7526f",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"4-substitution\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "id": "86f7ac4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_4\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            entries_str.append(f\"energy('{data_name}') - energy('{ref_name}')\")\n",
    "            weight_all.append(5e-2 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0778586d",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "id": "72916b18",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ac5bdf0",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "id": "10bbd3a1",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2b6e4f31",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fe8add98",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e972daa8",
   "metadata": {},
   "source": [
    "##### Forces"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2fb494f",
   "metadata": {},
   "source": [
    "Get metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "id": "3fe91916",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"4-substitution\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "838b1e3e",
   "metadata": {},
   "source": [
    "get systems molecules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "id": "eb816b7a",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_4\n",
    "\n",
    "data_mol = {}\n",
    "\n",
    "for name, job in data_j.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8b27e9fb",
   "metadata": {},
   "source": [
    "get vacancies neighbors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "id": "0f6f0588",
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "cut_off = 2.5  # [Ang]\n",
    "data_substitution_neighbors = {}\n",
    "max_substitution_per_type = 5\n",
    "\n",
    "for index_v in properties_data_df.index:\n",
    "    sim_name = \"SP-\" + index_v[:-4]\n",
    "    data_substitution_neighbors[sim_name] = []\n",
    "    for vacancies_r in properties_data_df.loc[index_v, \"substitution_site\"]:\n",
    "        dist = get_distances_atom(data_mol[sim_name], vacancies_r, mic=True)\n",
    "        indx = np.where(dist <= cut_off)[0].tolist()\n",
    "        data_substitution_neighbors[sim_name] += indx\n",
    "    data_substitution_neighbors[sim_name] = np.unique(data_substitution_neighbors[sim_name])\n",
    "    rg.shuffle(data_substitution_neighbors[sim_name])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4e2b905c",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "id": "04e126f1",
   "metadata": {},
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_job in data_j.items():\n",
    "    neighbors = data_substitution_neighbors[data_name]\n",
    "    if len(neighbors) > max_substitution_per_type:\n",
    "        neighbors = neighbors[:max_substitution_per_type]\n",
    "    for ai in neighbors:\n",
    "        entries_str.append(f\"forces('{data_name}', {ai})\")\n",
    "        weight_all.append(5e-2 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "        sigma_all.append(5e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "afea6821",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "id": "6b7cb789",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .15\n",
    "# sigma = 1.\n",
    "unit = (\"au/Ang\", scm.Units.convert(1, \"au/Ang\", \"au/Ang\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Forces\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "72a07cd3",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "id": "0f64a8c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb59b0da",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a58ac154",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8c9d0174",
   "metadata": {},
   "source": [
    "### Interstitial `Geometry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "id": "6a6f943d",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Interstitial (GO)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c936f21",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4c27cea9",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "id": "e7f38b9f",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"5-interstitial\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"0.1-pure_species\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_elem_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "id": "d0fce118",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_5\n",
    "elemets_j = jobs_01\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            vacancy_atoms_str = \"\"\n",
    "            for sim_name, el in zip(elemets_j.keys(), properties_elem_df[\"element\"]):\n",
    "                N_v = properties_data_df.loc[data_name[3:] + \".cif\", \"interstitial_element\"].count(el)\n",
    "                N_el = properties_elem_df.loc[sim_name[3:] + \".cif\", \"N\"]\n",
    "                vacancy_atoms_str += f\" -energy('{sim_name}')*{N_v}/{N_el}\"\n",
