{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load in \n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the \"../input/\" directory.\n# For example, running this (by clicking run or pressing Shift+Enter) will list the files in the input directory\n\nimport os\nprint(os.listdir(\"../input\"))\n\n# Any results you write to the current directory are saved as output.","execution_count":2,"outputs":[{"output_type":"stream","text":"['test.csv', 'mulliken_charges.csv', 'dipole_moments.csv', 'train.csv', 'structures.csv', 'magnetic_shielding_tensors.csv', 'potential_energy.csv', 'sample_submission.csv', 'scalar_coupling_contributions.csv', 'structures']\n","name":"stdout"}]},{"metadata":{"_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","trusted":true},"cell_type":"code","source":"train_df = pd.read_csv('../input/train.csv')\ntest_df = pd.read_csv('../input/test.csv')\nstructures = pd.read_csv('../input/structures.csv')","execution_count":4,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from torchvision import transforms, utils\nfrom torch.utils.data import Dataset, DataLoader\n\nimport random\nimport numbers\nimport math\nimport collections\n\nimport numpy as np\nimport torch","execution_count":5,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# should add CV\nclass MolecularDataset(Dataset):\n def __init__(self, df, structures, mode=\"train\", transform=\"augmentation\"):\n self.mode = mode\n self.transform = transform\n self.df = df\n self.structures = structures\n# self.df.reset_index(drop=True)\n# self.structures.reset_index(drop=True)\n self.types = ['1JHC', '2JHH', '1JHN', '2JHN', '2JHC', '3JHH', '3JHC', '3JHN']\n \n def __len__(self):\n if self.mode == \"train\":\n return 4000000\n else:\n return 658147\n \n def points_to_3D_map(self, points):\n '''points: n*4(x,y,z,index) '''\n grid = np.zeros((256,256,256))\n for point in points:\n# print(point)\n x = int(point[0]*10)+128\n y = int(point[1]*10)+128\n z = int(point[2]*10)+128\n grid[x][y][z] = point[3]\n return grid\n \n def type_to_num(self, type):\n num = 0\n for i in range(8):\n if type == self.types[i]:\n num = i\n break\n return num\n \n \n# index 0\n# id 0 [0]\n# molecule_name dsgdb9nsd_000001 [1]\n# atom_index_0 1 [2]\n# atom_index_1 0\n# type 1JHC\n# scalar_coupling_constant 84.8076\n# atom_count 5\n# Name: 0, dtype: object\n def __getitem__(self, idx):\n if self.mode == \"train\":\n infomation = np.zeros((3,))\n# print(\"**xxxxxxxxx\")\n# print(self.df.loc[idx][2])\n# print(self.df.loc[idx][3])\n# print(self.df.loc[idx][4])\n infomation[0] = self.df.loc[idx]['atom_index_0']\n infomation[1] = self.df.loc[idx]['atom_index_1']\n infomation[2] = self.type_to_num(self.df.loc[idx]['type'])\n \n name = self.df.loc[idx]['molecule_name']\n# print(\"here 3\")\n structures_of_name = structures.loc[structures['molecule_name'] == name]\n# print(\"here 4\")\n points = np.zeros((structures_of_name.shape[0], 4))\n for i in range(structures_of_name.shape[0]):\n points[i, 0] = structures_of_name.loc[i]['x']\n points[i, 1] = structures_of_name.loc[i]['y']\n points[i, 2] = structures_of_name.loc[i]['z']\n points[i, 3] = structures_of_name.loc[i]['atom_index']\n# print(\"here 5\")\n# grid = self.points_to_3D_map(points)\n \n target = np.array(self.df.loc[idx]['scalar_coupling_constant'])\n return torch.from_numpy(infomation), torch.from_numpy(points), torch.from_numpy(target)\n else:\n infomation = np.zeros((3,))\n infomation[0] = self.df.loc[idx]['atom_index_0']\n infomation[1] = self.df.loc[idx]['atom_index_1']\n infomation[2] = self.type_to_num(self.df.loc[idx]['type'])\n \n name = self.df.loc[idx][1]\n structures_of_name = structures.loc[structures['molecule_name'] == name]\n points = np.zeros((structures_of_name.shape[0], 4))\n for i in range(structures_of_name.shape[0]):\n points[i, 0] = structures_of_name.loc[i]['x']\n points[i, 1] = structures_of_name.loc[i]['y']\n points[i, 2] = structures_of_name.loc[i]['z']\n points[i, 3] = structures_of_name.loc[i]['atom_index']\n \n# grid = self.points_to_3D_map(points)\n \n target = np.array(self.df.loc[idx]['scalar_coupling_constant'])\n return torch.from_numpy(infomation), torch.from_numpy(points), torch.from_numpy(target)","execution_count":6,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"dataloader = MolecularDataset(df=train_df, structures=structures, mode='train')\nfor i in range(len(dataloader)):\n information, grid, target = dataloader[i]\n print(\"*****in *****\")\n print(information)\n print(\"*****grid*****\")\n print(grid)\n print(\"*****target*****\")\n print(target)\n if i == 0:\n break","execution_count":7,"outputs":[{"output_type":"stream","text":"*****in *****\ntensor([1., 0., 0.], dtype=torch.float64)\n*****grid*****\ntensor([[-1.2698e-02, 1.0858e+00, 8.0010e-03, 0.0000e+00],\n [ 2.1504e-03, -6.0313e-03, 1.9761e-03, 1.0000e+00],\n [ 1.0117e+00, 1.4638e+00, 2.7657e-04, 2.0000e+00],\n [-5.4082e-01, 1.4475e+00, -8.7664e-01, 3.0000e+00],\n [-5.2381e-01, 1.4379e+00, 9.0640e-01, 4.0000e+00]],\n dtype=torch.float64)\n*****target*****\ntensor(84.8076, dtype=torch.float64)\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.4","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}