exp.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch_geometric.datasets import Planetoid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = Planetoid('dataset', 'CiteSeer', split='public')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 182,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = Planetoid('dataset', 'Cora', split='public')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = dataset[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([140, 1433])"
      ]
     },
     "execution_count": 212,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.x[data.train_mask].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[(0, 1, 0), (0, 2, 0), (0, 3, 0)]\n",
      "[0.6853554  0.65664876] [0.14483912 0.6247451 ]\n",
      "[0.6853554  0.65664876] [0.12032649 0.839218  ]\n",
      "[0.6853554  0.65664876] [0.0429733  0.84485835]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[-1.8807],\n",
      "        [-0.0456],\n",
      "        [-0.9089],\n",
      "        [-1.6107]])\n"
     ]
    }
   ],
   "source": [
    "\n",
    "import torch\n",
    "import numpy as np\n",
    "import networkx as nx\n",
    "\n",
    "from torch_geometric.utils import to_networkx\n",
    "\n",
    "from torch_geometric.data import Data\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def draw_data(data):\n",
    "    G = to_networkx(data, to_multi=True)\n",
    "    print(G.edges)\n",
    "    pos = nx.random_layout(G)\n",
    "    nx.draw_networkx_nodes(G, pos, node_color='r')\n",
    "    nx.draw_networkx_labels(G, pos)\n",
    "    ax = plt.gca()\n",
    "    for e in G.edges:\n",
    "        print(pos[e[0]], pos[e[1]])\n",
    "        ax.annotate(\"\",\n",
    "                    xy=pos[e[0]], xycoords='data',\n",
    "                    xytext=pos[e[1]], textcoords='data',\n",
    "                    arrowprops=dict(arrowstyle=\"<-\", color=\"0.5\",\n",
    "                                    shrinkA=10, shrinkB=10,\n",
    "                                    patchA=None, patchB=None,\n",
    "                                    connectionstyle=\"arc3,rad=rrr\".replace('rrr',str(0.3*e[2])\n",
    "                                    ),\n",
    "                                    ),\n",
    "                    )\n",
    "    plt.axis('off')\n",
    "    plt.show()\n",
    "    \n",
    "    \n",
    "data = Data()\n",
    "data.x = torch.randn((4, 1))\n",
    "data.edge_index = torch.tensor([\n",
    "    [0,1],\n",
    "    [0,2],\n",
    "    [0,3]\n",
    "]).t().contiguous()\n",
    "draw_data(data)\n",
    "print(data.x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data(x=[11, 1433], edge_index=[2, 12], y=[11], train_mask=[11], val_mask=[11], test_mask=[11], n_id=[11], e_id=[12], input_id=[2], batch_size=2)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "EdgeIndex([[ 2,  3,  4,  5,  0,  6,  0,  7,  8,  9,  1, 10],\n",
       "           [ 0,  0,  1,  1,  2,  2,  3,  3,  4,  4,  5,  5]],\n",
       "          sparse_size=(11, 11), nnz=12, sort_order=col)"
      ]
     },
     "execution_count": 217,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from torch_geometric.loader import NeighborLoader\n",
    "\n",
    "dl = NeighborLoader(\n",
    "    data, \n",
    "    [2, 2],\n",
    "    input_nodes=data.train_mask,\n",
    "    batch_size=2,\n",
    "    directed=True\n",
    ")\n",
    "\n",
    "for batch in dl:\n",
    "    print(batch)\n",
    "    break\n",
    "    \n",
    "batch.edge_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([0, 1])"
      ]
     },
     "execution_count": 218,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch.input_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([ True,  True, False, False,  True, False, False, False, False, False,\n",
       "        False])"
      ]
     },
     "execution_count": 219,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch.train_mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "multignn",
   "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": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}