    "            entries_str.append(f\"energy('{data_name}') {vacancy_atoms_str} -energy('{ref_name}')\")\n",
    "            weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "81f9fabd",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "id": "c0bc023a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e88bd2ff",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 254,
   "id": "913ab8e7",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(elemets_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1bb0c1ba",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "067981d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "429a620f",
   "metadata": {},
   "source": [
    "##### Forces"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "00c0406b",
   "metadata": {},
   "source": [
    "Get metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "id": "5b8cac96",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"5.1-interstitial_PO\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cb1f5ad8",
   "metadata": {},
   "source": [
    "get systems molecules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "id": "b0cda025",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_5\n",
    "\n",
    "data_mol = {}\n",
    "\n",
    "for name, job in data_j.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c106a251",
   "metadata": {},
   "source": [
    "get vacancies neighbors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "id": "8af7e8eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "cut_off = 2.5  # [Ang]\n",
    "data_interstitial_neighbors = {}\n",
    "max_interstitial_per_type = 5\n",
    "\n",
    "for index_v in properties_data_df.index:\n",
    "    sim_name = \"GO-\" + index_v[:-4]\n",
    "    data_interstitial_neighbors[sim_name] = []\n",
    "    for interstitial_r in properties_data_df.loc[index_v, \"interstitial_site\"]:\n",
    "        dist = get_distances_atom(data_mol[sim_name], interstitial_r, mic=True)\n",
    "        indx = np.where(dist <= cut_off)[0].tolist()\n",
    "        data_interstitial_neighbors[sim_name] += indx\n",
    "    data_interstitial_neighbors[sim_name] = np.unique(data_interstitial_neighbors[sim_name])\n",
    "    rg.shuffle(data_interstitial_neighbors[sim_name])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "008134c6",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "id": "4237caba",
   "metadata": {},
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "f_prj = 0  # x component\n",
    "\n",
    "for data_name, data_job in data_j.items():\n",
    "    N_atoms = len(data_mol[data_name])\n",
    "    for ai in range(N_atoms):\n",
    "        entries_str.append(f\"forces('{data_name}', {ai}, {f_prj})\")\n",
    "        weight_all.append(0.8)\n",
    "        sigma_all.append(5e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a37635b1",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "id": "d3c99f69",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .15\n",
    "# sigma = 1.\n",
    "unit = (\"au/Ang\", scm.Units.convert(1, \"au/Ang\", \"au/Ang\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Forces\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a9c2209",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "id": "98627401",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8315c67a",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "07e4665d",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9a6c334e",
   "metadata": {},
   "source": [
    "##### Charge"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "45cd61cd",
   "metadata": {},
   "source": [
    "Get metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "id": "e793f101",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"5.1-interstitial_PO\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "id": "4b3be25d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# properties_data_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "609a19bb",
   "metadata": {},
   "source": [
    "get systems molecules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "id": "54e1bca4",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_5\n",
    "\n",
    "data_mol = {}\n",
    "\n",
    "for name, job in data_j.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "20d60140",
   "metadata": {},
   "source": [
    "get vacancies neighbors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "id": "cbe07d3a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "cut_off = 2.5  # [Ang]\n",
    "data_interstitial_atoms = {}\n",
    "max_interstitial_per_type = 5\n",
    "\n",
    "for index_v in properties_data_df.index:\n",
    "    sim_name = \"GO-\" + index_v[:-4]\n",
    "    data_interstitial_neighbors[sim_name] = []\n",
    "    Natoms = len(data_mol[sim_name])\n",
    "    Nint = len(properties_data_df.loc[index_v, \"interstitial_site\"])\n",
    "    data_interstitial_atoms[sim_name] = np.arange(Natoms - Nint, Natoms)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14bb842a",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "id": "9bc4bdc6",
