Equi(Kit)Script for workflows that compute, transform, join representations and estimate properties

examples/h5o2+.extxyz

@@ -0,0 +1,9 @@
+7
+Lattice='100 0 0 0 100 0 0 0 100'
+       O  5.40450e-01 -9.74850e-01 -2.16580e-01
+       O  2.92484e+00 -8.39252e-01  1.52397e-01
+       H  1.83860e-01 -1.25804e+00 -1.07875e+00
+       H  1.74329e+00 -9.09278e-01 -1.00395e-01
+       H  3.29707e+00 -1.59324e+00  6.47253e-01
+       H  3.53033e+00 -6.08138e-01 -5.76078e-01
+       H  4.23316e-02 -1.94860e-01  9.22810e-02

examples/multi_spectra_script.ipynb

@@ -0,0 +1,227 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "c86f17cd",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import rascaline\n",
+    "rascaline._c_lib._get_library()\n",
+    "\n",
+    "from copy import deepcopy\n",
+    "\n",
+    "import numpy as np\n",
+    "\n",
+    "from equisolve.numpy.scripts import MultiSpectraScript\n",
+    "from equisolve.numpy.models.linear_model import Ridge\n",
+    "from equisolve.numpy.preprocessing import StandardScaler\n",
+    "from equisolve.utils.convert import ase_to_tensormap\n",
+    "\n",
+    "import ase.io\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "a199029f",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "frames = ase.io.read(\"water_converted.extxyz\", \":20\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "baca7fb8",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "HYPERS = {\n",
+    "    \"cutoff\": 3.0,\n",
+    "    \"max_radial\": 6,\n",
+    "    \"max_angular\": 4,\n",
+    "    \"atomic_gaussian_width\": 0.3,\n",
+    "    \"center_atom_weight\": 1.0,\n",
+    "    \"radial_basis\": {\"Gto\": {}},\n",
+    "    \"cutoff_function\": {\"ShiftedCosine\": {\"width\": 0.5}},\n",
+    "}\n",
+    "multi_spectra_hypers = {}\n",
+    "#multi_spectra_hypers['Composition'] = {}\n",
+    "multi_spectra_hypers['SoapRadialSpectrum'] = deepcopy(HYPERS)\n",
+    "multi_spectra_hypers['SoapRadialSpectrum'].pop('max_angular')\n",
+    "multi_spectra_hypers['SoapPowerSpectrum'] = deepcopy(HYPERS)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "0233bbb9",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "y = ase_to_tensormap(frames, energy=\"TotEnergy\", forces=\"force\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "cdf7caa4",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "671.9636307759672"
+      ]
+     },
+     "execution_count": 5,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# Because all frames have the same species, we cannot scale, so we just subtract the mean\n",
+    "#transformer_composition = StandardScaler(parameter_keys=[\"values\", \"positions\"], with_mean=True, with_std=False)\n",
+    "# THIS IS BUGGY, something wrong in the standardizer when setting these flags to False\n",
+    "\n",
+    "transformer = StandardScaler(parameter_keys=[\"values\", \"positions\"])\n",
+    "estimator = Ridge(parameter_keys=[\"values\", \"positions\"])\n",
+    "script = MultiSpectraScript(multi_spectra_hypers,\n",
+    "                            #transformer_X=transformer,\n",
+    "                            #transformer_y=transformer,\n",
+    "                            estimator=estimator)\n",
+    "Xi = script.compute(systems=frames, gradients=[\"positions\"])\n",
+    "script.fit(Xi, y, **{\"estimator_kwargs\": {\"alpha\": 1e-15}})\n",
+    "script.score(Xi, y)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "902e9517",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pickle\n",
+    "with open(\"multi_spectra_script-water.pickle\", \"wb\") as file:\n",
+    "    pickle.dump(script, file)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "3b640034",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# RESTART NOTEBOOK"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "99a94bf0",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pickle\n",
+    "with open(\"multi_spectra_script-water.pickle\", \"rb\") as file:\n",
+    "    script = pickle.load(file)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "5369b107",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "-29201.186821558364\n",
+      "\n",
+      "[[-295.60671847   -4.62718863  -55.59864334]\n",
+      " [ 298.54620907    4.04942317   37.15507667]\n",
+      " [-221.68776379 -150.28232975 -242.81915489]\n",
+      " [  -7.67013877    1.6028996    48.61326349]\n",
+      " [ 176.03569875 -146.94124088  179.13481445]\n",
+      " [ 286.69810363  136.65180697 -172.50009028]\n",
+      " [-220.83316507  156.29491821  107.1942928 ]]\n"
+     ]
+    }
+   ],
+   "source": [
+    "# in ipi\n",
+    "\n",
+    "import ase.io\n",
+    "import numpy as np\n",
+    "\n",
+    "structure = ase.io.read(\"h5o2+.extxyz\", \"0\")\n",
+    "\n",
+    "Xi = script.compute(systems=structure, gradients=[\"positions\"])\n",
+    "y_pred = script.forward(Xi) # implicitely done in score function\n",
+    "energy = y_pred.block().values[0][0]\n",
+    "forces = np.array(y_pred.block().gradient(\"positions\").data.reshape(-1, 3))\n",
+    "\n",
+    "print(energy)\n",
+    "print()\n",
+    "print(forces)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "995fe798",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<equisolve.numpy.scripts.md_calculator.GenericMDCalculator at 0x7f8d31a0d6a0>"
+      ]
+     },
+     "execution_count": 10,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "from equisolve.numpy.scripts import GenericMDCalculator\n",
+    "\n",
+    "from ipi.utils.mathtools import det_ut3x3\n",
+    "from ipi.utils.units import unit_to_internal, unit_to_user\n",
+    "\n",
+    "\n",
+    "md_calc = GenericMDCalculator(\"multi_spectra_script-water.pickle\",\n",
+    "                    is_periodic=True,\n",
+    "                    structure_template=\"h5o2+.extxyz\",\n",
+    "                    atomic_numbers=[1,8])"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.8.0"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}