   "metadata": {},
   "outputs": [],
   "source": [
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_job in data_j.items():\n",
    "    interstitials = data_interstitial_atoms[data_name]\n",
    "    for ai in interstitials:\n",
    "        entries_str.append(f\"charges('{data_name}', {ai})\")\n",
    "        weight_all.append(0.2 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "        sigma_all.append(1e-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "677939b5",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "id": "8dcd4ac5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .1\n",
    "# sigma = 1.\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Charges\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d4c83df",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "id": "97e35f7b",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a69b6f9f",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8a7cf790",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5ad1a195",
   "metadata": {},
   "source": [
    "### Slabs `Geometry Optimization`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "id": "feea2223",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Slabs (GO)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "639cb704",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f62a15a",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "id": "bb83681c",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = jobs_0\n",
    "data_j = jobs_6\n",
    "entries_str = []\n",
    "cis_dir = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"6.1-slabs_PO\")\n",
    "data_path = os.path.join(cis_dir, \"properties.json\")\n",
    "properties_df = pd.read_json(data_path)\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[1])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            Na_ref = len(ref_job.results.get_input_molecule())\n",
    "            Na_data = len(data_job.results.get_input_molecule())\n",
    "            entries_str.append(f\"energy('{data_name}')/{Na_data} - energy('{ref_name}')/{Na_ref} \")\n",
    "            weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8db261a4",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "id": "83dcec9c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f67b9f50",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "id": "120cc3f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "341d7409",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cd33c015",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c7021be8",
   "metadata": {},
   "source": [
    "##### Bond distance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a37db105",
   "metadata": {},
   "source": [
    "get intila configurations for each job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "id": "798ef971",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol = {}\n",
    "\n",
    "for name, job in jobs_6.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "931b09ee",
   "metadata": {},
   "source": [
    "get molecule bonds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "id": "e724f27c",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol_bond = {}\n",
    "\n",
    "for name, ase_mol in data_mol.items():\n",
    "    data_mol_bond[name] = get_all_bonds(ase_mol, verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "39576ca5",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "max_bond_per_type = 10\n",
    "\n",
    "data_bond_indx = {}\n",
    "for name, ase_mol in data_mol.items():\n",
    "    surf_atoms = get_surf_atoms(ase_mol, cutoff=3.0, coord_cutoff=0.5)\n",
    "    bond_df = data_mol_bond[name]\n",
    "    for indx in bond_df.index:\n",
    "        bonds_all = bond_df.loc[indx, \"bonds\"]\n",
    "        # get only bond with surface atoms\n",
    "        bonds = []\n",
    "        for ai, aj in bonds_all:\n",
    "            if (ai in surf_atoms) or (aj in surf_atoms):\n",
    "                bonds.append((ai, aj))\n",
    "        if len(bonds) > max_bond_per_type:\n",
    "            rnd_indexs = np.arange(len(bonds))\n",
    "            rg.shuffle(rnd_indexs)\n",
    "            rnd_indexs = rnd_indexs[:max_bond_per_type]\n",
    "            data_bond_indx[name] = [bonds[bi] for bi in rnd_indexs]\n",
    "        else:\n",
    "            data_bond_indx[name] = bonds"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70c399c8",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "id": "e19270a7",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_6\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "\n",
    "for data_name, data_bonds in data_bond_indx.items():\n",
    "    for ai, aj in data_bonds:\n",
    "        entries_str.append(f\"distance('{data_name}', {ai}, {aj})\")\n",
    "        weight_all.append(0.2 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "        sigma_all.append(1e-2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "31a1d710",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "id": "4839e4f1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .1\n",
    "# sigma = 1.\n",
    "# unit = ('Angstrom', scm.Units.convert(1, 'au', 'Angstrom'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Distance\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d695cf30",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "id": "adbe006d",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d2849299",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e5f283da",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0db263aa",
   "metadata": {},
   "source": [
    "##### Angles"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79813b93",
   "metadata": {},
   "source": [
    "get intila configurations for each job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "id": "9cf313ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol = {}\n",
    "\n",
    "for name, job in jobs_6.items():\n",
    "    mol = job.results.get_input_molecule()\n",
    "    mol = SCMtoASE(mol)\n",
    "    data_mol[name] = mol.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "deca8bb4",
   "metadata": {},
   "source": [
    "get molecule angles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "id": "336c9429",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mol_angle = {}\n",
    "\n",
    "for name, ase_mol in data_mol.items():\n",
    "    data_mol_angle[name] = get_all_angles(ase_mol, verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "id": "22f69de1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# data_mol_angle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "id": "26bedc60",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "max_angle_per_type = 10\n",
    "\n",
    "data_angle_indx = {}\n",
    "for name, ase_mol in data_mol.items():\n",
    "    surf_atoms = get_surf_atoms(ase_mol, cutoff=3.0, coord_cutoff=0.5)\n",
    "    angle_df = data_mol_angle[name]\n",
    "    for indx in angle_df.index:\n",
    "        angles_all = angle_df.loc[indx, \"angles\"]\n",
    "        # get only angle with surface atoms\n",
    "        angles = []\n",
    "        for ai, aj, ak in angles_all:\n",
    "            if (ai in surf_atoms) or (aj in surf_atoms) or (ak in surf_atoms):\n",
    "                angles.append((ai, aj, ak))\n",
    "        if len(angles) > max_angle_per_type:\n",
    "            rnd_indexs = np.arange(len(angles))\n",
    "            rg.shuffle(rnd_indexs)\n",
    "            rnd_indexs = rnd_indexs[:max_angle_per_type]\n",
    "            data_angle_indx[name] = [angles[bi] for bi in rnd_indexs]\n",
    "        else:\n",
    "            data_angle_indx[name] = angles"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "871d1d34",
   "metadata": {},
   "source": [
    "Preper Entry"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "id": "e759130d",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_j = jobs_6\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for data_name, data_angles in data_angle_indx.items():\n",
    "    for ai, aj, ak in data_angles:\n",
    "        entries_str.append(f\"angles('{data_name}', ({ai}, {aj}, {ak}) )\")\n",
    "        weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1))\n",
    "        sigma_all.append(3.14)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7bc09ff",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "id": "a163a8b0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = .35\n",
    "# sigma = 1.\n",
    "# unit = ('Angstrom', scm.Units.convert(1, 'au', 'Angstrom'))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Angles\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "19841e53",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "id": "64f7d83c",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8114c030",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "193e1930",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "906e211a",
   "metadata": {},
   "source": [
    "### Amorphous bulk `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "id": "1c87870f",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Amorphous bulk  (SP)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4ed7548",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85bc0def",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "id": "c598dfbb",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"7.1-amorphous_frames\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "id": "34ec8868",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_7\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            entries_str.append(f\"energy('{data_name}') - energy('{ref_name}') \")\n",
    "            weight_all.append(1.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9847011d",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "id": "fc74ce68",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.\n",
    "# sigma = 1.\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce08ad19",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "id": "ef142ed4",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc3268da",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "aa5ced72",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "906e211a",
   "metadata": {},
   "source": [
    "### Bulk 300 K `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 319,
   "id": "1c87870f",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Bulk at 300 K (SP)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4ed7548",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85bc0def",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 320,
   "id": "c598dfbb",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"10.1-bulk_300K_frames\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 321,
   "id": "34ec8868",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_10\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            entries_str.append(f\"energy('{data_name}') - energy('{ref_name}') \")\n",
    "            weight_all.append(10.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9847011d",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 322,
   "id": "fc74ce68",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce08ad19",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 323,
   "id": "ef142ed4",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc3268da",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "aa5ced72",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "906e211a",
   "metadata": {},
   "source": [
    "### Bulk 500 K `Single Point`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 325,
   "id": "1c87870f",
   "metadata": {},
   "outputs": [],
   "source": [
    "systems_name = \"Bulk at 500 K (SP)\"\n",
    "N_df = len(dataset_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4ed7548",
   "metadata": {},
   "source": [
    "##### Energy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85bc0def",
   "metadata": {},
   "source": [
    "Get Reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 326,
   "id": "c598dfbb",
   "metadata": {},
   "outputs": [],
   "source": [
    "cis_dir_ref = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"1-supercell\")\n",
    "data_pathr_ref = os.path.join(cis_dir_ref, \"properties.json\")\n",
    "properties_ref_df = pd.read_json(data_pathr_ref)\n",
    "cis_dir_data = os.path.join(MAIN_DIR_JNB1, CHEM_FORMULA, \"11.1-bulk_500K_frames\")\n",
    "data_path_data = os.path.join(cis_dir_data, \"properties.json\")\n",
    "properties_data_df = pd.read_json(data_path_data)\n",
    "supercell_size = properties_data_df[\"unitcell_repetition\"].unique()[0]\n",
    "refernce_cis = properties_ref_df.loc[properties_ref_df[\"unitcell_repetition\"] == supercell_size].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 327,
   "id": "34ec8868",
   "metadata": {},
   "outputs": [],
   "source": [
    "referece_j = {}\n",
    "for refernce_cis_ in refernce_cis:\n",
    "    ref_name = refernce_cis_[:-4]\n",
    "    for name, job in jobs_1.items():\n",
    "        if ref_name in name:\n",
    "            referece_j[name] = job\n",
    "\n",
    "data_j = jobs_11\n",
    "entries_str = []\n",
    "weight_all = []\n",
    "sigma_all = []\n",
    "\n",
    "for ref_name, ref_job in referece_j.items():\n",
    "    ref_ID = int(ref_name.split(\"-\")[2])\n",
    "    for data_name, data_job in data_j.items():\n",
    "        data_ID = int(data_name.split(\"-\")[2])\n",
    "        if ref_ID == data_ID:\n",
    "            entries_str.append(f\"energy('{data_name}') - energy('{ref_name}') \")\n",
    "            weight_all.append(10.0 / (float(data_name.split(\"-\")[2]) + 1.0))\n",
    "            sigma_all.append(2e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9847011d",
   "metadata": {},
   "source": [
    "Add entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 328,
   "id": "fc74ce68",
   "metadata": {},
   "outputs": [],
   "source": [
    "# weight = 1.\n",
    "# sigma = 1.\n",
    "# unit = ('kJ/mol', scm.Units.convert(1, 'au', 'kJ/mol'))\n",
    "unit = (\"au\", scm.Units.convert(1, \"au\", \"au\"))\n",
    "\n",
    "for entry, weight, sigma in zip(entries_str, weight_all, sigma_all):\n",
    "    dataset.add_entry(entry, weight, sigma=sigma, unit=unit)\n",
    "    dataset_df.loc[N_df, \"entry_sigma\"] = sigma\n",
    "    dataset_df.loc[N_df, \"chem_formula\"] = CHEM_FORMULA\n",
    "    dataset_df.loc[N_df, \"systems_name\"] = systems_name\n",
    "    dataset_df.loc[N_df, \"entry_type\"] = \"Energy\"\n",
    "    dataset_df.loc[N_df, \"entry_weight\"] = weight\n",
    "    N_df += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce08ad19",
   "metadata": {},
   "source": [
    "Evaluate all the entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 329,
   "id": "ef142ed4",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_jobs = {}\n",
    "all_jobs.update(referece_j)\n",
    "all_jobs.update(data_j)\n",
    "all_results = {}\n",
    "\n",
    "for name, job in all_jobs.items():\n",
    "    all_results[name] = job.results\n",
    "\n",
    "dataset.calculate_reference(all_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc3268da",
   "metadata": {},
   "source": [
    "Show last 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "aa5ced72",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(dataset[-5:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f3d73f1",
   "metadata": {},
   "source": [
    "### Storeing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4871333e",
   "metadata": {},
   "source": [
    "check CONSISTENCY"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 333,
   "id": "b2a782d1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ALL GOOD\n"
     ]
    }
   ],
   "source": [
    "bad_ids = dataset.check_consistency(jobcollection)\n",
    "if len(bad_ids) > 0:\n",
    "    raise RuntimeError(f\"The Dataset entries {len(bad_ids)} are INCONSISTENT with the JobCollection\")\n",
    "else:\n",
    "    print(\"ALL GOOD\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab29c50a",
   "metadata": {},
   "source": [
    "### Saving `DataSet`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "37b45516",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Remove problematic\n",
    "job_ids = set(dataset.jobids)\n",
    "print(len(job_ids))\n",
    "for job_name in dataset.jobids:\n",
    "    if \"-0.200\" in job_name:\n",
    "        print(f\"{job_name} removed\")\n",
    "        job_ids.remove(job_name)\n",
    "\n",
    "print(len(job_ids))\n",
    "dataset = dataset.from_jobids(job_ids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 336,
   "id": "bdf34c7e",
   "metadata": {},
   "outputs": [],
   "source": [
    "now = datetime.datetime.now(TIMEZONE)\n",
    "comments = dict(\n",
    "    info=f\"\"\"Full Datasets entries\n",
    "@date = {now.strftime(\"%Y-%m-%d %H:%M:%S %Z\")} ({TIMEZONE.zone})\n",
    "@authors = Paolo De Angelis\n",
    "@mail = paolo.deangelis@polito.it\n",
    "@PLAMS = {scm.__version__}\n",
    "@ParAMS = {params.__version__}\"\"\"\n",
    ")\n",
    "dataset.header = comments\n",
    "dataset.store(os.path.join(MAIN_DIR_JNB3, \"dataset.yaml\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2e486ae",
   "metadata": {},
   "source": [
    "split in `dataset`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 337,
   "id": "b696f9ca",
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.random import PCG64, Generator\n",
    "\n",
    "pcg64 = PCG64(seed=SEED)\n",
    "rg = Generator(pcg64)\n",
    "\n",
    "job_ids = set()\n",
    "# crystal\n",
    "job_ids = job_ids.union(set(jobs_0.keys()))\n",
    "# supercell\n",
    "# drop last system\n",
    "jobs_sc = []\n",
    "for job_name in jobs_1.keys():\n",
    "    if \"2-LiF_Pm-3m\" not in job_name:\n",
    "        jobs_sc.append(job_name)\n",
    "job_ids = job_ids.union(set(jobs_sc))\n",
    "# strain\n",
    "jobs_strain = []\n",
    "for job_name in jobs_2.keys():\n",
    "    if \"2-LiF_Pm-3m\" not in job_name:\n",
    "        jobs_strain.append(job_name)\n",
    "N = len(jobs_strain)\n",
    "rg.shuffle(jobs_strain)\n",
    "job_ids = job_ids.union(set(jobs_strain[: int(N * 0.7)]))\n",
    "# 300 K\n",
    "job_ids = job_ids.union(set(jobs_10.keys()))\n",
    "# 500 K\n",
    "job_ids = job_ids.union(set(jobs_11.keys()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "33fe137b",
   "metadata": {},
   "outputs": [],
   "source": [
    "job_ids = set(dataset.jobids)\n",
    "print(len(job_ids))\n",
    "for job_name in dataset.jobids:\n",
    "    if \"LiF_Pm-3m\" in job_name:\n",
    "        job_ids.remove(job_name)\n",
    "\n",
    "print(len(job_ids))\n",
    "dataset_2over3 = dataset.from_jobids(job_ids)\n",
    "\n",
    "job_ids = set(dataset.jobids)\n",
    "print(len(job_ids))\n",
    "for job_name in dataset.jobids:\n",
    "    if \"LiF_Pm-3m\" in job_name or \"LiF_P6_3mc\" in job_name:\n",
    "        job_ids.remove(job_name)\n",
    "\n",
    "print(len(job_ids))\n",
    "dataset_1over3 = dataset.from_jobids(job_ids)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Training Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 342,
   "id": "93cb4f6a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/3 set = 1153 with 184 jobs\n"
     ]
    }
   ],
   "source": [
    "now = datetime.datetime.now(TIMEZONE)\n",
    "comments = dict(\n",
    "    info=f\"\"\"Train Set entries\n",
    "@date = {now.strftime(\"%Y-%m-%d %H:%M:%S %Z\")} ({TIMEZONE.zone})\n",
    "@authors = Paolo De Angelis\n",
    "@mail = paolo.deangelis@polito.it\n",
    "@PLAMS = {scm.__version__}\n",
    "@ParAMS = {params.__version__}\"\"\"\n",
    ")\n",
    "dataset_1over3.header = comments\n",
    "dataset_1over3.store(os.path.join(MAIN_DIR_JNB3, \"training_set.yaml\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Test Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ea1344f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "now = datetime.datetime.now(TIMEZONE)\n",
    "comments = dict(\n",
    "    info=f\"\"\"Train Set entries\n",
    "@date = {now.strftime(\"%Y-%m-%d %H:%M:%S %Z\")} ({TIMEZONE.zone})\n",
    "@authors = Paolo De Angelis\n",
    "@mail = paolo.deangelis@polito.it\n",
    "@PLAMS = {scm.__version__}\n",
    "@ParAMS = {params.__version__}\"\"\"\n",
    ")\n",
    "dataset_2over3.header = comments\n",
    "dataset_2over3.store(os.path.join(MAIN_DIR_JNB3, \"test_set.yaml\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db09b9b9",
   "metadata": {},
   "source": [
    "Save dataset metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 346,
   "id": "775c9e89",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_df.to_json(os.path.join(MAIN_DIR_JNB3, \"dataset_metadata.json\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1d37a928",
   "metadata": {},
   "source": [
    "## Plot Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "87d28309",
   "metadata": {},
   "source": [
    "### Entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0a594ca3",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_df = pd.read_json(os.path.join(MAIN_DIR_JNB3, \"dataset_metadata.json\"))\n",
    "dataset_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 348,
   "id": "50d166ac",
   "metadata": {},
   "outputs": [],
   "source": [
    "cmap = mp.cm.get_cmap(\"bone\")\n",
    "inner_color = cmap(np.arange(14) / 14)\n",
    "cmap = mp.cm.get_cmap(\"tab10\")\n",
    "outer_color = cmap(np.arange(14))\n",
    "\n",
    "entry_type_label = []\n",
    "entry_type_count = []\n",
    "entry_type_colors = []\n",
    "system_label = []\n",
    "system_count = []\n",
    "system_colors = []\n",
    "system_names = dataset_df[\"systems_name\"].unique()\n",
    "\n",
    "for i, entry_type in enumerate(dataset_df[\"entry_type\"].unique()):\n",
    "    sub_df = dataset_df.loc[dataset_df[\"entry_type\"] == entry_type]\n",
    "    entry_type_label.append(entry_type)\n",
    "    entry_type_count.append(len(sub_df))\n",
    "    entry_type_colors.append(outer_color[i].tolist())\n",
    "    for j, systems_name in enumerate(sub_df[\"systems_name\"].unique()):\n",
    "        sub_sub_df = sub_df.loc[sub_df[\"systems_name\"] == systems_name]\n",
    "        system_label.append(systems_name)\n",
    "        system_count.append(len(sub_sub_df))\n",
    "        nc = np.where(system_names == systems_name)[0][0]\n",
    "        system_colors.append(inner_color[nc].tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 349,
   "id": "f32b02cb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 700x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=[7, 4], dpi=100, facecolor=\"white\")\n",
    "\n",
    "bbox_props = dict(boxstyle=\"square,pad=0.1\", fc=\"w\", ec=\"none\", lw=0.72)\n",
    "kw = dict(arrowprops=dict(arrowstyle=\"-\"), bbox=bbox_props, zorder=0, va=\"center\")\n",
    "\n",
    "with plt.style.context(\"seaborn\"):\n",
    "    spec = gridspec.GridSpec(ncols=1, nrows=1, figure=fig)\n",
    "    ax = fig.add_subplot(spec[0])\n",
    "    wedges, texts = ax.pie(\n",
    "        entry_type_count,\n",
    "        radius=1,  # labels=entry_type_label,\n",
    "        colors=entry_type_colors,\n",
    "        wedgeprops=dict(width=0.3, edgecolor=\"white\"),\n",
    "    )\n",
    "    ax.pie(system_count, radius=0.7, colors=system_colors, wedgeprops=dict(width=0.3, edgecolor=\"white\"))\n",
    "    ax.annotate(f\"{len(dataset_df)}\", xy=(0.0, 0.0), xytext=(0.0, 0.0), horizontalalignment=\"center\")\n",
    "    # labels\n",
    "    for i, p in enumerate(wedges):\n",
    "        ang = (p.theta2 - p.theta1) / 2.0 + p.theta1\n",
    "        y = np.sin(np.deg2rad(ang))\n",
    "        x = np.cos(np.deg2rad(ang))\n",
    "        horizontalalignment = {-1: \"right\", 1: \"left\"}[int(np.sign(x))]\n",
    "        connectionstyle = f\"angle,angleA=0,angleB={ang}\"\n",
    "        kw[\"arrowprops\"].update({\"connectionstyle\": connectionstyle})\n",
    "        ax.annotate(\n",
    "            entry_type_label[i],\n",
    "            xy=(x, y),\n",
    "            xytext=(1.2 * np.sign(x), 1.2 * y),\n",
    "            horizontalalignment=horizontalalignment,\n",
    "            **kw,\n",
    "        )\n",
    "    inner_patch = []\n",
    "    for i, label in enumerate(dataset_df[\"systems_name\"].unique()):\n",
    "        inner_patch.append(mp.patches.Patch(color=inner_color[i], label=label))\n",
    "    ax.legend(handles=inner_patch, loc=\"upper left\", bbox_to_anchor=(1.3, 1.0))\n",
    "fig.savefig(os.path.join(\"plots\", \"dataset_entries.png\"), dpi=330, bbox_inches=\"tight\", facecolor=\"none\")\n",
    "fig.savefig(os.path.join(\"plots\", \"dataset_entries.pdf\"), dpi=330, bbox_inches=\"tight\", facecolor=\"none\")"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "cba73819b033846e8a8e00b60503f464f1519761ae2cd4b097b44cbe8a75715b"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  },
  "toc-autonumbering": false
 },
 "nbformat": 4,
 "nbformat_minor": 5
}